Communications in Computer and Information Science
254
Azizah Abd Manaf Shamsul Sahibuddin Rabiah Ahmad Salwani Mohd Daud Eyas El-Qawasmeh (Eds.)
Informatics Engineering and Information Science International Conference, ICIEIS 2011 Kuala Lumpur, Malaysia, November 14-16, 2011 Proceedings, Part IV
13
Volume Editors Azizah Abd Manaf Advanced Informatics School (UTM AIS) UTM International Campus Kuala Lumpur, 54100, Malaysia E-mail:
[email protected] Shamsul Sahibuddin Advanced Informatics School (UTM AIS) UTM International Campus Kuala Lumpur, 54100, Malaysia E-mail:
[email protected] Rabiah Ahmad Advanced Informatics School (UTM AIS) UTM International Campus Kuala Lumpur, 54100, Malaysia E-mail:
[email protected] Salwani Mohd Daud Advanced Informatics School (UTM AIS) UTM International Campus Kuala Lumpur, 54100, Malaysia E-mail:
[email protected] Eyas El-Qawasmeh King Saud University Information Systems Department Riyadh, Saudi Arabia E-mail:
[email protected] ISSN 1865-0929 e-ISSN 1865-0937 ISBN 978-3-642-25482-6 e-ISBN 978-3-642-25483-3 DOI 10.1007/978-3-642-25483-3 Springer Heidelberg Dordrecht London New York Library of Congress Control Number: 2011941089 CR Subject Classification (1998): C.2, H.4, I.2, H.3, D.2, H.5 © Springer-Verlag Berlin Heidelberg 2011 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting, reproduction on microfilms or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Typesetting: Camera-ready by author, data conversion by Scientific Publishing Services, Chennai, India Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)
Message from the Chair
The International Conference on Informatics Engineering and Information Science (ICIEIS 2011)—co-sponsored by Springer—was organized and hosted by Universiti Teknologi Malaysia in Kuala Lumpur, Malaysia, during November 14–16, 2011 in association with the Society of Digital Information and Wireless Communications. ICIEIS 2011 was planned as a major event in the computer and information sciences and served as a forum for scientists and engineers to meet and present their latest research results, ideas, and papers in the diverse areas of digital information processing, digital communications, information security, information ethics, and data management, and other related topics. This scientific conference comprised guest lectures and 210 research papers for presentation over many parallel sessions. This number was selected from more than 600 papers. For each presented paper, a minimum of two reviewers went through each paper and filled a reviewing form. The system involves assigning grades to each paper based on the reviewers’ comments. The system that is used is open conference. It assigns grades for each paper that range from 6 to 1. After that, the Scientific Committee re-evaluates the paper and its reviewing and decides on either acceptance or rejection. This meeting provided a great opportunity to exchange knowledge and experiences for all the participants who joined us from all over the world to discuss new ideas in the areas of data and information management and its applications. We are grateful to Universiti Teknologi Malaysia in Kuala Lumpur for hosting this conference. We use this occasion to express thanks to the Technical Committee and to all the external reviewers. We are grateful to Springer for co-sponsoring the event. Finally, we would like to thank all the participants and sponsors. Azizah Abd Manaf
Preface
On behalf of the ICIEIS 2011 conference, the Program Committee and Universiti Teknologi Malaysia in Kuala Lumpur, I have the pleasure to present the proceedings of the International Conference on Informatics Engineering and Information Science’ (ICIEIS 2011). The ICIEIS 2011 conference explored new advances in digital information and data communications technologies. It brought together researchers from various areas of computer science, information sciences, and data communications to address both theoretical and applied aspects of digital communications and wireless technology. We hope that the discussions and exchange of ideas will contribute to advancements in the technology in the near future. The conference received more than 600 papers of which 530 papers were considered for evaluation. The number of accepted papers 210. The accepted papers were authored by researchers from 39 countries covering many significant areas of digital information and data communications. Each paper was evaluated by a minimum of two reviewers.
Organization
General Chair Azizah Abd Manaf
Universiti Teknologi Malaysia, Malaysia
Program Chair Ezendu Ariwa Mazdak Zamani
London Metropolitan University, UK Universiti Teknologi Malaysia, Malaysia
Program Co-chairs Yoshiro Imai Jacek Stando
Kagawa University, Japan Technical University of Lodz, Poland
Proceedings Chair Jan Platos
VSB-Technical University of Ostrava, Czech Republic
Publicity Chair Maitham Safar Zuqing Zhu
Kuwait University, Kuwait University of Science and Technology of China, China
International Program Committee Abdullah Almansur Akram Zeki Ali Dehghan Tanha Ali Sher Altaf Mukati Andre Leon S. Gradvohl Arash Habibi Lashkari Asadollah Shahbahrami Chantal Cherifi Craig Standing
King Saud University, Saudi Arabia International Islamic University Malaysia, Malaysia Asia Pacific University, Malaysia American University of Ras Al Khaimah, UAE Bahria University, Pakistan State University of Campinas, Brazil University Technology Malaysia (UTM), Malaysia Delft University of Technology, The Netherlands Universit´e de Corse, France Edith Cowan University, Australia
X
Organization
D.B. Karron Duc T. Pham E. George Dharma Prakash Raj Eric Atwell Estevam Rafael Hruschka Eyas El-Qawasmeh Ezendu Ariwa Fouzi Harrag Genge Bela Gianni Fenu Guo Bin Hamid Jahankhani Hend Al-Khalifa Hocine Cherifi Isamu Shioya Isao Nakanishi Jim Yonazi Jose Filho Juan Martinez Khaled A. Mahdi Kosuke Numa Ladislav Burita Laxmisha Rai Manjaiah D.H. Majid Haghparast Malinka Ivanova Martin J. Dudziak Mazdak Zamani Mirel Cosulschi Mohd Abd Wahab Monica Vladoiu Nan Zhang Nazri Mahrin Noraziah Ahmad Pasquale De Meo Paulino Leite da Silva Piet Kommers Prabhat Mahanti Rabiah Ahmad
Computer Aided Surgery and Informatics, USA Cardiff University, UK Bharathidasan University, India University of Leeds, UK Carnegie Mellon University, USA King Saud University, Saudi Arabia London Metropolitan University, UK UFAS University, Algeria University of Targu Mures, Romania University of Cagliari, Italy Institute Telecom & Management SudParis, France University of East London, UK King Saud University, Saudi Arabia Universit´e de Bourgogne, France Hosei University, Japan Tottori University, Japan The Institute of Finance Management, Tanzania University of Grenoble, France Gran Mariscal de Ayacucho University, Venezuela Kuwait University, Kuwait The University of Tokyo, Japan University of Defence, Czech Republic Shandong University of Science and Technology, China Mangalore University, India Islamic Azad University, Shahre-Rey Branch, Iran Technical University, Bulgaria Stratford University, USA Universiti Teknologi Malaysia, Malaysia University of Craiova, Romania Universiti Tun Hussein Onn Malaysia, Malaysia PG University of Ploiesti, Romania George Washington University, USA Universiti Teknologi Malaysia, Malaysia Universiti Malaysia Pahang, Malaysia University of Applied Sciences of Porto, Portugal ISCAP-IPP University, Portugal University of Twente, The Netherlands University of New Brunswick, Canada Universiti Teknologi Malaysia, Malaysia
Organization
Radhamani Govindaraju Ram Palanisamy Riaza Mohd Rias Salwani Mohd Daud Sami Alyazidi Shamsul Mohd Shahibudin Talib Mohammad Valentina Dagiene Viacheslav Wolfengagen Waralak V. Siricharoen Wojciech Mazurczyk Wojciech Zabierowski Yi Pan Zanifa Omary Zuqing Zhu Zuqing Zhu Zuraini Ismail
XI
Damodaran College of Science, India St. Francis Xavier University, Canada University of Technology MARA, Malaysia Universiti Teknologi Malaysia, Malaysia King Saud University, Saudi Arabia Universiti Teknologi Malaysia, Malaysia University of Botswana, Botswana Institute of Mathematics and Informatics, Lithuania JurInfoR-MSU Institute, Russia University of the Thai Chamber of Commerce, Thailand Warsaw University of Technology, Poland Technical University of Lodz, Poland Georgia State University, USA Dublin Institute of Technology, Ireland The University of Science and Technology of China, China University of Science and Technology of China, China Universiti Teknologi Malaysia, Malaysia
Reviewers Morteza Gholipour Geshnyani Asadollah Shahbahrami Mohd Faiz Hilmi Brij Gupta Naeem Shah Shanmugasundaram Hariharan Rajibul Islam Luca Mazzola K.P. Yadav Jesuk Ko Mohd Wahab Luca Mazzola Anirban Kundu Hamouid Khaled Muhammad Naveed Yana Hassim Reza Moradi Rad Rahman Attar Zulkefli Bin Mansor Mourad Amad Reza Ebrahimi Atani Vishal Bharti
University of Tehran, Iran University of Guilan, Iran Universiti Sains Malaysia, Malaysia Indian Institute of Technology, India Xavor Corporation, Pakistan B.S. Abdur Rahman University, India University Technology Malaysia, Malaysia Universit`a della Svizzera Italiana, Italy Acme College of Engineering, India Gwangju University, Korea Universiti Tun Hussein Onn Malaysia, Malaysia Universit`a della Svizzera Italiana, Italy West Bengal University of Technology, India Batna University, Algeria Iqra University, Pakistan Universiti Tun Hussein Onn Malaysia, Malaysia University of Guilan, Iran University of Guilan, Iran Universiti Teknologi MARA, Malaysia Bejaia University, Algeria University of Guilan, Iran Dronacharya College of Engineering, India
XII
Organization
Mohd Nazri Ismail Nazanin Kazazi Amir Danesh Tawfig Eltaif Ali Azim Iftikhar Ahmad Arash Lashkari Zeeshan Qamar N. Mohankumar Irfan Syamsuddin Yongyuth Permpoontanalarp Jorge Coelho Zeeshan Qamar Aurobindo Ogra Angkoon Phinyomark Subarmaniam Kannan Babak Bashari Rad Ng Hu Timothy Yap Tzen Vun Sophia Alim Ali Hussein Maamar Tong Hau Lee Rachit Mohan Hamma Tadjine Ahmad Nadali Kamaruazhar Bin Daud Mohd Dilshad Ansari Pramod Gaur Ashwani Kumar Velayutham Pavanasam Mazdak Zamani Azrina Kamaruddin Mazdak Zamani Rajendra Hegadi Javad Rezazadeh A.K.M. Muzahidul Islam Asghar Shahrzad Khashandarag
University of Kuala Lumpur, Malaysia University Technology Malaysia, Malaysia University of Malaya, Malaysia Photronix Technologies, Malaysia COMSATS Institute of Information Technology, Pakistan King Saud University, Saudi Arabia University Technology Malaysia, Malaysia COMSATS Institute of Information Technology, Pakistan Amrita Vishwa Vidyapeetham, India State Polytechnic of Ujung Pandang, Indonesia King Mongkut’s University of Technology, Thailand Polytechnic Institute of Porto, Portugal COMSATS Institute of Information Technology, Pakistan University of Johannesburg, South Africa Prince of Songkla University, Thailand Multimedia University, Malaysia University Technology of MalaysiaMalaysia Multimedia University, Malaysia Multimedia University, Malaysia University of Bradford, UK Faculty of Electronic Technology, Libya Multimedia University, Malaysia Jaypee University of Information Technology, India IAV GmbH, Germany Islamic Azad University, Iran Universiti Teknologi MARA, Malaysia Jaypee University of Information Technology, India Wipro Technologies, India Jaypee University of Information Technology, India Adhiparasakthi Engineering College, India Universiti Teknologi Malaysia, Malaysia UiTM Shah Alam, Malaysia Universiti Teknologi Malaysia, Malaysia Pragati College of Engineering and Management, India Universiti Teknologi Malaysia (UTM), Iran Universiti Teknologi Malaysia, Malaysia Islamic Azad University, Iran
Organization
Thaweesak Yingthawornsuk Chusak Thanawattano Ali AL-Mazari Amirtharajan Rengarajan Nur’Aini Abdul Rashid Mohammad Hossein Anisi
XIII
University of Technology Thonburi, Thailand Thailand AlFaisal University, Kingdom of Saudi Arabia SASTRA University, India Universiti Sains Malaysia, Malaysia Universiti Teknologi Malaysia (UTM), Malaysia Mohammad Nazir University Technology of Malaysia, Malaysia Desmond Lobo Burapha University International College, Chonburi, Thailand Salah Al-Mously Koya University, Iraq Gaurav Kumar Chitkara University, India Salah Eldin Abdelrahman Menoufia University, Egypt Vikram Mangla Chitkara University, India Deveshkumar Jinwala S V National Institute of Technology, India Nashwa El-Bendary Arab Academy for Science, Technology & Maritime Transport, Egypt Ashish Rastogi Guru Ghasidas Central University, India Vivek Kumar Singh Banaras Hindu University, India Sude Tavassoli Islamic Azad University, Iran Behnam Dezfouli University Technology Malaysia (UTM), Malaysia Marjan Radi University Technology Malaysia (UTM), Malaysia Chekra Ali Allani Arab Open University, Kuwait Jianfei Wu North Dakota State University, USA Ashish Sitaram Guru Ghasidas University, India Aissa Boudjella Jalan Universiti Bandar Barat, Malaysia Gouri Prakash HSBC Bank, USA Ka Ching Chan La Trobe University, Australia Azlan Mohd Zain Universiti Teknologi Malaysia, Malaysia Arshad Mansoor SZABIST, Pakistan Haw Su Cheng Multimedia University (MMU), Malaysia Deris Stiawan Sriwijaya University, Indonesia Akhilesh Dwivedi Ambedkar Institute of Technology, India Thiagarajan Balasubramanian RVS College of Arts and Science, India Simon Ewedafe Universiti Tun Abdul Rahman, Malaysia Roheet Bhatnagar Sikkim Manipal Institute of Technology, India Chekra Allani The Arab Open University, Kuwait Eduardo Ahumada-Tello Universidad Autonoma de Baja California, Mexico Jia Uddin International Islamic University Chittagong, Bangladesh Gulshan Shrivastava Ambedkar Institute of Technology, India Mohamad Forouzanfar University of Ottawa, Canada
XIV
Organization
Kalum P. Udagepola Muhammad Javed Partha Sarati Das Ainita Ban Noridayu Manshor Syed Muhammad Noman
BBCG, Australia Dublin City University, Ireland Dhaka University of Engineering, Bangladesh Universiti Putra Malaysia, Malaysia Universiti Putra Malaysia, Malaysia Sir Syed University of Engineering and Technology, Pakistan Zhefu Shi University of Missouri, USA Noraini Ibrahim Universiti Teknologi Malaysia (UTM), Malaysia Przemyslaw Pawluk York University, Canada Kumudha Raimond Addis Ababa University, Ethiopia Gurvan Le Guernic KTH- Royal Institute of Technology, Sweden Sarma A.D.N Nagarjuna University, India Utku Kose Afyon Kocatepe University, Turkey Kamal Srivastava SRMCEM, India Marzanah A. Jabar Universiti Putra Malaysia, Malaysia Eyas ElQawasmeh King Saud University, Saudi Arabia Adelina Tang Sunway University, Malaysia Samarjeet Borah Sikkim Manipal Institute of Technology, India Ayyoub Akbari Universiti Putra Malaysia, Malaysia Abbas Mehdizadeh Universiti Putra Malaysia (UPM), Malaysia Looi Qin En Institute for Infocomm Research, Singapore Krishna Prasad Miyapuram Universit` a degli Studi di Trento, Italy M.Hemalatha Karpagam University, India Azizi Nabiha Annaba University of Algeria, Algeria Mallikarjun Hangarge Science and Commerce College, India J. Satheesh Kumar Bharathiar University, India Abbas Hanon AlAsadi Basra University, Iraq Maythem Abbas Universiti Teknologi PETRONAS, Malaysia Mohammad Reza Noruzi Tarbiat Modarres University, Iran Santoso Wibowo CQ University Melbourne, Australia Ramez Alkhatib AlBaath University, Syrian Arab Republic Ashraf Mohammed Iqbal Dalhousie University, Canada Hari Shanker Hota GGV Central University, India Tamer Beitelmal Carleton University, Canada Azlan Iqbal Universiti Tenaga Nasional, Malaysia Alias Balamurugan Thiagarajar College of Engineering, India Muhammad Sarfraz Kuwait University, Kuwait Vuong M. Ngo HCMC University of Technology, Vietnam Asad Malik College of Electrical and Mechincal Engineering, Pakistan Anju Sharma Thapar University, India Mohammad Ali Orumiehchiha Macquarie University, Australia Khalid Hussain University Technology Malaysia, Malaysia
Organization
Parvinder Singh Amir Hossein Azadnia Zulkhar Nain Shashirekha H.L. Dinesh Hanchate Mueen Uddin Muhammad Fahim Sharifah Mastura Syed Mohamad Baisa Gunjal Ali Ahmad Alawneh Nabhan Hamadneh Vaitheeshwar Ramachandran Ahmad Shoara Murtaza Ali Khan Norshidah Katiran Haniyeh Kazemitabar Sharifah Mastura Syed Mohamad Somnuk Phon-Amnuaisuk Prasanalakshmi Balaji Mueen Uddin Bhumika Patel Sachin Thanekar Nuzhat Shaikh Safiye Ghasemi Nor Laily Hashim Joao Pedro Costa S. Parthasarathy Omar Kareem Jasim Balasubramanian Thangavelu Lee Chai Har Md Asikur Rahman Renatus Michael Shin-ya Nishizaki Sahadeo Padhye Faith Shimba Subashini Selvarajan
XV
Deenbandhu Chhotu Ram University of Science and Technology, India University Technology Malaysia (UTM), Malaysia American University, United Arab Emirates Mangalore University, India Vidypratishthan’s College Of Engineering, India Universiti Teknologi Malaysia (UTM), Malaysia Kyung Hee University, Korea Universiti Sains Malaysia, Malaysia Amrutvahini College of Engineering, India Philadelphia University, Jordan Murdoch University, Australia Tata Consultancy Services, India Farabi Higher Education Institute, Iran Royal University for Women, Bahrain Universiti Teknologi Malaysia, Malaysia Universiti Teknologi PETRONAS, Malaysia Universiti Sains Malaysia, Malaysia Universiti Tunku Abdul Rahman, Malaysia Bharathiar University, India Universiti Teknologi Malaysia, Malaysia CKPithawalla College of Engineering and Technology, India University of Pune, India MES College of Engineering, India Islamic Azad University, Iran Universiti Utara Malaysia, Malaysia University of Coimbra, Portugal Thiagarajar College of Engineering, India Maaref College University, Iraq SVM Arts and Science College, India Multimedia University (MMU), Malaysia Memorial University of Newfoundland, Canada The Institute of Finance Management, Tanzania Tokyo Institute of Technology, Japan Motilal Nehru National Institute of Technology, India The Institute of Finance Management, Tanzania Annamalai University, India
XVI
Organization
Valentina Emilia Balas Muhammad Imran Khan Daniel Koloseni Jacek Stando Yang-Sae Moon Mohammad Islam Joseph Ng Umang Singh Sim-Hui Tee Ahmad Husni Mohd Shapri Syaripah Ruzaini Syed Aris Ahmad Pahlavan Aaradhana Deshmukh Sanjay Singh Subhashini Radhakrishnan Binod Kumar Farah Jahan Masoumeh Bourjandi Rainer Schick Zaid Mujaiyid Putra Ahmad Abdul Syukor Mohamad Jaya Yasir Mahmood Razulaimi Razali Anand Sharma Seung Ho Choi Safoura Janosepah Rosiline Jeetha B Mustafa Man Intan Najua Kamal Nasir Ali Tufail Bowen Zhang Rekha Labade Ariffin Abdul Mutalib Mohamed Saleem Haja Nazmudeen Norjihan Abdul Ghani Micheal Arockiaraj A. Kannan Nursalasawati Rusli Ali Dehghantanha Kathiresan V. Saeed Ahmed Muhammad Bilal
University of Arad, Romania Universiti Teknologi PETRONAS, Malaysia The Institute of Finance Management, Tanzania Technical University of Lodz, Poland Kangwon National University, Korea University of Chittagong, Bangladesh University Tunku Abdul Rahman, Malaysia ITS Group of Institutions, India Multimedia University, Malaysia Universiti Malaysia Perlis, Malaysia Universiti Teknologi MARA, Malaysia Islamic Azad University, Iran Pune University, India Manipal University, India Sathyabama University, India Lakshmi Narain College of Technology, India University of Chittagong, Bangladesh Islamic Azad University, Iran University of Siegen, Germany Universiti Teknologi MARA, Malaysia Universiti Teknikal Malaysia Melaka, Malaysia NUST SEECS, Pakistan Universiti Teknologi MARA, Malaysia MITS, LAkshmangarh, India Seoul National University of Science and Technology, Korea Islamic Azad University, Iran RVS College of Arts and Science, India University Malaysia Terengganu, Malaysia Universiti Teknologi PETRONAS, Malaysia Ajou University, Korea Beijing University of Posts and Telecommunications, China Amrutvahini College of Engineering, India Universiti Utara Malaysia, Malaysia Universiti Tunku Abdul Rahman, Malaysia University of Malaya, Malaysia Loyola College, India K.L.N.College of Engineering, India Universiti Malaysia Perlis, Malaysia Asia-Pacific University, Malaysia RVS College of Arts and Science, India CIIT,Islamabad, Pakistan UET Peshawar, Pakistan
Organization
Ahmed Al-Haiqi Dia AbuZeina Nikzad Manteghi Amin Kianpisheh Wattana Viriyasitavat Sabeen Tahir Fauziah Redzuan Mazni Omar Quazi Mahera Jabeen A.V. Senthil Kumar Ruki Harwahyu Sahel Alouneh Murad Taher Yasaman Alioon Muhammad Zaini Ahmad Vasanthi Beulah Shanthi A.S. Siti Marwangi Mohamad Maharum Younes Elahi Izzah Amani Tarmizi Yousef Farhang Mohammad M. Dehshibi Ahmad Kueh Beng Hong Seyed Buhari D. Christopher NagaNandiniSujatha S Jasvir Singh Omar Kareem Faiz Asraf Saparudin Ilango M.R. Rajesh R. Vijaykumar S.D. Cyrus F. Nourani Faiz Maazouzi Aimi Syamimi Ab Ghafar Md. Rezaul Karim Indrajit Das Muthukkaruppan Annamalai Prabhu S. Sundara Rajan R. Jacey-Lynn Minoi Nazrul Muhaimin Ahmad Anita Kanavalli Tauseef Ali
XVII
UKM, Malaysia KFUPM, Saudi Arabia Islamic Azad University, Iran Universiti Sains Malaysia, Malaysia University of Oxford, UK UTP Malaysia, Malaysia UiTM, Malaysia UUM, Malaysia Saitama University, Japan Hindusthan College of Arts and Science, India Universitas Indonesia, Indonesia German Jordanian University, Jordan Hodieda University, Yemen Sharif University of Technology, Iran Universiti Malaysia Perlis, Malaysia Queen Mary’s College, India Loyola College, Chennai, India Universiti Teknologi Malaysia, Malaysia UTM, Malaysia Universiti Sains Malaysia, Malaysia Universiti Teknologi Malaysia, Malaysia IACSIT, Iran Universiti Teknologi Malaysia, Malaysia Universiti Brunei Darussalam, Brunei Darussalam RVS College of Arts and Science, India K.L.N. College of Engineering, India Guru Nanak Dev University, India Alma’arif University College, Iraq Universiti Teknologi Malaysia, Malaysia K.L.N. College of Engineering, India Bharathiar University, India RVS College of Arts and Science, India AkdmkR&D, USA LabGED Laboratory, Algeria Universiti Teknologi Malaysia, Malaysia Kyung Hee University, Korea VIT University, India Universiti Teknologi MARA, Malaysia Loyola College, India Loyola College, India Universiti Malaysia Sarawak, Malaysia Multimedia University, Malaysia M.S. Ramaiah Institute of Technology, India University of Twente, The Netherlands
XVIII
Organization
Hanumanthappa J. Tomasz Kajdanowicz Rehmat Ullah
University of Mangalore, India Wroclaw University of Technology, Poland University of Engineering and Technology, Peshawar, Pakistan Nur Zuraifah Syazrah Othman Universiti Teknologi Malaysia, Malaysia Mourad Daoudi University of Sciences and Technologies Houari Boumediene, Algeria Mingyu Lee Sugnkyunkwan University, Korea Cyriac Grigorious Loyola College, India Sudeep Stephen Loyola College, India Amit K. Awasthi Gautam Buddha University, India Zaiton Abdul Mutalip Universiti Teknikal Malaysia Melaka, Malaysia Abdu Gumaei King Saud University, Saudi Arabia E. Martin University of California, Berkeley, USA Mareike Dornh¨ ofer University of Siegen, Germany Arash Salehpour University of Nabi Akram, Iran Mojtaba Seyedzadegan UPM, Malaysia Raphael Jackson Kentucky State University, USA Abdul Mateen Federal Urdu University of Science and Technology, Pakistan Subhashini Ramakrishnan Dr G.R. Damodaran College of Science, India Randall Duran Singapore Management University, Singapore Yoshiro Imai Kagawa University, Japan Syaril Nizam University Technology Malaysia, Malaysia Pantea Keikhosrokiani Universiti Sains Malaysia, Malaysia Kok Chin Khor Multimedia University, Malaysia Salah Bindahman Universiti Sains Malaysia, Malaysia Sami Miniaoui University of Dubai, United Arab Emirates Intisar A.M. Al Sayed Al Isra University, Jordan Teddy Mantoro International Islamic University Malaysia, Malaysia Kitsiri Chochiang PSU University, Thailand Khadoudja Ghanem University Mentouri Constantine, Algeria Rozeha A. Rashid Universiti Teknologi Malaysia, Malaysia Redhwan Qasem Shaddad Taiz University, Yemen MuhammadAwais Khan COMSATS Institute of Information and Technology, Pakistan Noreen Kausar Universiti Teknologi PETRONAS, Malaysia Hala Jubara UTM, Malaysia Alsaidi Altaher Universiti Sains Malaysia, Malaysia Syed Abdul Rahman Al-Haddad Universiti Putra Malaysia, Malaysia Norma Alias Universiti Teknologi Malaysia, Malaysia Adib M. Monzer Habbal University Utara Malaysia, Malaysia Heri Kuswanto Institut Teknologi Sepuluh Nopember, Indonesia
Organization
Asif Khan Tufail Habib Amin Shojaatmand Yasser K. Zahedi Vetrivelan N. Khalil Ullah Amril Syalim Habib Ullah Michal Kratky Suyeb Khan Heng Yaw Ling Zahid Mahmood Sebastian Binnewies Mohammadreza Khoei Zahid Mahmood Thawanrat Puckdeepun Wannisa Matcha Sureena Matayong Sapna Mishra Qaim Mehdi Rizvi Habib Ullah
XIX
FAST NUCES Peshawar Campus, Pakistan Aalborg University, Denmark Islamic Azad University, Iran Universiti Teknologi Malaysia, Malaysia Periyar Maniammai University, India National University of Computing and Emerging Sciences, Pakistan Kyushu University, Japan COMSATS Institute of IT, Pakistan VSB-Technical University of Ostrava, Czech Republic Electronics and Communication Engineering, India Multimedia University, Malaysia COMSATS, Institute of Information Technology, Pakistan Griffith University, Australia Universiti Teknologi Malaysia, Malaysia COMSATS IIT, Pakistan Universiti Teknologi PETRONAS, Malaysia Universiti Teknologi PETRONAS, Malaysia Universiti Teknologi PETRONAS, Malaysia Dayanand Academy of Management Studies, India SRMCEM, India COMSATS Institute of Information Technology, Wah Campus, Pakistan
Table of Contents – Part IV
Information and Data Management Risk Management Determinants in Vendor Selection of IT Tender Evaluation Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Syaripah Ruzaini Syed Aris, Haryani Haron, and Nor Azlin Murat A Survey on Data Integration in Bioinformatics . . . . . . . . . . . . . . . . . . . . . . Cheo Thiam Yui, Lim Jun Liang, Wong Jik Soon, and Wahidah Husain
1 16
Structural Approach to the Formation of Information Culture of Individuals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rasim Alguliev and Rasmiya Mahmudova
29
Smart Card Based on Hash Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zulkharnain and Ali Sher
41
RDF Data Models in Oracle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zdenka Telnarova
52
Order Processing in Supply Chain Management with Developing an Information System Model: An Automotive Manufacturing Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mohammad Reza Khoei, Misam Kashefi, Pezhman Ghadimi, Amir Hossein Azadnia, Mat Rebi Abdul Rani, and Morteza Lalmazloumian Spatial Information Databases Integration Model . . . . . . . . . . . . . . . . . . . . Mustafa Man, Mohd. Shafry Mohd Rahim, Mohammad Zaidi Zakaria, and Wan Aezwani Wan Abu Bakar
65
77
Web Applications and Software Systems A Framework of Hybrid Semantic Speech Query via Stemmer for Quran Documents Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mohd Amin MohdYunus, Roziati Zainuddin, and Noorhidawati Abdullah A Semantic Layer for a Peer-to-Peer Based on a Distributed Hash Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mohammed Ammari, Dalila Chiadmi, and Laila Benhlima
91
102
XXII
Table of Contents – Part IV
Design and Development of a Ubiquitous Cancer Care System: A Collaborative Communication among Cancer Community . . . . . . . . . . . Ariffin Abdul Mutalib and Norlaily Hashim
115
ICA-WSS: A QOS-Aware Imperialist Competitive Algorithm for Web Service Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shahriar Lotfi and Khalil Mowlani
135
Flexi-adaptor: A Nobel Approach for Adapting Web Content for Mobile Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rajibul Anam, Chin Kuan Ho, and Tek Yong Lim
149
Multimedia Performance Evaluation of the Developed Automatic Musical Note Translating System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chooi Ling Si Toh, Chee Kyun Ng, and Nor Kamariah Noordin
164
Motion Tweening for Skeletal Animation by Cardinal Spline . . . . . . . . . . . Murtaza Ali Khan and Muhammad Sarfraz
179
Preliminary Investigation on the Use of Augmented Reality in Collaborative Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wannisa Matcha and Dayang Rohaya Awang Rambli
189
Ad Hoc Networks Development of 6LoWPAN Adaptation Layer with Fragmentation and Reassembly Mechanisms by Using Qualnet Simulator . . . . . . . . . . . . . . . . . Chiaw Wei Chan, Gee Keng Ee, Chee Kyun Ng, Fazirulhisyam Hashim, and Nor Kamariah Noordin
199
Mobile Computing Performance Evaluation of Switching between Routing Protocols in Mobile Ad Hoc Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Annapurna P. Patil, K. Rajani Kanth, Desai Neel, N. Ganesh, and P. Amala
213
Miscellaneous Topics in Digital Information and Communications Processing of Multidimensional Range Query Using SIMD Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Peter Chovanec and Michal Kr´ atk´y
223
Table of Contents – Part IV
A Comparative Evaluation of Web Service Discovery Approaches for Mobile Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nor Azizah Saadon and Radziah Mohamad Planning and Project Control with EXCEL Software in Asalouyeh 5th Olefin Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Adnan Keshvari, Hamed Mirzapour, Jalil Ebrahimpour, and Nikzad Manteghi
XXIII
238
253
Partial Plagiarism Detection Using String Matching with Mismatches . . . Tetsuya Nakatoh, Kensuke Baba, Yasuhiro Yamada, and Daisuke Ikeda
265
A Surveillance System for a Farm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-Ram Kim, Sang-Yong Rhee, Seung-Joon Seok, Hong-Chang Lee, and Young-Baek Kim
273
Privacy in Health Information Systems: A Review . . . . . . . . . . . . . . . . . . . . Salah Bindahman and Nasriah Zakaria
285
The Genetics Lab: An Innovative Tool for Assessment of Intelligence by Mean of Complex Problem Solving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cyril Hazotte, H´el`ene Mayer, Younes Djaghloul, Thibaud Latour, Philipp Sonnleitner, Martin Brunner, Ulrich Keller, Eric Francois, and Romain Martin Energy Efficient Mutually Exclusive Path Finding Protocol for Wireless Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anita Kanavalli, G.P. Bharath, B.N. Nuthan Prasad, P. Deepa Shenoy, K.R. Venugopal, and L.M. Patnaik
296
311
Conceptual Design System for Monitoring Electrical Loads . . . . . . . . . . . . Murad Ahmed Ali Taher and Ali Abdo Mohammed Al-Kubati
321
Part of Speech Tagging Approach to Designing Compound Words for Arabic Continuous Speech Recognition Systems . . . . . . . . . . . . . . . . . . . . . . Dia AbuZeina, Moustafa Elshafei, and Wasfi Al-Khatib
330
A Discriminative Non-linear Manifold Learning Technique for Face Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bogdan Raducanu and Fadi Dornaika
339
Investigation of Collaborative Interaction Using ITB-Based Setting vs. Computer-Based Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thawanrat Puckdeepun, Jafreezal B. Jaafar, Mohd Fadzil B. Hassan, and Fawnizu Azmadi B. Hussin
354
XXIV
Table of Contents – Part IV
Point-to-Point Communication on Gigabit Ethernet and InfiniBand Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Roswan Ismail, Nor Asilah Wati Abdul Hamid, Mohamed Othman, Rohaya Latip, and Mohd Azizi Sanwani
369
Detecting Unknown Anomalous Program Behavior Using API System Calls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Md. Rafiqul Islam, Md. Saiful Islam, and Morshed U. Chowdhury
383
Taxonomical Classification of Closely Related Reads of Genus Bacillus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wenmin Wang
395
System Level Performance Analysis Design Patterns . . . . . . . . . . . . . . . . . . Carolyn Pe Rosiene
405
Synthesizing 3D Face Shapes Using Tensor-Based Multivariate Statistical Discriminant Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jacey-Lynn Minoi, Carlos Eduardo Thomaz, and Duncan Fyfe Gillies
413
Automotive Camera Integration: Technologies and Limits . . . . . . . . . . . . . Hadj Hamma Tadjine and Karsten Schulze
427
Parallel versus Perpendicular Plots: A Comparative Study . . . . . . . . . . . . Raja Jamilah Raja Yusof, Roziati Zainuddin, and Zulkifli Mohd Yusoff
436
Intelligent Guard: A Novel Approach toward Software Protection . . . . . . Arash Salehpour, Mohammad Etemad, and Morteza Mokhtari Nazarlu
449
Mining Data Stream from a Higher Level of Abstraction: A Class Window Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Abdullah-Al-Mamun, Md. Anowarul Abedin, Md. Al Arman, M.A. Mottalib, and Mohammad Rezwanul Huq
461
Development of Water Quality Monitoring System Prototype for Fresh Water Fish Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. Razzemi, Z. Abdul Halim, M. Kusairay Musa, and K. Hasbullah
470
A Dynamic Timestamp Discrepancy against Replay Attacks in MANET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nourddine Enneya, Aziz Baayer, and Mohammed ElKoutbi
479
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
491
Risk Management Determinants in Vendor Selection of IT Tender Evaluation Process Syaripah Ruzaini Syed Aris, Haryani Haron, and Nor Azlin Murat Faculty of Computer and Mathematical Sciences Universiti Teknologi MARA Malaysia {ruzaini,haryani}@salam.uitm.edu.my,
[email protected]
Abstract. This study reports on the determinants of Information Technology tender activity focusing on vendor selection process. The evaluation process in the selection of vendors is crucial because this is the phase where risks occur frequently. A survey is conducted among government agencies, governmentlinked companies and private sectors which are involved in IT procurement. Sixty percent of the population responded. A factor analysis is performed and four risk categories are extracted with a total of 11 determinants identified. The four categories of risks are vendor financial stability, hazards, operation and strategic. Determinants for vendor financial stability include pricing and the determinants for operation are performance, and deliverable. Meanwhile, reporting, customer demand, business enabler and supply chain are the determinants for strategic and agreement, security and support are the determinants for hazard. This study contributes towards new knowledge by proposing a risk management model for vendor selection in IT Tender Evaluation process. Keywords: IT Governance, Risk management, Determinants for vendor Selection, IT Tender Evaluation.
1 Introduction IT outsourcing is becoming a trend today. Governments are actively involve in IT outsourcing because it allows them to operate more effectively (McDougal, 2003). Malaysian government also strongly supports the idea of outsourcing. Malaysia is ranked as the third attractive location for offshore outsourcing in 2003. The achievement continues until recently and a report from ATKearney (2009) mentioned that the trend of IT outsourcing in Malaysia will continue to grow. One of the important criteria in IT outsourcing is to find the right vendor (Syaripah Ruzaini et al, 2008; Handfield, 2006). The chosen of the right vendor will increase the probability of IT outsourcing success. The vendor should be selected properly to ensure they are the best vendor that is able to deliver accordingly. This paper focuses on the tender evaluation process. According to Department of Treasury and Finance (2006), tender evaluation is crucial and critical because it not only provide benefit to both organization and vendor, but also affect their reputations in the future. Despite A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 1–15, 2011. © Springer-Verlag Berlin Heidelberg 2011
2
S.R.S. Aris, H. Haron, and N.A. Murat
the facts that this activity is important, there are many researchers that mentioned that there are many risks and problems associated with this activity. Among the risk include are not being transparent in the decision making (Suraya Yusof, 2006) and corruption (The Central Vigilance Commission, 2010). NSW Department of Commerce Tendering Manual (2004) has described that risk management should always be used in a tender process as it will minimize problems associated with it. However, step-by-step guidelines in evaluating tender are in limited number. The guidelines for the determinants in the vendor selection process are also in limited number. Hence, it is very important for a research to be carried out in this area especially regarding managing risk in tender evaluation activities. Therefore, it is the intention of this paper to report the potential determinants affecting IT tender process. This paper starts with introduction and literature review that will discuss on IT outsourcing, tender evaluation process, risk in tender evaluation process, managing risk in the tender evaluation process and determinants used in vendor selection criteria. Quantitative analysis has been conducted to confirm the determinants of vendor selection in IT tender evaluation process from Malaysian perspective. This study also contributes towards new knowledge by proposing a risk management framework for vender selection in IT tender evaluation process.
2 Literature Review Literature review will discuss about IT outsourcing, tender evaluation process, risk in tender evaluation process, managing risk in the tender evaluation process and determinants used in vendor selection criteria. 2.1 IT Outsourcing IT outsourcing is not a new phenomena. It has started in 1960s whereby the focusing is on the hardware and the approach in on service and facility management. In late 1990a, outsourcing is widely practiced and total solution of services are shifted to a vendor, rather than done in house (Jae et al., 2003). IT outsourcing can be defined as the practice of commissioning part or all of an organization’s IT assets, people, and/or activities to one or more external providers. Some of the activities being outsourced include system planning, application development, operation and maintenance, facilities management and so on (Dhar & Balakrishnan, 2006). In Malaysia, the pioneer of IT outsourcing is Bumiputra-Commerce Bank (BCB) (recently known as CIMB) that took its step to outsource its IT function to Electronic Data System (EDS) after Central Bank of Malaysia (Bank Negara) has been pushing local banks to seek out outsourcing partners to handle non-critical functions of the business (ITWorld, 2002). Since then, Malaysian government had outsourced many of IT project including Smart School, Human Resources Management Information System (HRMIS) and e-procurement project (Suhaimi, 2005). In ability to manage the IT outsourcing activity, researchers have come up with ways to govern IT outsourcing which include choosing the right vendor. Researchers have highlighted that choosing the right vendor in IT outsourcing in very important and it can increase the probability of IT outsourcing success (Handfield, 2006; Syaripah Ruzaini et al., 2008). While
Risk Management Determinants in Vendor Selection of IT Tender Evaluation Process
3
choosing for the right vendor, the organization should thoroughly review the tender evaluation process. 2.2 Tender Evaluation Process Tender is a purchasing procedure whereby potential vendor are invited to make a firm and unequivocal offer of the price (Douh, 2009). McGeorge and Adam (2003) describe tendering as a call for tender or merely ‘Invitation to treat’. According to Malaysia Treasury Circular Letter: “Surat Pekeliling Perbendaharaan Bil 8 Tahun 1995: Panduan Am Perolehan Kerajaan”, there are three main types of public procurement namely tender, quotation and direct purchase. Tender is for the procurement’s value more than RM 500,000 and any vendors which have registered with Ministry of Finance (MOF) are able to participate. In quotation, procurement value must be in the range of RM 50,000 to RM 500,000 and registered vendors are invited to participate. Meanwhile, in direct purchase, the procurement process is only involving potential vendor. Evaluation of tenders is the process to assess and compare tenders in the frame of tendering. It is very critical phase as it involves many activities such as receiving, opening, examination, comparison and classification of bids, reporting, and recommendation of the most appropriate vendor (Douh, 2009). The process involve in evaluation of tenders are the assessing or scoring, comparing and ranking bids. Each phase is given a score and finally ranked in order of merit. Tenders are scores according to the information provided, and tender evaluation process ends with a report, written by the panel. The report recommends the suitable vendor with regards to the legal rules and principles. The report is then submitted to tender committee for approval. Generally, the tender with high total score is recommended as winner. 2.3 Risk in Tender Evaluation Process Carmarthenshire County Council in Wales (2004) has viewed risk in procurement process into two (2) which are strategic risk (Political, Economic, Social, Technological, Legislative, Competitive and Customer/citizen) and operational risk (Professional, Financial, Legal, Physical, Contractual, Technological and Environmental). IT procurement falls into technological (in both strategic and operational risk). For strategic risk, it concern about those associated with a capacity to deal with the pace/scale of technological change or its ability to use technology to address changing demands such as a failure to procure the appropriate software to allow the for efficient financial management of the authority. For operational risk, it concerns is about those related to a reliance on operational equipment such as exclusive reliance on an e-procurement system to deliver critical supply acquisition. Rohana Othman et al (2010) shows that there are three (3) stages of risk occur in procurement processes which are Stage 1 risks (Inadequate choices of procedures, Lack of adequate assessment and planning, Insufficient timeframe preparation and Inconsistent practices across the bidders in first stage), Stage 2 risks (The selection of suppliers had inconsistent bidders’ information, Conflict of interest situations were prevalent, Lack of access to record procedures, Biased towards favoured bidders and Collusive bidding resulting in incorrect prices) and Stage 3 risks (Contract
4
S.R.S. Aris, H. Haron, and N.A. Murat
administration had insufficient monitoring process, No transparency in contracts being awarded, Lack of work or goods verification and Lack of separation of financial duties especially involving the payment process). Out of these three stages of risk, she claims that the most common risk is in stage 2, where the selection of supplier led to the award of a procurement contract to a vendor. At this stage, cost is not a major criterion to select the vendor but the availability and credibility of vendor to complete the task given are measured. 2.4 Managing Risk in the Tender Evaluation Process Risk management is the process of identifying risks, analysing their consequences and devising appropriate responses. Risk management should always be used in a tender process. The effort expended in managing risks in a tender process should be consistent with the estimated procurement cost and complexity, significance and nature of the process (NSW Department of Commerce Tendering Manual, 2004). Dependant on the nature, cost and complexity of the procurement, the evaluation committee may need to develop a risk management plan. An assessment of risk should be undertaken to provide greater certainty that this tendering and contracting process will produce a successful outcome. 2.5 Determinants Used as Vendor Selection Criteria From the literature, this research identifies several criteria which are commonly used in vendor selection in tender evaluation process which are performance, deliverable, supply chain, financial, support, confidentiality, business enables, reporting, customer demand, resource, security and agreement. The descriptions of each criteria and its author is as tabulated in Table 1. Table 1. Determinants in Vendor Selection Process Criteria Performance
Deliverable
Description Performance has been identified as one of the critical areas contributing to customer/vendor relationship in achieving outsourcing success (McFarlan & Nolan, 1995). Encyclopedia of Business (2010) describe performance as when vendors are given standing, status, or title according to their attainment of some level of performance, such as delivery, lead time, quality, price or some combination of variables. By using this definition, time, quality, vendor’s skill, availability and sustainability are used as a performance’s variable. The deliverable in IT outsourcing can either be services or product. According to European Commission (2006), services are “all services related to information technology, such as (but not limited to) training, consultancy, removal, logistics, integration work, engineering, development, maintenance and writing of documentation”. Meanwhile products are “any software, hardware or telecommunications product”. It is important to include deliverable as the determinant in vendor selection because it allows the specific vendor with the right skill to work on the project.
Risk Management Determinants in Vendor Selection of IT Tender Evaluation Process Table 1. (continued) Supply chain
Financial
Support
Confidentiality
Business enabler
Reporting
Customer demand
Wu and Olson (2008) described supply chain as network of vendors, manufacturers, distributors and retailers that are connected by transportation, information, and financial infrastructure. Current supply chain technology includes potentially many vendors linked to one or more manufactures and/or assemblers, who often use multiple distributors to supply many retailers. Hence, supply chain should be part of the determinants. Technology Business Research Inc. (TBR) describe that financial stability is a strong evidence of a company’s track record in meeting customer requirements, as well as an indicator of future performance. Therefore, in order to select for the best vendor, organization should review their financial stability and business seeking stability. Every purchasing has come with after sales services as a maintenance or support for it business. Technical support objective is to resolve the customer’s problem as efficient as possible (Qualstar, 2010). Therefore, vendor must provide efficient, effective and reliable technical support in order to align good service/product given. Therefore, this element should be considered as the determinants in vendor selection. Transparency and accountability are keys for enhancing integrity throughout the whole procurement cycle, including in needs assessment and contract management (OECD, 2007). They also mentioned that lack of transparency and accountability were recognized as a major threat to integrity in public procurement. Therefore, it is important for this criterion to one of the determinants. Prakash (2009) describe that the strategic use of IT systems helps an organization to review and align the performance of various business functions with the organization’s strategy by tracking and analyzing key performance index via end-to-end integrated systems. It also helps in finding out the causes of underperformance, take action to reduce costs and optimize profitability with the various business areas such as sales, production, customer, services and like (Prakash, 2009). AIRMIC, IRM & ALARM (2002) defines that formal reporting should address the control methods, the process used to identify risks and they are addressed by the risk management system, the primary control systems in place to manage significant risks and the monitoring and review system in place. MAMPU (2009) describe that monitoring and reviewing systems are the best way to avoid risk in the IT project. Therefore, it is important that reporting to and the reporting ability to be the determinants in vendor selection. For the purpose of this research, customer demand represents requirement from client organization to vendor. In tender document, the term can be represented trough Request for Information (RFI). Sourcing4oem.com (2010) defines RFI as “a request for information of various kinds from different vendors, most likely with the aim for future deliveries of products or services. The RFI is normally a document containing a description on why the information is requested and a specification to what information that actually is requested”. By using RFI, vendor are able to capture the requirements from the client orgnization (Method123TM,2003).
5
6
S.R.S. Aris, H. Haron, and N.A. Murat Table 1. (continued) Resources
Security
Agreement
Resource in IT project include time, money, people, tools, technology (Marchewka, 2010). It is very important to make sure that vendor have enough resources to complete the IT project. Therefore, this criterion is very important. Security has been number one issues in IT outsourcing (Syaripah Ruzaini, 2010). Therefore, it is very importance that vendor has a formal security procedure so that actual or perceived breach of confidentiality such as complaints or mistrust from vendor can be avoided. Unwillingness of the supplier to accept the contract makes delays in delivery and need to restart procurement. The other risks are failure of either party to fulfill the conditions of the contract, loss of intellectual property, failure to meet liabilities of third parties and unauthorized increased in scope of work. This is the common risk in project and the contract agreement between organization and vendor are the best way to legally bind the two parties.
3 Methodologies The objective of this study is to identify the potential determinants affecting IT outsourcing tender process. In ability to capture the potential determinant, research model adapted and modified from AIRMIC, ALARM & IRM (2002) was used. The research model is as shown in Figure 1. To achieve the objective of this study, quantitative method has been applied. In order to test the reliability of the questionnaire, pilot study was conducted. A sample of 20 organizations was chosen randomly to answer the questionnaire. The cronbach’s alpha values are more than 0.7. Therefore, it can be concluded that the scales for these constructs were deemed to exhibit an adequate reliability. The questionnaires were then distributed to 150 organizations with 48 returned back questionnaires. From Figure 1, the hypotheses are as follows: H1 H2 H3 H4 H5 H6 H7 H8 H9 H10 H11 H12 H13 H14 H15 H16 H17
: : : : : : : : : : : : : : : : :
There is a relationship between reporting and strategic There is a relationship between supply chain and strategic There is a relationship between business enabler and strategic There is a relationship between customer demand and strategic There is a relationship between performance and operation There is a relationship between deliverable and operation There is a relationship between resource and operation There is a relationship between stability and financial There is a relationship between pricing and financial There is a relationship between agreement and hazard There is a relationship between security and hazard There is a relationship between support and hazard There is a relationship between confidentiality and hazard There is a relationship between strategic and risk management There is a relationship between operation and risk management There is a relationship between financial and risk management There is a relationship between hazard and risk management
Risk Management Determinants in Vendor Selection of IT Tender Evaluation Process
Customer Demand
7
Business Enabler Supply Chain
Reporting
H4
H3
STRATEGIC
H2
H1
Stability
Performance
H14
H8
H5 H16
H15
RISK MANAGEMENT
FINANCIAL
OPERATION
H9
H7
Resource
H6 H17
Pricing H10
Deliverable H12
HAZARD H11
Agreement
Security
Support
H13
Confidentiality
Fig. 1. The research model
4 Results and Findings 4.1 Demographic Profile Table 2 shows the demographic profile of the respondent. From Table 2, it can be seen that the respondent gender is equally distributed by 50%. 54.2% of the respondents are from the government sector, followed by the private sector (18.8%). Table 2 also shows that most of the respondents have more than 4 years of working experience. Table 2. Demographic Profile Gender Male Female Company Sector Government Semi-Government Government Link Company (GLC) Private Years of Experience Less than 3 years 4-10 years 11-20 years
Number 24 24
Percent (%) 50 50
26 8 5 9
54.2 16.7 10.4 18.8
15 29 4
31.3 60.4 8.3
The analysis of the data also revealed that the respondents have involve in IT procurement activity and their roles include as a decision maker, technical committee, finance committee, procurement committee and project manager. They were also
8
S.R.S. Aris, H. Haron, and N.A. Murat
asked about the guidelines that have been used in the tender evaluation process. From the analysis, the top three guidelines are: 1. 2. 3.
Arahan Perbendaharaan Surat Pekeliling Perbendaharaan/Akta Kewangan 1967. Financial Guideline/Procurement Guideline Government Procedure and Internal Group Objective
Acara
The analysis shows that although guidelines are available, there are no standardized guideline to be used and most of the guideline are prepared by the centre agencies that control the procurement process. No specific attention were given to the tender evaluation process. This finding shows that there is a constraint in current practices. Therefore, the next section describe about the potential determinants that should be considered in the vendor evaluation process. 4.2 Risk Management Determinants in Vendor Selection Process In ability to come up with the determinants, this paper firstly conducted a factor analysis. The aim of factor analysis is to confirm the construct validity of the scales. Then, Spearman Correlation is computed to identify the correlation between the variables. The detail discussions on the findings are as follow. Factor Analysis. A factor analysis was done to determine the construct validity. The KMO and Bartlett’s test are done for each risk categories and gave a significant value of 0.00 for each category and extracted 20 factors from strategic, 16 factors from operation, 9 factors from financial and 20 factors from hazard. Strategic. Strategic category has four determinants which are reporting, customer demand, business enabler and supply chain. Using Kaiser’s criterion, the results of factor analysis is shown in Table 3. The first factors consist of three items, second factors consist of eight items, third factors consist of five items and last factor consist of four items. Factor loading of less than 0.6 was removed and reliability test was done again for each factor. The results of the reliability are as in Table 4. The results exceeding the minimum alpha of 0.6, thus the construct measures are deemed reliable (Hair et al, 1998). Table 3. Factor analysis for Strategic Category
Supplier performance report is available Vendor evaluation system procedure is available to monitor vendor's performance Vendor evaluation system procedure are documented
0.789 0.913 0.90
Supply Chain
Business Enabler
Customer Demand
Reporting
Component
Risk Management Determinants in Vendor Selection of IT Tender Evaluation Process
9
Table 3. (continued) RFI included: RFI: vendor reputation RFI: vendor history RFI: resource distribution RFI: experience-customer environment RFI: project budget RFI: customer-base/references RFI: subcontractors’/partner IT management understand business needs Business management recognize that IT project is critical to the organization's success Business management view IT project spending as a strategic investment Business and IT management jointly contribute to strategic planning of IT project IT project is perceived as bridging value to the business RFP included: RFP: Vendor background RFP: Financial performances RFP: Project staff and management
0.896 0.911 0.84 0.859 0.748 0.740 0.734 0.720 0.847 0.669 0.862 0.833 0.853 0.902 0.774
Table 4. Reliability Analysis for Strategic Factor Factor Reporting Customer demand Business Enabler Supply Chain
Cronbach’s Alpha 0.857 0.927 0.811 0.865
N of item 3 8 5 4
Operation. Operation category has three factors which are performance, deliverable and resource. Using Kaiser’s criterion, the results of factor analysis is shown in Table 5. The first factors consist of seven items, second factors consist six of items and third factor consist of three items. Factor loading less than 0.6 was suppressed and reliability test has been done again to each factor. A Cronbach’s Alpha values more than 0.7 was taken and the results are as in Table 6. The Cronbach’s Alpha for performance and deliverable factors are exceeding the minimum alpha of 0.6, thus the construct measures are deemed reliable (Hair et al, 1998). However, Cronbach’s Alpha of resources is less than 0.6. Therefore, this item is removed and not included for further investigation.
10
S.R.S. Aris, H. Haron, and N.A. Murat Table 5. Factor Analysis for Operation
Project team skill Project team availability Contribution of vendor's skill or expertise to organization Project timeline On-time delivery Delivered product/service on time Service performance Delivered correct product/service Suitable product or service obtained Quality performance of deliveries Quality of the product or service Team collegiality Vendor sustainability Vendor provide the full need of equipment and facilities Description of the vendor's financial equipment Particular of the vendor's average annual staff for the past three years.
Resource
Deliverable
Performance
Component
0.879 0.831 0.649 0.623 0.671 0.774 0.683 0.793 0.711 0.942 0.920 0.763 0.803 0.704 0.651 0.79
Table 6. Reliability Analysis for Operation Factor Performance Deliverable * Resource
Cronbach’s Alpha 0.930 0.827 0.623
N of item 7 6 3
*Note : Item removed
Financial. Financial category has two factors which are stability and pricing. Using Kaiser’s criterion, the results of factor analysis is shown in Table 7. The first factors consist of five items and the second factors consist of four items. Factor loading of less than 0.6 was suppressed and reliability test were computed again to each factor. A Cronbach’s Alpha value more than 0.7 was taken and the results is as in Table 8. The results exceeding the minimum alpha of 0.6, thus the construct measures are deemed reliable (Hair et al, 1998).
Risk Management Determinants in Vendor Selection of IT Tender Evaluation Process
11
Table 7. Factor Analysis for Financial
A company's balance sheet Statement of turnover for the past years transactions Financial position: sales Financial position: profitability Financial position: return of investment (ROI) Financial position: debt ratio Financial position: transparency of finances Lowest cost Agreement on prices avoid a cost variation
Pricing
Stability
Component
0.727 0.637 0.790 0.892 0.850 0.814 0.752 0.634 0.687
Table 8. Reliability’s analysis for financial Factor Stability Pricing
Cronbach’s Alpha 0.893 0.720
N of item 5 4
Hazard. Hazard has four determinants which are agreement, security, support and confidentiality. Using Kaiser’s criterion, the results of factor analysis is shown in Table 9. Table 9. Factor Analysis for Hazard
Contract awarding Clear specification Project pricing Project ownership Scope of service Project management procedure SLA The security of service level Security breach
0.896 0.834 0.905 0.751 0.888 0.867 0.832 0.866 0.873
Confidentiality
Support
Security
Agreement
Component
12
S.R.S. Aris, H. Haron, and N.A. Murat Table 9. (continued) How often must organization provide staff with appropriate technical skill training Technical assistant After sales service Warranty Documentation: Specific technical support procedure Documentation: Detail response time Vendor provide warranty services for the product Warranty are offered in respect of specification of the system/product Information sharing between organization and vendor Variety of offers avoid a biased specification Are conflicts among project committee in IT project selection occurring?
0.782 0.893 0.844 0.754 0.859 0.75 0.782 0.705 0.696 0.693 -0.604
The first factors consist of seven items, second factors consist five of items, third factor consist of five items and the last items consists of three items. Factor loading of less than 0.6 was removed and reliability test has been done again to each factor. A Cronbach’s Alpha values more than 0.7 was taken and the results are as in Table 10. The Cronbach’s Alpha for agreement, security and support factors are exceeding the minimum alpha of 0.6, thus the construct measures are deemed reliable (Hair et al, 1998). However, Cronbach’s Alpha of confidentiality is less than 0.6. Therefore, this item is removed and not included for further investigation. Table 10. Reliability’s analysis for hazard Factor Agreement Security Support * Confidentiality
Cronbach’s Alpha 0.942 0.932 0.857 0.689
N of item 7 5 5 3
*Note : Item removed
Spearman’s Correlation. In order to determine the correlation between the variables, correlation test was conducted. Before the hypotheses were tested, normality test were computed to measure the normality of the data. Since the sample size is less than 100, Shapiro–Wilks statistics is used (Coakes, 2005). p value is 0.000 thus proves that the distribution of the data is not normal. Therefore, Spearman’s correlation is used to analyze the hypotheses. The result is as tabulated in Table 11.
Risk Management Determinants in Vendor Selection of IT Tender Evaluation Process
13
Table 11. Spearman’s Correlation Result between Risks Type with Determinants Risk Type Strategic
Operation Financial Hazard
Determinants
Correlations
Reporting Supply Chain Business Enabler Customer Demand Performance Deliverable Stability Pricing Agreement Security Support
0.561 0.468 0.433
Sig. Value 0.000 0.001 0.002
Interpretation
-0.441
0.002
Negative moderate correlation
0.649 0.695 0.873 0.626 0.503 0.721 0.658
0.000 0.000 0.000 0.000 0.000 0.000 0.000
Positive moderate correlation Positive moderate correlation Positive strong correlation Positive moderate correlation Positive moderate correlation Positive strong correlation Positive moderate correlation
Positive moderate correlation Positive moderate correlation Positive moderate correlation
Spearman’s correlation test was also carried out to measure the relationship between financial, hazard, operation and strategic with risk management. The results of the hypotheses are as shown in Table 12. From the analysis of the hypotheses, the Risk Management determinants model for vendor selection in IT Tender Evaluation process is proposed and is shown in Figure 2. Table 12. Spearman’s Correlation Result between Financial, Hazard, Operation and Strategic with Risk Management Construct Financial Hazard Operation Strategic
Correlation 0.731 0.678 0.667 0.662
Sig. Value 0.000 0.000 0.000 0.000
Interpretation Positive high correlation Positive moderate correlation Positive moderate correlation Positive moderate correlation
As shown in Figure 2, financial shows a positive high correlation. It indicate that there is a high correlation between financial and risk management. One variable under the financial construct: stability shows a positive high correlation. It proved that vendor’s stability is highly correlated with vendor financial. Pricing also shows positive moderate relationship which indicate that there are medium association between pricing and financial. The analysis also revealed that there is positive medium relationship between hazard and risk management. From the three construct in hazard, security has the highest correlation which represents positive high correlation. It proves that security is highly associated with hazard. This results support the finding of other researchers that rose up that security is number one issues in IT project. The other two variables (agreement and support) show a positive medium relationship with hazard construct. Operations construct and its variable (performance and deliverable) shows a positive medium relationship with risk management. Strategic construct shows a positive medium correlation with risk management. Three variables under the strategic construct namely reporting, business enabler and supply chain show a positive medium relationship. Meanwhile, customer demand shows a negative medium relationship which indicate inverse relationship between customer demand and strategic.
14
S.R.S. Aris, H. Haron, and N.A. Murat Customer Demand Reporting
Business Enabler -0.441
0.561
Stability
Supply Chain
0.433
STRATEGIC
0.468
Performanc
0.662
0.873
0.649 0.731
FINANCIAL
0.667
RISK MANAGEMENT
OPERATION
0.626
0.649 0.678
Pricing 0.503
Agreement
HAZARD 0.721
Deliverable 0.658
Support
Security
Fig. 2. Risk Management determinants model for vendor selection in IT Tender Evaluation process
6 Conclusion Vendor selection has been identified as one of the most important factors that contribute to project success. One of the processes in vendor evaluation is tender evaluation. This research is conducted to find the determinant used in tender evaluation process. From the research, it can be concluded that the following determinants are associated with tender evaluation process: stability, pricing (financial); security, support, agreement (hazard); deliverable, performance (operation); reporting, supply chain, business enabler and customer demand (strategic). The result also revealed that financial construct and security variable have positive high relationship with risk management. The association confirms on the determinant and thus a framework for the determinant is proposed. For future work, it is recommend that more organizations will involved since only small sample involved in this research. Since small sample since might affect the result of the association, it is worth to see whether bigger sample would yield similar result.
References 1. McDougal, P.: Outsourcing’s on in a Big Way. Communication Convergence 17 (2003) 2. Kearney, A.T.: Geography of Offshoring is Shifting, According to A.T. Kearney Study. USA, (2009) 3. Aris, S.R.S., Arshad, N.H., Mohamed, A.: Conceptual Framework on Risk Management in IT Outsourcing Projects. Journal of WSEAS Transactions on Information Science & Applications 5, 5 (2008) 4. Handfield: A brief history of outsourcing. Retrieved from College of Management at North Carolina State University portal (2006), http://scm.ncsu.edu/public/facts/facs060531.html 5. Department of Treasury and Finance.: Manual on small earth dams (2006), Retrieved from http://www.fao.org/docrep/012/i1531e/i1531e04.pdf 6. Yusof, S.: Pembangunan sistem pangkalan data bagi proses penilaian tender dalam industri pembinaan (Doctoral dissertation) (2006), http://eprints.utm.my/4603/1/SurayaYusofKPFKA2006TTT.pdf
Risk Management Determinants in Vendor Selection of IT Tender Evaluation Process
15
7. The Central Vigilance Commission.: Tender stage (2010), Retrieved from http://www.cvc.nic.in/3%20Tender%20Stage.pdf 8. NSW Department of Commerce Tendering Manual. Risk Management in a Tender Process. PWM-0633 Issue: Edition 2 Version 4.00 - Tendering Manual (October 2004), retrieved from http://www.nswprocurement.com.au/psc/Tendering-Manual/ Appendix-3A.aspx (2006) 9. Jae, Huynh, Ron Chi, Shih: IT outsourcing evolution: past, present, and future. Wireless networking security 46(5), 84–89 (2003) 10. Dhar, S., Balakrishnan, B.: Risks, Benefits and Challenges in Global IT Outsourcing Perspectives and Practices. Journal of Global Information Management 14(3), 39–69 (2006) 11. ITWorld, BCB Pioneers Bank IT Outsourcing in Malaysia (2002), http://www.ITWorld.com (retrieved January 17, 2006) 12. Douh: The development of performance criteria of tender evaluation process for public waorks in chad (2009), http://dspace.knust.edu.gh/dspace/bitstream/ 123456789/1084/1/THE%20DEVELOPMENT%201.pdf 13. Carmarthenshire County Council in Wales. Mitigating Risk in Procurement (2004), retrieved from http://www.tenderwise.com/docs/ MitigatingProcurementRisk091205.doc 14. McFarlan, F.W., Nolan, R.L.: How to Manage an IT Outsourcing Alliance. Sloan Management Review 36(2), 9–23 (1995) 15. Encyclopaedia of Business. Vendor rating (2010), retrieved from http://www.referenceforbusiness.com/management/ Tr-Z/Vendor-Rating.html 16. European Commission. General terms and conditions for information technologies contracts (2006), retrieved from http://ec.europa.eu/taxation_customs/resources/documents/ common/tenders_grants/tenders/it_general_conditions.pdf 17. Wu, Olson: Supply Chain Risk, Simulation, and Vendor Selection. International Journal of Production Economics 114, 2 (2008) 18. Technology Business Research, Inc. Benchmarking Security Vendor Infrastructure and Stability, retrieved from http://eval.symantec.com/mktginfo/enterprise/ white_papers/b-tbr_security_vendor_benchmark_WP.en-us.pdf 19. Qualstar. Technical support & customer service procedures, (2010), retrieved from http://www.qualstar.com/Tech_Support_Procedures_Web.pdf 20. OECD Publishing. Integrity in Public Procurement. Good practice from A o Z (2007), retrieved from http://www.oecd.org/dataoecd/43/36/38588964.pdf 21. Prakash: IT as a business enabler (2009), http://www.expresscomputeronline.com/ 20090525/1000thissue18.shtml 22. AIRMIC, ALARM & IRM. A risk management standard (2002), retrieved from http://www.theirm.org/publications/documents/ Risk_Management_Standard_030820.pdf 23. Sourcing4oem.com. RFI-Request for information template and description (2010), retrieved from http://www.sourcing4oem.com/purchasing-related-articles/ purchasing-articles/rfi-description-and-template 24. Method123 TM. Project Management Glossary (2003), http://www.method123.com/project-management-glossary.php 25. Marchewka, J.T.: Information Technology Project Management. Wiley (2010) 26. Aris, S.R.S.: Risk Management in IT Outsourcing Framework in Malaysia, Doctoral dissertation, Universiti Teknologi MARA (2010) 27. Hair, J.F., Anderson, R.E., Tatham, R.L., Black, W.C.: Multivariate Data Analysis. Prentice-Hall (1998)
A Survey on Data Integration in Bioinformatics Cheo Thiam Yui, Lim Jun Liang, Wong Jik Soon, and Wahidah Husain School of Computer Sciences, Universiti Sains Malaysia, 11800 Minden, Penang, Malaysia
[email protected],{s_ven86,soon_legion}@hotmail.com,
[email protected]
Abstract. The need for data integration is widely acknowledged in bioinformatics. There are several huge biological databanks now available across the world in different formats. To characterize or apply data mapping between several data sources requires integration of all related data fields. The problem of integration may be addressed using a variety of approaches; some are widely used and some are less so, having failed to achieve the basic requirements of data integration. In this paper, we discuss three techniques for data integration: th e federated database system approach, th e data warehousing approach and th e lin k-driven approach. While each approach has its strengths and weaknesses, it is important to identify which approach is best suited to a given user’s needs. We also discuss some database systems which use these three different approaches to solving the problem of data integration. Keywords: Data warehouse, Link-driven approach, Federated databases, Data integration, Bioinformatics.
1 Introduction The emergence of a large number of bioinformatics data sources (DSs) spread across the Internet in a wide variety of formats and different locations has resulted in a need for flexible and efficient approaches to integrating information from multiple bioinformatics DSs and services. Research into bioinformatics data integration is usually related to applied mathematics, informatics, statistics, computer science, artificial intelligence, chemistry, biochemistry, and biomedical information management and analysis [1]. Besides that, research has also examined methods for importing or extracting many types of data, including structured, semi-structured and unstructured. DS data are often stored in web-accessible biological databases like SwissProt, Genbank, EMBL, DDBJ, or Enzyme [2]. Thus, analytical methods are needed to carry out bioinformatics investigative tasks such as identifying and characterizing regions of functional interest in a genomic sequence or retrieving correct or nearly correct biological data from multiple DSs [3]. With the expansion of biological DSs available across the Web, integration of all of this remote data is a major challenge for system designers wishing to develop an ideal A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 16–28, 2011. © Springer-Verlag Berlin Heidelberg 2011
A Survey on Data Integration in Bioinformatics
17
system which can meet all users' needs. Alternatively, some proposed systems target individual web databases directly in an effort to preserve the system’s autonomy. Examples of these systems include TAMBIS and BASIIS [4]. These systems use a wide range of underlying architectures. For instance, TAMBIS is a federated database built on a mediator-wrapper architecture that integrates various DSs. Every integration tool or system, whether commercial or academic, has its own strengths and drawbacks. Choosing the right tool for the right task is critical to the success of any data integration project. The problem of integration may be addressed via data warehousing, link-driven systems, federated databases and so forth [1]. Since each approach has strengths and weaknesses, it can be difficult to identify which approach best suits a particular need without perfectly understanding the data integration landscape. This paper gives an introduction to the problem of data integration in bioinformatics. First, we provide an overview of the landscape of biological data integration (section 2). Then, we discuss the existing tough problems in bioinformatics integration such as data standardization (section 3). In section 4 we present various data integration approaches (namely, data warehousing, link-driven systems and federated approach) along with a means of categorizing and understanding individual approaches being used by researchers. In section 5 we discuss the strengths and weaknesses of each approach. Finally, we show examples of each approach in section 6 and propose the solution for data integration in bioinformatics using the hybrid approach in section 7.
2 Overview of Data Integration Biological databases like SwissProt, Genbank, EMBL, DDBJ, and Enzyme store data in different formats, reside in different locations and have different user interfaces [5]. To improve productivity in biopharmaceutical research and development (R&D), integration is necessary to combine these databases so that they look like a single database. Several examples exist of integrated multiple databases, such as Kleisli [3], TAMBIS [4], SRS [6], and DiscoveryLink [7]. Biological databases often contain large and complex data structures which reflect the richness of their research results; they also cover similar domains, such as genes, proteins, sequence annotations or microarray results. Several data models -- flat files, relational and object-oriented relational structures, extensible markup language (XML) and others -- have been introduced to store biological data in the database. Users wishing to exchange or merge information between DSs must understand the database schemas used in each source system in order to use data from different sources effectively and translate accurately among the schemas. There is thus a need for tools and interfaces that are powerful enough to convert bioinformatics data from one database schema to another while being flexible enough to deal with the redundant and conflicting data frequently found among different data sources. These systems should be able to search for similar data across many databases, despite the differences in database structure or field names.
18
C.T. Yui et al.
3 Problems in Bioinformatics Integration The major problem in bioinformatics data integrity is the enormous number of bioinformatics DSs; the number keeps increasing from year to year, which increases the difficulty of identifying an effective solution in query processing. Because biological DSs are spread across the Web and are different in semantics, interfaces and data formats [1], it is almost impossible to develop a perfect universallyaccepted standard of representation for central biological concepts (e.g., gene, protein, sequence, polymorphism, and pathway) or for querying protein and DNA interaction networks. Moreover, new varieties of DSs are being added daily, raising the difficulty level for gathering and organizing DSs. On the other hand, scientists need both database query activities and proper usage of computational analysis tools to effectively merge and analyse biological data derived from multiple, diverse, heterogeneous sources freely and readily. For example, the programming languages bioperl, biojava, and biopython can employ XML to deal with large data sets of up to about one terabyte [5]. However, the problems of modeling, storing and querying have still not been solved satisfactorily. Research like the present study is extremely necessary to identify all the problems that currently exist in bioinformatics and figure out a best solution, because data integration is widely needed in the bioinformatics industry. Since the data needed in bioinformatics is spread across the Internet in different formats, data integration is obviously crucial to any integrated view of all relevant data. The presence or absence of this integrated view plays an important role in determining bioinformatics-related activities' success or failure. Thus, we were inspired to conduct this research to figure out the best solution to data integration in bioinformatics.
4 Approaches to Integration Many studies have been done in bioinformatics-related fields and various approaches for integrating all the diverse biological data sources have been explored. In this paper, we examine three integration approaches being adopted to address the issue of interoperability among biological databases, namely: navigational or link-driven integration, mediator-based or federated databases, and data warehousing. 4.1 Data Warehouse Approach Data warehousing is the process of generating an architected information-management solution that enables informational and analytical processing across platform, application, organizational and other barriers [1]. The architecture of a data warehouse is shown in Fig. 1. In this approach, data is migrated from multiple data sources into a single DBMS which acts as a central repository, typically a relational database [2]. The imported data may be filtered and/or cleaned and its structure may be transformed in order to retrieve the desired data more precisely. Since the warehouse is a copy, it must be refreshed frequently to ensure it contains the latest data [9]. A data warehouse may involve deployment of "data marts" [10], subset warehouses designed to support a specific activity or inquiry. For instance, Initiative Management Data
A Survey on Data Integration in Bioinformatics
19
Fig. 1. Architecture of a Data Warehouse [8]
Mart is used to integrate and optimize data for generating project management reports whereas Quality Data Mart is used to manage and solve the complex problem of associating lab workflow information with the quality of the sequencing process. 4.2 Federated Database or Mediator Approach A federated database is a logical association which integrates multiple autonomous databases together into a single logical large database. A federated database system also called a remote database is a collection of several data sources whose formats may include flat files, spreadsheets, and text documents [11]. The architecture of a federated database consists of five levels of schema: local schema, component schema, export schema, federated schema, and external schema, as shown in Fig. 2. • A Local Schema is a conceptual schema expressed in the primary data model of each independent component DBMS. • A Component Schema translates the several local schemas into a common data model (CDM) or canonical data model [12] by mapping like fields. A transforming processor uses this mapping to transform the component schema back to their local schema on demand. • An Export Schema uses specified access controls to limit the operations that the federated schema is allowed to submit to its component schemas. The process used by the export schema is also called a filtering process [12]. • The Federated Schema is a dynamic view of the logically integrated component schemas. At this level, query decomposition is performed as is the merging of data between one or more export schemas. • An External Schema is a secondary level of abstraction created for a specific class of user or application [13]. A federated database can be very large and complex; an external schema can be customized from the larger federated schema to meet different users' needs. For example, an external schema could be used to map data (or subsets of data) from the federated schema to a visual representation; visualization makes large or abstract data easier to read or understand.
20
C.T. Yui et al.
Fig. 2. Five-level schema architecture of a federated database system [12]
A federated database system (FDBS) can be characterized as loosely coupled or tightly coupled. A loosely coupled system contains multiple federated schemas because each group of users needs to create and maintain their respective schema. In a tightly coupled FDBS, the administrator has full control over creating and maintaining the component database systems as well as the export schema. A tightly coupled FDBS may have single or multiple federation schemas depending on the constraints of the integration. Different architectures address different problems. A loosely coupled FDBS is typically used when potential constraints [12] of federation cannot be identified at the early stage, so changes to the component database may occur frequently. On the other hand, a loosely coupled FDBS enables federation users to make dynamic changes in their schemas. A tightly coupled FDBS aims to provide location, replication and distribution transparency [12]. A tightly coupled FDBS with a single federation schema integrates multiple export schemas into one standardized federated schema. Federation users are thus able to use the same standardized queries to access different component databases. When there are many export schemas that need to be controlled, a tightly coupled FDBS with multiple federation schemas is recommended. The primary disadvantage of a loosely coupled FDBS is the risk of redundant data, since users are free to define their own local schemas. Also, an FDBS is often sophisticated and it is difficult for users to understand the available export schemas, which may be defined in different semantic [12] and query languages. In a tightly coupled FDBS, any changes in the component schema or export schema require updates to the federated schema. This implies that the administrator will need to do redundant work for any changes in this type of FDBS.
A Survey on Data Integration in Bioinformatics
21
4.3 Link-Driven Approach A link-driven approach is also known as a navigational or browsing approach. This approach emerged from the fact that an increasing number of websites require users to manually browse through several web pages and data sources in order to obtain the desired information [14]. A link-driven system, such as NCBI Entrez [1] or SRS [6], provides static links between data or records in different data sources. These links constitute a kind of workflow in which the output of one source or tool is redirected to become the input of the next source, until the requested information is located [1] as shown in Fig. 3. Sources are defined as sets of pages with their interconnections and specific entrypoints, as well as additional information such as content, path constraints, and optional or mandatory input parameters [3]. Marc Friedman, Alon Levy, and Todd Millstein [15] deepened the research on path-based approaches by analyzing paths between biological sources. They showed how to extend a data integration system to incorporate data webs, and defined a data webs mode and a language that made querying multiple web-structured sources possible. Their goal was to enhance the query-answering algorithm and complexity results to identify the best of several potential execution paths to answer a given query.
Fig. 3. Schematic diagram of the browsing approach [1]
5 Existing Bioinformatics Systems In this section we review the details of existing well-known bioinformatics systems. This section also categorizes currently available systems in terms of the integration approach used. The following three systems are discussed:
22
C.T. Yui et al. • • •
AllGenes, a data warehousing system GenomEUtwin, a federated database system NCBI Entrez, a web-based link-driven system.
5.1 ALLGenes AllGenes[1] is a data warehousing system designed to provide access to a database integrating every known and predicted human and mouse gene; it is generated by clustering and assembling publicly available EST (Expressed Sequence Tags) and messenger RNA (mRNA) sequences into DoTS (Database of Transcribed Sequences) [16], a human and mouse transcript and gene index. AllGenes is a webbased system that integrates various types of data such as EST sequences, genomic sequence and expression data. Integration is implemented in a structured manner using a relational database called GUS (Genomics Unified Schema), an extensive genomics warehouse that performs significant cleansing and transformation before the remote data is loaded into AllGenes for analysis, refinement, clustering and assembly. Vocabulary and ontology control are implemented while the remote data is being integrated. The AllGenes interface uses GUS's boolean query and query history facilities, which allow users to compose sophisticated queries built from more basic queries [1]. A sample query might find all DNA repair genes that are expressed in the central nervous system; this is shown in Fig. 4 and the results of the query are shown in Fig. 5. The user can access a summary page for each qualifying assembly from the query result, which includes all the valuable information as predicted. Besides that, the user can activate hyperlinks for direct access to GeneCards, the Mouse Genome Database (MGD), GenBank, ProDom, and so on. The system also allows a user to search the top 10 best hits against the GenBank non-redundant protein database (nr) and the top 10 best protein domain/motif hits [1].
Fig. 4. The AllGenes query builder [1]
A Survey on Data Integration in Bioinformatics
23
Fig. 5. Results of the AllGenes query [1]
5.2 GenomEUtwin Genometwin is a data resource for researchers, genetic epidemiologists, molecular geneticists and mathematicians. The major challenges for Genometwin are the storage and integration of extensive amounts of post-genome data from different sources in different countries [13]. Genometwin achieves this using the TwinNET federated database system; the general topology of TwinNet is shown in Fig. 6.
Fig. 6. General topology of TwinNET [13]
24
C.T. Yui et al.
5.3 NCBI Entrez Entrez is a link-driven system used at NCBI for the databases PubMed, Nucleotide and Protein Sequences, Protein Structures, Complete Genomes, Taxonomy, and others [16]. Entrez's portal page enables users to find information on any biomedical topic by searching all 24 databases at once rather than each individually. Searching across all Entrez databases is done by entering a simple search term or phrase in the “Search across database” query box. The results will be displayed in the global query page with the number of hits in each database displayed adjacent to the database name as shown in Fig. 7. More complex searches can also be done using Boolean operators and combinations of one of more limits. Limits restrict a search to a defined subset of the database. Because the contents of each Entrez database vary, the limits for each database also vary. Entrez is more powerful than many other services because most of its records are linked to other records, both within a given database (such as Nucleotide) and between databases. Links within a database are called “neighbors” (e.g. Nucleotide neighbors) [16]. Links between databases (such as Protein and Nucleotide) are identified using an algorithm called BLAST, which is used to compare the search term(s) (e.g. "amino acid" or "DNA sequence") to all related fields in the database.
Fig. 7. NCBI Entrez Global Query [16]
6 Strengths and Weaknesses of the Various Approaches In the previous section we have presented multiple approaches to database integration in the bioinformatics domain. However, there is no single best approach that outperforms all others; each has its pros and cons and is best adapted to a particular set of integration needs.
A Survey on Data Integration in Bioinformatics
25
The strengths of data warehousing are many. First, it permits the remote data to be filtered and cleaned as the copy of the data is stored locally, according to a single global schema [1]. Second, if the original data source is not compatible with the warehouse-specific format, the data retrieved and integrated in the warehouse will indeed eventually have to be converted or transformed into a more amenable structure [2]. However, it is still questionable how this process could be achieved efficiently and, more specifically, how the data could be validated, transformed or modified effectively without costly and time-consuming human intervention [3]. Normally, the remote data sources exert a load only at data refresh times, and any changes to the original data sources do not directly affect the availability of data stored in the warehouse. This is actually a major weakness of the data warehousing approach, because the warehouse copy of the data must be refreshed frequently to ensure users have access to up-to-date content. Thus, this approach is not the best method for integrating enormous numbers of data sources that use high-level query languages like SQL, or for querying systems that change on an hourly or even daily basis [4]. In a federated database system, data remains stored in the component databases. Instead of copying a huge amount of data into a single data warehouse, federation users are able to directly access disparate data sources through the logical federated database. Federation users are able to see up-to-date information [1] all the time. As the number of data source increases, the query load in the federated database will also increase, which affects the response time of the system. In addition, a high level of programming is required to translate heterogeneous data sources into a single logical federated database. Besides that, adding new schemas and simultaneous update of a subset of the data in the federated database system will reduce the overall consistency of data in the system. The interactive nature of the link-driven approach makes it well suited to exploring the data landscape when an investigator has not yet formulated a specific question.[1] The user has the advantage of retrieving information about a single object and optionally drilling down for more details or following hyperlinks to related objects. Users without query language knowledge are able to use the system without much trouble due to the user-friendly interface which only involves the simple tasks of point-and-click, similar to web browsing [3]. The weakness of this approach is that the functionality is extremely limited. The interface is also inefficient and does not support bulk queries. [14] It is not suitable for large data sets, since the user will have to go through tedious and very cumbersome work to search from page to page for the desired information, while a short SQL query could much more easily perform the task, with similar results.
7 Proposed Hybrid Approach We propose a hybrid approach to building integration systems as the best strategy to overcome the problem of data integration in bioinformatics. The ideal integration system should take into consideration the wishes of those who will be using the
26
C.T. Yui et al.
system so that it will be able to satisfy the variety of integration needs present in a typical organization. As mentioned earlier, each integration approach has its strengths and weaknesses. Data warehousing is probably the best approach when it is critical to clean, transform, or hand curate data, and for which only the best query performance is adequate [3]. However, the data warehouse approach does have its cons: results reliability and overall system maintenance are questionable as there is the possibility of returning outdated results. If the warehouse must fetch data from outside the organization, it is best if the remote DSs do not change frequently, so that the work and costs of maintenance in merging updates is not too burdensome. In cases where the remote DSs do change frequently, the federated database is the best choice due to its relatively low maintenance costs and high extensibility, reliability and scalability [1]. An FDBS allows easy prototyping and replacing of new DSs for old in evaluation mode. It meshes well with the landscape of many remote autonomous DSs which the bioinformatics community currently boasts. Thus, all the data in an FDBS are always up-to-date, and they permit integration of external data that is not accessible for duplicating internally, such as data only available via a webbased system or a website. Secondary sequence comparison, clustering, test mining and so on are also allowed via the integration of special purpose search algorithms. What kind of bioinformatics integration system do biologists and researchers actually desire? It is clear that they want a system which can automate the maximum number of tasks in the least amount of time and with minimal user interaction. Because of the diverse and dynamic nature of biological data, manual analysis of a data source schema will cost a great deal and be very time consuming. Thus, we propose a hybrid approach which can automate processes and thereby reduce the cost and time necessary to develop full-scale integration systems that can match the pace at which biological data is generated. The ideal bioinformatics integration system must be able to automatically extract source descriptions and gather source statistics in order to analyze or refine query plans and enhance the overall performance and functionality of the system even as the source data evolve. Examples of the statistics that such a system should take into account are average response time, overlap between sources, freshness of data and density of each result to determine the accuracy of data. Wherever possible, strategies should be generic, except for one-time, one-use programs or where hard-coding is needed to fine-tune a limited set of operations over a limited set of data. Both browsing and querying are important for different levels of users and different needs. For access to data in batch mode, the most common queries can be pre-written and parameterized and offered to users via a form. Both semantic and syntactic data integration are needed, although semantic integration is just beginning to be explored and understood. Due to the evolution and maturity of the technology and the strength of its tools, both warehousing and federation use a relational data model as the method of choice for large integration efforts. As mentioned above, a middle software layer may be provided to expose biological objects to users. At present, given the state of the industry, relational databases are arguably the best choice for underlying data storage, query planning, and optimization. Fig. 8 shows the architecture of the proposed hybrid approach of data integration for public bioinformatics databases.
A Survey on Data Integration in Bioinformatics
27
Fig. 8. Proposed hybrid approach
8 Conclusion This paper began with an overview of data integration in the field of bioinformatics. Then it described the problems encountered in integrating data from diverse bioinformatics data sources. Various approaches exist to perform data integration in bioinformatics. This paper explored three approaches, namely: the data warehouse, the federated database and the link-driven system. Existing bioinformatics systems (e.g., AllGenes, GenomEUtwin, Entrez) that employ these approaches were given, and for each approach the strengths and weaknesses were discussed. In the final section, a proposed hybrid approaches for data integration in bioinformatics was discussed and its ideal characteristics given. The proposed system should be able to automate a maximum number of tasks in the least amount of time and with minimal human intervention, and to integrate data at the same pace as the bioinformatics field is advancing. The system should also take user types into consideration and provide an interface tailored for their needs -- for example a basic, intuitive one for inexperienced users and a sophisticated one with more functionality for advanced users. Acknowledgements. The authors would like to thank the Universiti Sains Malaysia (USM), Penang for supporting this study.
28
C.T. Yui et al.
References [1] Lacroix, Z., Critchlow, T.: Bioinformatics Managing Scientific Data. Morgan Kaufman Publishers (2003) [2] Eckman, B.A., Lacroix, Z., Raschid, L.: Optimized Seamless Integration of Biomolecular Data. In: Bioinformatics and Bioengineering Conference, Proceedings of the IEEE 2nd International Symposium, pp. 23–32 (2001) [3] Hernandez, T., Kambhampati, Z.: Integration of Biological Sources: Current Systems and Challenges Ahead. SIGMOD Record 33(3) (2004) [4] Stevens, R., Paton, N.W., Baker, P., Ng, G., Goble, C.A., Bechhofer, S., Brass, A.: TAMBIS Online: A Bioinformatics Source Integration Tool. In: Eleventh International Conference on Scientific and Statistical Database Management,1999, p. 280 (1999) [5] Yan, L., Vincent, S., Murphy, M.C.: Integrating Bioinformatics Data Sources over the SFSU ER Design Tools XML Databus. ACM International Conference Proceeding Series, vol. 155(19) (2006) [6] Wong, L.S.: Technologies for Integrating Biological Data. Laboratories for Information Technology 3, 389–404 (2002) [7] Davidson, S.B., Crabtree, J., Brunk, B., Schug, J., Tannen, V., Overton, C., Toeckert, C.: K2/Kleisli and GUS: Experiments in Integrated Access to Genomic Data Sources. IBM System Joural. Deep Computing for the Life Sciences 40(31), 512–531 (2001) [8] Kirsten, T., Lange, J., Rahm, E.: An Integrated Platform for Analyzing MolecularBiological Data Within Clinical Studies. In: Grust, T., Höpfner, H., Illarramendi, A., Jablonski, S., Fischer, F., Müller, S., Patranjan, P.-L., Sattler, K.-U., Spiliopoulou, M., Wijsen, J. (eds.) EDBT 2006. LNCS, vol. 4254, pp. 399–410. Springer, Heidelberg (2006) [9] Thuraisingham, B., Iyer, S.: Extended RBAC–Based Design and Implementation for a Secure Data Warehouse. In: The Second International Conference on Availability, Reliability, and Security (ARES), pp. 821–828 (2007) [10] Robert, M.R.: Bringing the Data Mart into the Curriculum. In: ACM-SE 38: Proceedings of the 38th Annual on Southeast Regional Conference, pp. 129–134 (2000) [11] Richard, M.C.: How Federated Databases Benefit Bioinformatics Research, http://www.b-eye-network.com/view/2164 [12] Amit, S.P., James, A.L.: Federated Database Systems for Managing Distributed, Heterogeneous, and Autonomous Databases. ACM Computing Surveys (CSUR) 22(3), 7–23 (1990) [13] Muilu, J., Peltonen, L., Litton, J.-E.: The Federated Database – A Basis for BiobankBasedpost-Genome Studies, Integrating Phenome and Genome Data from 600 000 Twin Pairs in Europe. European Journal of Human Genetics, 1–6 (2007) [14] Davidson, S., Overton, C., Buneman, P.: Challenges in Integrating Biological Data Source. Journal of Computational Biology 2(4), 557–572 (1995) [15] Friedman, M., Levy, A., Millstein, T.: Navigational Plans For Data Integration. In: Proceedings of the National Conference on Artificial Intelligence (AAAI), pp. 67–73 (1999) [16] The Computational Biology and Informatics Laboratory. AllGenes: A Website Providing Access to an Integrated Database of Known and Predicted Human and Mouse Genes. Center for Bioinformatics, University of Pennsylvania (2004), http://www.allgenes.org [17] Information U.S. National Library of Medicine National Center for Biotechnology Information, http://www.ncbi.nlm.nih.gov
Structural Approach to the Formation of Information Culture of Individuals Rasim Alguliev and Rasmiya Mahmudova Institute of Information Technology, Azerbaijan National Academy of Sciences, Baku, Azerbaijan
[email protected],
[email protected]
Abstract. The article explores the issues of formation of information culture of the individuals. Various views, approaches to the concept of information culture are studied and a new structural approach is proposed in the article. Knowledge and capabilities related to structural elements of individual information culture – reception, storage, protection, as well as processing and presentation of information are explored. Necessity of comprehensive development of these knowledge and capabilities for formation of information culture in people is justified. Keywords: information culture, information reception, information storage, information processing, information protection, information presentation.
1 Introduction It is well known that a human being is constantly in an interaction with the information environment (infosphere) around him. A large part of the human activity is related to information: reading, watching TV, listening to music, studying, communicating, discussing something, preparing any document, etc. Despite that people are engaged in processing and exchanging information for centuries since their creation, nowadays information activities, communication skills and abilities necessary for properly dealing with the information are of great importance. This is associated with the recent emergence and development of an information society. As information is considered as the key strategic resource in the information society, irrespective of the sphere of his/her activity each individual is required to deal with the information on a high level, to obtain the necessary information on time and apply it in his/her activities. Therefore, besides a number of factors, successful development of an information society is characterized by the level of information culture of the people and society on the whole. As a result of the rapid increase in the volume of information, people are loaded with too much information, and as well, this has a negative impact on their normal activities. In other words, human beings are obliged to receive, process, remember new information more than ever and apply it in their activities, which leads to the information weariness characteristic to modern period. Researches show that information weariness reduces A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 29–40, 2011. © Springer-Verlag Berlin Heidelberg 2011
30
R. Alguliev and R. Mahmudova
thinking ability, negatively affects normal activity, and causes medical and psychological complications [1]. Therefore emerges the necessity to prepare the individuals for life conditions in this society, and teach them to rationally treat the inforamtion.
2 The Structural Approach to Information Culture Conducted researches demonstrate that the definition of information culture involves various approaches. This is connected with that information culture is an integrative type of culture i.e. it is a new form of culture uniting other culture forms and has a very broad coverage area. On the other hand, experts of various fields of science have conducted numbers of researches on these problems and different approaches have emerged based on individual aspects. Let’s review some of them: “information culture – is ability of using information approach, analyzing information conditions and increasing the efficiency of information systems [2]. “information culture – is a knowledge level allowing a human being to freely behave in the information environment, to participate in its formation and to implement interaction with information” [3]. “information culture – is quality indicator of human activity in the field of information acquisition, transfer, storage and application” [4]. “information culture – is harmonization of the inner world of an individual in the process of information acquisition of social importance” [5]. “information culture – being one of the integral parts of human culture in general, is a compilation of skills and knowledge, as well as information outlook providing independent, objective action to meet individual information demand in an optimal way using both - traditional and new information technologies” [6]. Various scientific explanations of this definition have caused the emergence of different approaches to formation of an individual's information culture. In general, the content of "information culture" definition represents the interaction of a human being with the information environment. As it is known, interacting with the surrounding infosphere, a human being receives information on the one hand, stores the information, processes it (information consumer), on the other hand synthesizes new information and knowledge, creating and presenting it to the infosphere (information producer). In this respect, information culture – can be defined as a compliance of knowledge and skills of an individual related to receiving information, information storage, processing, preserving and presenting it, as well as conforming to legal and ethical norms while using information. Therefore, formation of an individual information culture can be implemented by exploring and eliminating the problems emerging during interaction of a human being with information. A group of people has a very high capacity for reception of information, its acquisition, storage, while the others have a skill of keeping in mind. But there is another group of people possessing a high level of analytical thinking, sharp intellect, and produce new and more capacious ideas based on obtained information. Some people possess a highly developed ability to present information. They are able to deliver the obtained information to the society or any audience in a laconic way, and take control over the audience easily with their distinct and fluent speech.
Structural Approach to the Formation of Information Culture of Individuals
31
Development of one or more of these skills can be observed in people, certain qualities can be found in each. However, all the qualities can hardly ever be concentrated in one and the same person in a composite form. Depending on their profession features, these or other skills and habits are developed more. This allows to state, that these skills can be developed in a complex form through special assignments, and trainings. As in terms of level these skills differ in people, general level of information culture differs, as well. The lack in the level of information culture negatively affects the individual’s success which can be obtained during education and professional activities. Therefore, considering that in accordance with the structure of general information culture, the human being depends directly on: information reception, memory storage, processing, protection, if necessary, and presentation skills and abilities, formation of general information culture of an individual can be achieved through investigation and development of these skills separately. Information reception
Initial storage of information
Information processing
Information protection
Information presentation
Fig 1. The structure of information culture of individuals
3 Information Reception Culture of Individuals Capability of accurate and adequate reception of information can be considered as the initial stage in the mutual relations between the Infosphere and an individual. Development of the person and his/her identification as a personality initiates with reception of substantial information about the environment through the sense organs.
32
R. Alguliev and R. Mahmudova
After the process of information processing based on initial knowledge, consciousness and outlook are formed. It should be noted that information reception depends on several factors: methods for information reception, its form and content, objective set forward, interest, a person's ability to observe, concentration ability, health and etc. It is known that capabilities of information reception channels (sense organs) of each person are different. People can be divided into several subgroups based on their capabilities of information reception - visuals, audiles, tactile learners and digitals. Visuals accept information by vision, audiles by hearing organs. Tactile learners are affected by other types of sensation - smell, touch, movement, etc. Digitals achieve logical understanding based on signs, numbers and logical results. Therefore, the levels of information reception by the people depends on the form (text, sound, images, etc.) the information is presented and how the person corresponds to the key channel of reception. The process of information reception requires a number of particular skills and abilities, one of which is observation. Observation is an ability to feel very important, but less noticeable details, features, attributes etc in events and subjects. It is a mean of obtaining necessary information and is a forethoughtfully planning and independent receiver [7]. Observation relies on life experience and knowledge, as well as willing of the person to learn and know. At the same time, as the other psychological characteristics of the person, observation ability is not an innate talent either, it can be developed by means of special exercises and trainings. High-level of observation ability is of great importance for comprehensive development of an individual. Comprehension of received information through different senses depends on concentration of attention as well as willing and other similar factors. Attention of the human being is structured in such way that any sound, bright vision, etc. draws attention faster. As a result, it is impossible to direct one’s attention to necessary information, rather an individual is distracted from the objective which results in incomplete or imprecise information reception. For full and proper reception of the information, it is very necessary to learn to focus on important information and receive only essential messages. This should be studied in particular. According to experts, reception and perception of information relies on the human’s interest circle, as well as the goal pursued. For example, if a person wants to engage in some activity, he/she tries to accumulate and acquire all information about that activity. Naturally, in this case more information can be accumulated. Information reception is partially connected with the health of an individual. Thus, being sick or too tired, or being in a bad mood are of factors negatively affecting information reception and comprehension. Although information reception depends on a number of different factors, the key factor is whether the information is appropriate to a person’s reception channel capabilities. Presently applied traditional lesson-class system doesn’t consider the diversity of information reception channels of the students. Therefore, the students whose information reception method matches the teacher's information presentation method acquire the information better, while the others experience difficulties.
Structural Approach to the Formation of Information Culture of Individuals
33
4 Initial Information Storage Culture in Individuals One of the most important components providing an individual with high-level behavior toward information is his/her memory. A human being is developing, improving and contributing to the development of the society through remembering knowledge and skills obtained from years of education, personal experience, exploring the experience of others, and through applying them in his/her activity. An individual is linked to the society through his/her memory. The role and importance of memory to succeed either in educational or professional activity is great. At present, the rapid increase in the volume of information cause overloading of people’s memory. People are helpless before the huge volume of information flow, reception and storage of such amount of information is difficult due to limited capabilities of memory. Experts call this problem emerging between information growth rate and people's ability to adopt the information as “information crisis” [8]. In order to facilitate the operation of the memory, it is necessary to master special ways and methods of information storage, and achieve development of the memory through its training. It is known that everyone possesses a different level of memory. As people differ from each other with distinct characteristics, they differ by nature and types of memory, as well. Variety in development level of the memory types is reflected in the individual's behavior with information and affects the results achieved in life on a certain level. Thus, studying the characteristics of people’s memory and development of these characteristics from young ages must be considered as an important stage of formation of information culture. The study of the characteristics of the memory starts with acquaintance with memory types, focusing on own memory and assimilation of tasks for the development of the memory. It has been proved that the information storage has a selective nature, information most vital for a person is remembered for a long time, sometimes for a life-time. Everything related to human behavior, needs, interests, desires, and purposes are easily stored in the human conscience and easily recalled. Information unapplied in life and professional activity is forgotten. Therefore, not to forget the information, its constant repetition and application in one’s activity is recommended. However, only repetition is not sufficient for better memorizing, the information should be classified in groups and coordinated with each other. To memorize the information easily it should be properly structured. It means that, coordination of any newly submitted information with existing information base, and its conscious interpretation is essential. There are several methodologies existing in modern psychology aimed at improving storage capabilities of information of text, sound, images and other types. For example, method of algorithmization of the verbal text used to remember texttype information more effectively is well known. This method is applied to facilitate recording the scientific, massive-scientific texts, verbal reports and speeches. The main point of the method is that, primarily the important information is separated from the general content, in the next stage key ideas are segregated from the selected information, and in their turn, these ideas are suppressed to the image-words. As a result, these words play the role of the keys which assist the speaker in presentation of a systematic report.
34
R. Alguliev and R. Mahmudova
In order to improve the function of memory, an individual must know the main methods of storage the information and be capable of applying them. According to specialists there are numbers of memorizing rules, which can be used to improve the memory [9]. There are a number of factors that negatively affect the functioning of memory: malnutrition, nonobservance of sleep pattern, exhaustion, harmful habits such as smoking, drinking and etc. Therefore, everyone should be aware of the memory hygiene and it must be followed in order to preserve memory. Otherwise, it can lead to memory weakness and some anomalies. Thus, exhaustion, information overload, a strong emotional stress can result in a number of memory diseases. Thus, everyone should be informed of certain issues related to memory, its types and characteristics, methods of information storage, as well as development of memory, the memory hygiene.
5 Information Processing Culture of Individuals One of the most important components, characterizing information culture of than individual are the skills and habits related to the information processing, which play a very important role in human-information behavior. Not only information itself, but also its selection, evaluation, systemization and processing skills are necessary in the world of information. The supreme function of a human being is always to generate new information and knowledge through processing continuously received information. However, the processing capacity of all people is not on the same level. If a person isn’t capable of producing materials of different content using gained information and knowledge, or he/she can not use them in his/her activity through analyzing, it is due to the poor processing skills and mentality. Abundance and diversity of information, as well as the lack of precise information and importance of its selection and evaluation force people to engage in the activities of analytical information in their daily life. In other words, rapid changes occurring in the society require any specialist to analyze updates and changes in his subject field, to make conclusions, to forecast and be ready to make a decision. This means that people should be taught not only to use existing knowledge, as well as to investigate it to solve any specific problem, to produce new information and knowledge through processing collected ones. Nowadays it is rather important to prepare different types of products – information products using different types of information as primary raw material. Information processing process is a process of obtaining new information as a result of performance of certain operations during accumulation of primary data. Processing results in modification of form or contents of information. Processing associated with acquisition of new information and content is analytical processing. Analytical processing of information is a creative process, reflecting operations of comprehension, analysis and assessment of its content for selection of important information from the initial one. Analysis allows selecting more valuable information, allocating secondary information and data, i.e. to perform certain analytical
Structural Approach to the Formation of Information Culture of Individuals
35
operations. Along with analysis process of text, its synthesis takes place which is a full and systematical combination process of information obtained as a result of analytical operations[10]. Apparently, the process of information processing is not an easy procedure, therefore requires an individual’s analytical skills. These include information search, selection and evaluation, analysis, systematization, generalization and design skills. The process of analytical processing of information has several consecutive stages in order to meet information needs of human. The crucial term to meet information demand is to understand which particular information is necessary to solve the set forward problem, and exact expression of the demand in a form of information request [11]. After defining the information demand accurately, an individual begins to collect corresponding information. From this point of view, possessing certain knowledge and skills for the prompt search of necessary information and notion about all types of information sources are initial conditions of the processing procedure. Selection and assessment of acquired information depends on certain objectives, thus, the more information helps to achieve the goal more valuable it is. Development of components of information culture related to information processing can be achieved through inculcating the people with the skills for analysis and synthesis of information. For example, planning, writing articles, taking notes, preparation of documents such as summaries and reviews, etc. can aid the formation of these skills and habits. Using of the project method can contribute to development of skills and habits related to information processing in students. The project method is a pedagogical technology developed by American pedagogue W. Klipatrick in 1920s. In the project method the main purpose is ensuring the students to acquire knowledge independently in the process of practical tasks or solving problems requiring knowledge integration of different subject fields, and here a teacher is a coordinator and an expert [12]. The work on the project will contribute to the formation of features such as comparing information collected from various sources, defining its significance in terms of problem solution, setting up conclusion, expressing one’s opinion.
6 Information Protection Culture of Individuals Consideration of information as a key capital of the modern period, necessity of information protection has increased in connection with an increase in its value and importance. Because, valuable things are always exposed to danger of harming the owner: they are stolen, sold, destroyed, etc. Therefore, skills and habits related to the information protection can also be included into the components of information culture of an individual. In former times, people stored information mainly in their brains and on paper carriers. Nowadays, computers are used in almost all fields of human activity for collection, processing and storage of information. Today, large quantities of various types of information about the activities of enterprises, state and public organizations and individual citizens are being collected and stored in computer networks and various electronic databases. Under such
36
R. Alguliev and R. Mahmudova
conditions, the threat of using this information in order to harm the citizens and institutions emerges [13]. Therefore, people working with information systems have to be aware of the importance of protection of collected information and of what’s to be done in order to provide information security and should be responsible for it. Up to modern times, main attention was focused on the development of various technical and technological methods and means (development of the antivirus programs, internet work screens, spam-filters, variety of methods to control and limit access, registration of incidents, etc.) while solving the problem of information security. The application of information-communication technologies - ICT in all spheres, comprehension of the fundamental role of information in society, inclusion of information into the category of important strategic resources have resulted in understanding that information security depends not only on the technical and technological solutions, but rather on the culture of the society and its members [14]. It is not accidental that in 2003, UN adopted a resolution in connection with "Creation of global information security culture". The resolution identified nine items of requirements for state bodies, enterprises, other organizations and individual users generating, controlling and using information systems and networks. It indicates the level of importance of information security issues. In addition to information protection issues, problems of individual thought protection, as well as the measures taken over the information ethics, protection of human psychology can be considered a part of the information security problems. It is known that an individual’s enviroment – family, education instituion, society, as well as inforamtion envriroment play an important role in maturing of an individual as a personality, formation of one’s world view. Under the conditions of information society, ICT and Internet have great influence on formation of one’s demands, interests, views, values, and world-view in general. Along with important and useful information resources, there is a great amount of information disseminating religious extremism and nationalism ideas, and promoting harmful habits placed on the Internet. And this type of information rather negatively impacts the psychology of the people, especially children and teenagers. At present, Internet offers different kinds of services to the people, usage of which enables committing certain crimes against them. For example, by registering on social networks, people post lots of personal information there: photos of family members, including children and properties, videos of daily life (work, home, leisure, events attended), information about occupation, hobbies, future plans and etc. In this case, danger of organization and implementation of crimes against those individuals and their family members using this information emerges. Thus, knowledge and skills on providing information protection and its safety in the modern world become of great importance. Therefore, since young ages people should be taught how to behave in the information space, be informed about harmful Internet contents and their negative impact on one’s psychology, as well as methods of protection from these effects. People must know, if one doesn’t use information resources properly, others’ rights are violated and he/she is responsible for it. Everyone should be able to protect his/her own information and its use in order to harm others must not be permitted.
Structural Approach to the Formation of Information Culture of Individuals
37
7 Information Presentation Culture of Individuals One of the basic indicators characterizing the high-level information culture of an individual are the skills related to the information presentation. These skills are of great importance in the mutual relations between the human and an environment. It is known that, information presentation by an individual is mainly conducted by means of natural language in written or verbal form. Furthermore, information is also transmitted in a variety of different ways such as gestures, mimicry, dancing, music, painting, etc. People’s ability to distinctly deliver information to others is very essential in the information society. We must regretfully note that, only a certain part of graduates are capable of correctly expressing their views verbally and in written form. In this regard, the development of information presentation skills should be considered as one of the stages of information culture formation. As presentation of information is carried out by means of speech, culture of speech can be considered one of the components of information culture. Development of human speech capacity and lexicon starts since childhood and continues all lifelong. Reciting poems, retelling the lesson, acting as a certain literary character, active conversation with classmates at school stimulate the development of ability to express one’s opinion and speech culture. However, formation of speech culture is not characterized systematically. We would also like to note that, application of test method during examination and assessment of knowledge in secondary and higher education institutions, superiority of communication over internet among young people has resulted in intensification of problems in this field. In higher education system, with some exceptions, consistent and purposeful work is not carried out in the direction of formation of speech culture. However, this capacity is of great importance during all kinds of mutual relations of people is with environ. An individual presents himself/herself to the society through speech; in this way intellectuality, resplendence of his/her inner world is displayed [15]. Information must be clearly submitted to its recipients. In other words, it should include as many full sentences as possible; avoiding too many special terms and complexity of grammatical structure, expressing the essence of the idea. Features such as pithiness, sequence and efficiency should to be taken into account for a good speech. A wide range of vocabulary resources is required while making one’s speech interesting and expressing ideas adequately. However, only a range of vocabulary resources is not sufficient. Clarity of speech also relies on the ability of the speaker to adapt the speech to the audience's terminology base. From this perspective, the participants of the audience must be taken into consideration during presentation of information. Using mimicry and gestures during the speech aids the expression of information, as well as its reception by the listeners. In this context, studying the methods of speechless information expression means such as using gestures and mimicry are particularly expedient. Presentation of information in a written form is of a completely different character. During the preparation of the written text, determining its precise structure and
38
R. Alguliev and R. Mahmudova
arrangement style in advance is very importance. The highest form of written language is the ability to express one’s views consistently and coherently. To convey views completely and consecutively, one should be well informed of the language grammar and possess a large lexicon. Meanwhile, the words for the text composition must be selected in such manner that readers understand it clearly. In order to compose a clear and understandable text, one should know the described area thoroughly and investigate the topic carefully. In other words, it is required to address multiple sources, carefully read and analyze the collected information and be capable of creating a new text in order to prepare a good material regarding any topic. Correct construction of a structure of the sentence is very essential while preparing the written text. The structure of the sentence should be simple and unfatiguing for the reader. One of the important moments is the sequence of sentences and logical consecutiveness. Each new idea must be the logical continuation of the previous one and not contradict it. Repetition harms the content of the text. Therefore, it is necessary to read the text carefully several times and reduce repeated ideas. At present, there are many styles and types of written language and they must be presented from school years. Although, from this perspective writing essays in native language classes is of great importance, it is not of systematic character. Writing skills must be taught specifically.
8 Conclusion Analysis carried out in the article allows conclude that the level of information culture of an individual is related to development level of skills such as information reception, storage, processing and protection, as well as its presentation (transmittal). As these skills are not on the same level for everyone, the level of their information culture also differs. This can lead to the formation of so-called new type of information disparity among people. Hence, curriculum of comprehensive schools should include the subject of "information culture" and skills noted within this subject be developed. In order to avoid it, information culture of individuals must be formed and capabilities defining its structure must be specifically developed: • Individuals should acquire knowledge and skills related to the correct reception of information. It is necessary to broaden people’s capabilities through grouping their key reception channels from young ages, other reception channels should be developed gradually. • People must be trained regarding information storage methods, as well as memory control and development, observation of memory hygiene, its conservation and improvement skills. • Individuals’ analytical skills related to information processing must be developed. For this reason, skills related to search, selection, assessment, systematization, analysis, synthesis and generalization must be taught to people.
Structural Approach to the Formation of Information Culture of Individuals
39
• At present, transformation of information into a valuable resource has highlighted its protection and security provision. In its turn, acquirement of essential knowledge and capabilities for protection of information, as well as fulfillment of ethical and legal norms during interaction with information environment is required. • One of the features characterizing the information culture of an individual is the ability to present information. As transmittal of information is primarily carried out by means of oral and writing speech, corresponding skills and capabilities, as well as other means of transmitting information such as gesture, mimics, graphical images etc must be specifically taught. The education system is primarily responsible for formation of information culture meeting the requirements of the new society. The education system can fulfill this duty by submitting “Information culture” class to the curriculum of secondary education institutions. Within this class, harmonic development of skills such as information reception, storage, processing, protection and transmittal can be achieved from early ages of the students. As general information culture of individuals consists of compilation of abovementioned skills, they must be on a high level, so that in the future they would not experience any difficulties while mastering and applying rapidly increasing information and knowledge regardless of their activity field.
References 1. Slabunova, E.E.: Information Culture in the concept of lyceum education, http://vio.fio.ru/ (in Russian) 2. Vorobyev, G.G.: School of the Future Begins Today: teacher’s book. Molodaya Gvardiya, Moscow (1991) (in Russian) 3. Medvedyeva, E.A.: Basics of Information Culture. Federal Educational portal Ecology, Sociology, Management, http://www.ecsocman.edu.ru (in Russian) 4. Khangeldiyeva, I.G.: About the concept of information culture. Personal Information Culture: Past, Present and Future. In: Intern. scientific. conf., Krasnodar-Novorossiysk. Abstracts Rep. Krasnodar, pp. 7–8 (1993) (in Russian) 5. Zinovyeva, N.B.: Personal Information Culture: Introduction to the Course: Handbook. Manual for schools of culture and art. Krasnodar (1996) (in Russian) 6. Gendina, N.I.: Information Culture and Information literacy of an individual: international and Russian approaches to the problem. J. Open Education 5(64), 58–69 (2007) 7. Regush, L.A.: Workshop on Monitoring and Observation. Sankt-Peterburg, Piter (2001) (in Russian) 8. Vasilyeva, E.: Memory laws and efficient techniques of memorization, Centre of distance education, http://www.elitarium.ru/ (in Russian) 9. Provotorov, V.I.: Annotating and reviewing as instructional techniques in training working on specific texts. J. Scientific Notes, El. Scientific journal of Kursk State University, http://scientific-notes.ru/ (in Russian) 10. Sherbitsky, G.I.: Information and educational needs. BSU publication house, Minsk (1983) (in Russian)
40
R. Alguliev and R. Mahmudova
11. Kurbatova, L.M.: Project activities of a student as a technology of assigning information culture of the future specialist. J. Information science and Education 1, 114–118 (2008) (in Russian) 12. Colin, K.K.: Information security as a humanitarian problem. J. Open Education. 1(54), 86–93 (2006) (in Russian) 13. Malyuk, A.A., Litinskaya, L.V., Kovalenko, S.Y.: Formation of information culture of the society - key factor in ensuring information security. J. Information Technology Safety 4, 17–24 (2009) (in Russian) 14. Starkova, G.I.: Speech culture as an indicator of general human culture. Economics and Law Institute of Altay (in Russian), http://www.aeli.altai.ru 15. Site about psychology, http://www.psyznaiyka.net/ (in Russian)
Smart Card Based on Hash Function Zulkharnain and Ali Sher American University of Ras Al Khaimah, P.O. Box 10021, RAK, United Arab Emirates {znain,asher}@aurak.ae Abstract. With the ongoing new trends adopted for hacking smart card, there is need of updating the technology used in smart card, and particularly the math techniques used in it. The potential for counterfeiting and fraud may pose significant financial risks to institutions issuing payment obligations in these systems and to other participants in these areas. The world-wide use of public key algorithms to maintain electronic cash security, in smart card should be modified. However, public key algorithms use many computation resources, and make electronic cash schemes inefficient for Smart cards to compute. Furthermore, the denominations of electronic cash are fixed, and the electronic cash spending date is hard to record. It will cause the lack of flexibility while applying. This paper proposes a new electronic cash scheme that utilizes trapdoor hash function technology. The proposed scheme not only allows users to apply for desired denominations of electronic cash and to record transaction dates, but also decreases calculation costs. This is also an efficient procedure capable to be used in today’s smart card. Keywords: Smart card, E-Cash, E-Commerce, mobile devices, signature, hash function.
1 Introduction Smart cards make use of E-cash and since the first electronic cash scheme was presented by D. Chaum [1], multiple electronic cash schemes [2][3][4][5][6] and [7] have been published in the literature. All of these published schemes utilize blind signature to maintain user anonymity. Information on electronic cash must be stored in a database to check for double spending. Yet when there is no expiration date for such information, database storage capacity is depleted quickly which increases maintenance costs. The partially blind signature technique [8][9][10] and [11] has allowed consumers to exchange information with banks, and to store information, such as withdrawal and expiration dates, onto electronic cash, resolving the problems of storage capacity and maintenance costs. There are three life stages for electronic cash: 1. Consumers pay cash to banks in exchange for electronic cash; 2. Consumers pay for goods using electronic cash; 3. Merchants exchange electronic cash at banks. A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 41–51, 2011. © Springer-Verlag Berlin Heidelberg 2011
42
Zulkharnain and A. Sher
Consumers might not purchase goods immediately after receiving their electronic cash. Therefore, they believe banks should pay interest during this period, and give rise to technology that stamps consumption dates on electronic cash [12][ 13] and [14]. The scheme in [13] encodes the consumption dates into hash operations. Consumption dates must be expressed in an encoded format, causing such calculations to be extremely inefficient for mobile devices with low computation abilities [15] and [16]. Banks create consumption dates in [12] which increasing the bank’s needs for communication. Electronic cash in [14] is embedded with the public key of the consumption date, and it allows consumers to create a signature indicating the date of consumption. This method, however, is incompatible for devices with low computing abilities, especially as consumers purchase more. This research proposes a new electronic cash scheme with a trapdoor hash function that provides a consumption date stamp technology without increasing communication. The proposed scheme simultaneously decreases the need for calculating transactions while consumers are shopping. It only requires one integer multiplication and two integer additions, and makes the technology compatible for mobile devices with low computing abilities. Section 2 provides a summary of the trapdoor hash function. Section 3 describes the requirements that electronic cash systems must fulfill. Sections 4 and 5 discuss the security and efficacy of electronic cash.
2 Review of Trapdoor Hash Functions Hash functions are commonly applied to digital signature techniques, and digital signature algorithms can be broken down into three phases: signing key generation, signing document (generating signature), and signature verification. Generally, the procedures for signing document are as follows: Hash functions extract the abstract of the document that is needed to be signed, after which a digital signature algorithm signs the abstract. Collision-resistance is one of the main features of traditional hash functions. For Chameleon functions [17] or trapdoor hash functions [18][19] and [20], the feature of collision-resistance is optional; the owner of a trapdoor key can easily find other collided pre-images and produce the same hash value. For instance, assuming TH ( ) represents trapdoor hash function and the hash value v = TH(h1), after knowing h1, the owner of a trapdoor key can then calculate h2; hence, v = TH(h2) = TH(h1). Computing the value of Chameleon functions online requires a multiplication and modulo operation. Online computation means the amount of computations are required once the target message is determined. In the literature [15], only one modulo operation is required for this computation. In the literature [18] and [19], the computational requirement is further reduced to only one integer multiplication and addition. It is suitable for mobile devices with limited computational resources. The techniques mentioned in literature [18] and [19] are as follows: Let p, q, t, P and Q be prime numbers, the compound number n is the product of P and Q; that is, n = P ⋅ Q, P = 2 ⋅ p ⋅ t + 1, Q = 2 ⋅ q + 1. |P|, |Q|, and |p| represent the encoded bit length of P, Q, and p. Their lengths can be chosen as follows: |P| = |Q| = 512, |p| = l = 160.
Smart Card Based on Hash Function
43
The order of g ∈ Z*n is p. Randomly selecting x ∈R {0, 1}l; Calculating y = gx mod n. The trapdoor key is TK = x, and the public key is HK = (g, n, y). If a message m1 ∈R {0, 1}l, the hash operation is to compute the hash value of the message m1. The processes are as follows. 1. Random Selection: r1 ∈R {0, 1}2⋅l + k, 2. Calculation of the hash value: v = THHK(m1, r1), i.e., v = gr1 ⋅ ym1 mod n. After determining r1, then the hash value of message m1 is v = THHK(m1, r1). The owner of the trapdoor key can begin a trapdoor operation to obtain m2 and r2 such that v = THHK(m1, r1) = THHK(m2, r2). The detailed processes of trapdoor operation are shown as follows: 1. Determining message: m2 ∈ {0, 1}l, 2. Calculating r2, i.e., THTK(m2) = r2 = r1 + (m1 - m2) ⋅ x. Although integer arithmetic is used for calculating r2, the confidential information (m1 m2) ⋅ x is still properly hidden behind the random number r1 because |r1| = 2 ⋅ l + k, often k = 80. Similar information hiding techniques can also be seen in the literature [21][22] and [23].
3 The Proposed Electronic Cash Scheme This research proposes an electronic cash scheme built with an RSA blind signature technique and trapdoor hash function allowing banks to file relevant information, such as: electronic cash denominations, withdrawal dates, expiration dates, and consumption dates. Electronic cash schemes are composed of three roles: consumers (apply for and spend electronic cash), merchants (provide services and goods to consumers), and banks (authorize, distribute, and manage electronic cash), as depicted in Figure 1.
1
3 Customer 6
2
4 5
Merchant
7 Fig. 1. Electronic cash scheme
Electronic cash scheme procedures: 1. Consumers apply for electronic cash from banks 2. Banks distribute electronic cash to consumers
Bank
44
Zulkharnain and A. Sher
3. 4. 5. 6.
Consumers purchase with electronic cash Merchants check with bank for issues of double spending Banks confirm validity of electronic cash Merchants confirm that transaction was successful, pay consumers interest for the duration between application and spending, and provide goods and services 7. Merchants deposit electronic cash into banks, which pay merchants the amount of denomination and interest for the duration between application and deposit dates. (To incorporate the concept of fair electronic payment [24] and [25] into Figure 1, banks should confirm that the consumer have received goods and services before paying the money to merchants.)
3.1 Definition of Notations The parameters and symbols of electronic cash scheme can be divided into: systems, banks, and consumers, described as follows: The parameters and symbols for systems: 1. EXD: Expiration date of electronic cash 2. APD: Application date of electronic cash 3. SPD: Electronic cash purchase date 4. h(⋅): collision-free one-way hash function 5. ||: bit concentration symbol 6. k, l: parameters of confidentiality in accordance with the development level of security, for example: k = 80, l = 160 The parameters and symbols for banks: 1. p, q, t, P and Q: both are prime numbers, with P and Q the same code length, and P = 2 ⋅ p ⋅ t + 1, Q = 2 ⋅ q + 1. |P| = |Q| = 512, |p| = l 2. product of two large prime numbers, such as n = P ⋅ Q 3. g g ∈ Z*n with an order of p 4. (dsig, ever): bank utilizes the RSA cryptosystems to create keys of signing and verifying, ever ∈R Zn, gcd(ever, φ(n)) = 1, φ(n) = (P - 1)(Q - 1), ever ⋅ dsig = 1 modφ(n); ever is public key for verifying signatures, dsig is secret key for signing documents.
:
The parameters and symbols for consumers: 1. x: consumer chooses a secret key for his trapdoor hash function, for example: TK = x, and x ∈R {0, 1}l 2. HK: public key of consumer’s trapdoor hash function, for example: HK = (g, n, y = gx mod n) 3. m1, r1: consumers choose two messages randomly, for example: m1 ∈R {0, 1}l, r1 ∈R {0, 1}2⋅l + k 3.2 Stages of the Proposed Electronic Cash Scheme There are four stages (phases) of the electronic cash schemes: 1. apply for electronic cash, 2. exterminate blind factors, i.e. receive electronic cash, 3. spend electronic cash, 4. deposit electronic cash into banks. Detailed descriptions of each step are as follows:
Smart Card Based on Hash Function
45
(1) Applying for electronic cash Consumers apply for w of electronic cash by first discussing relevant dates with bank and performing the following calculations once consensus has been reached: 1. 2. 3. 4. 5. 6. 7.
create a random trapdoor key x ∈R {0, 1}l, calculate y = gx mod n create a random message m1 ∈R {0, 1}l and a random number r1 ∈R {0, 1}2⋅l + k execute hash operation, A = THHK(m1, r1) = gr1 ⋅ ym1 mod n application date and denomination α′ = (EXD || APD || w) select a blind factor r ∈R Z*n use a one way hash function to generate messages h(A || EXD || APD || w ) calculate α = rever ⋅ h(A || EXD || APD || w) mod n
Consumer sends messages to bank after the calculation is complete, and store m1, r1, x for backup. The bank will verify the accuracy of the dates upon receiving the information and sign with private keys. The calculations are as follows: 1. 2.
verify date based on information (EXD || APD || w) create blind signature t = αdsig ⋅ (EXD || APD || w) mod n
The bank resends the blind signature to consumers while deducting w amount of electronic cash from the consumers’ accounts. The process of applying for electronic cash is shown in Figure 2, and is the first step in the electronic cash system procedures (shown in Figure 1).
Bank
Customer l
x
x∈R{0, 1} , y = g mod n m1 ∈R {0, 1}l, r1 ∈R {0, 1}2⋅l + k A = THHK(m1, r1) = gr1 ⋅ ym1 mod n
α′ = (EXD || APD || w), r ∈R Z*n α = rever ⋅ h(A || EXD || APD || w)
α, α′ t = αdsig ⋅ (EXD || APD || w) mod n
mod n Fig. 2. Stages of applying for electronic cash
(2) Eliminate blind factor (receive electronic cash) After users receive the blind signature t from banks, they can eliminate blind factors and receive electronic cash by performing the following calculations: 1.
Eliminate blind factor, namely calculate s = t ⋅ r-1 = h(A || EXD || APD || w)dsig ⋅ (EXD || APD || w) mod n.
46
2.
Zulkharnain and A. Sher
Confirm signature, that is, verify sever ?= h(A || EXD || APD || w) ⋅ (EXD || APD || w)ever mod n. Assuming the signature is correct, ((A || EXD || APD || w), s) represents w denomination of electronic cash.
The process for eliminating the blind factor is depicted in Figure 3, and the step 2 of Figure 1 of electronic cash system is completed here.
Bank
Customer t s = t ⋅ r-1 = h(A || EXD || APD || w)dsig ⋅ (EXD || APD || w) mod n sever
? =
h(A || EXD || APD || w) ⋅ (EXD || APD || w)ever mod n
ECash = ((A || EXD || APD || w), s)
Fig. 3. Eliminate blind factor (receive electronic cash)
(3) Consumption (spend electronic cash) When a customer wishes to purchase an item, he/she performs the following steps to complete transaction. 1. consumers sign on documents m2 = (EXD || APD || SPD); contents include expiration, application, and consumption dates of electronic cash 2. consumers withdraw information stored in database (m1, r1, x) 3. use trapdoor key x and carry out trapdoor operation, THTK(m2) = r2 = (m1 - m2) ⋅ x + r1 Additional information (EXD, APD, SPD, s, r2, w, y) is sent to merchants after calculation. Purchasing procedures are described in Figure 4, and the process of the third step in Figure 1 is completed here. (4) Depositing Merchants perform the following calculations upon receiving electronic cash from consumers: 1. decision document m2 = (EXD || APD || SPD) 2. calculate A = THHK(m2, r2) = gr2 ⋅ ym2 mod n 3. verify sever ?= h(A || EXD || APD || w)(EXD || APD || w)ever mod n
Smart Card Based on Hash Function
Customer
47
Merchant
m2 = (EXD || APD || SPD) r2 = (m1 - m2) ⋅ x + r1 EXD, APD, SPD, s, r2, w, y m2 = (EXD || APD || SPD) A = gr2 ⋅ ym2 mod n sever
? =
h(A || EXD || APD || w)(EXD ||
APD || w)ever mod n
Fig. 4. Consumption (spend electronic cash)
Bank
Merchant EXD, APD, SPD, s, r2, w, y
m2 = (EXD || APD || SPD) A = gr2 ⋅ ym2 mod n sever
? =
h(A || EXD || APD || w)(EXD || APD || w)ever mod n
Store (EXD, APD, SPD, s, r2, w, y) in data base
Fig. 5. Depositing
Assuming that the above calculations are accurate, merchants may request banks to confirm that the electronic cash is not involved in double spending. If the electronic cash is verified, merchants may then accept the transaction and pay interest to the consumers while the bank records the information into their databases to prevent double spending. The bank then directs the w amount of cash that is spent by consumers and deposits interest into the merchants’ accounts. The depositing stage is depicted in Figure 5, steps 4-7 of the electronic cash system in Figure 1 are completed in this stage.
48
Zulkharnain and A. Sher
4 Security Analysis This section discusses the nature of electronic cash regarding its accuracy, uniqueness, credibility, anonymity, double spending, and fraud interest. Accuracy: Both the electronic cash that banks distributed to consumers and the electronic cash with consumption dates can be verified by any entity. The bank’s public key is sufficient to verify the electronic cash. Unforgeability: Due to the factorization is a mathematical assumption, blind signatures are secure. Forgers would have to know the bank’s private key in order to forge successfully a signature. Consumers will use their own private keys to calculate hash collision values. If anyone attempted to change the relevant dates, he or she would need to know the consumer’s private keys to reach the same hash collision values. Undeniability: Banks must give private signatures before electric signatures can be created and provided to consumers. Consumers then use the public keys provided by banks to verify the cash so banks cannot deny issuing electronic cash. Consumers use private trapdoor key to calculate the pre-imaged of a pre-determined hash value, and transfer relevant information and electronic cash to merchants for the latter to verify. Upon receiving the necessary information, merchants can calculate the accurate pre-determined hash value so consumers cannot deny ever paying the electronic cash. Anonymity and un-traceability: When consumers are applying for electronic cash, their information is mixed with blind factors before being submitted to banks. Once banks return the information with bank-authorized signatures, consumers then eliminate the blind factors to receive the actual message. No one is able to track the information without knowing which blind factor consumers randomly chose, making such process safe and anonymous. For consumers who request special amounts of electronic cash, they are able to arrange for specific denominations so that their transactions remain anonymous and untraceable. Double spending: In our proposal, the bank serves as a trustworthy third party. All information stored in the bank’s database is legal and credible. The records of spending electronic cash are stored in the bank and will only be deleted when they have expired. Upon receiving electronic cash, merchants first verify the cash, and then connect to the bank to confirm whether this electronic cash is stored in the bank database. It is easily preventing cases of double spending. Fraud interest: when applying for electronic cash, relevant dates are encoded within the cash and are signed by banks. Due to consumers and merchants have conflicting interests, the two parties cannot conspire together to gain fraud interest.
Smart Card Based on Hash Function
49
5 Performance Analysis Electronic cash systems [12] require banks to mark consumption dates; therefore, communication costs have increased. Table 1 can only analyze electronic cash systems [14]. Major roles involved in the electronic cash system include banks, merchants, and consumers. The life cycle of electronic cash includes receiving, spending, and depositing. Table 1 compares the amounts of electronic cash for each role during the cash’s life cycle. In Table 1, the difference between the proposed scheme and Juang’s is in the calculation of the consumption stage. Juang’s scheme requires one modular exponentiation operation (Exp), two modular multiplications (MM), one modular inverse operation (Inv), and one hash function operation (H); whereas this study only requires one integer multiplication (IM). According to an experiment recorded in [19], the calculation time for MM is roughly 1.4 times longer than IM. Assuming that the index is 160 bits, and a modular exponentiation operation requires an average of 240 modular multiplications, Juang’s scheme’s calculation time is roughly 336 (1.4 * 240) times longer than the proposed scheme. Although banks and merchants tend to have enough calculation resources, most consumers do not, especially those consumers with mobile devices or personal digital assistants (PDAs), make the proposed scheme more practical and user-friendly. If the date and coding are not taken into account, the electronic cash in Juang’s scheme (r, s, r′, s′) has a code length of 1504 (1024, 160, 160, 160) bits. The electronic cash in this study (s, r2, y) has a code length of 2448 (1024, 400, 1024) bits, which is an additional 944 bits. If the transmission rate is originally 100 kbps, then the rate will be reduced by 10 ms (ms = 10-3 second). However, most mobile devices run on memory capacities of many hundreds of K bytes (1 byte = 8 bits); therefore, 944 bits is an insignificant increase. Table 1. Comparison chart between proposed scheme and Juang’s scheme [14]
Receiving electronic cash
Consuming
Depositing
Consumer Bank Juang’s Our scheme Juang’s scheme [14] scheme [14] 6 Exp 6 Exp 1 Exp 6 MM 4 MM 1 MM 2H 1H 1 Inv 1 Inv 1 Exp 1 IM 2 MM 1H 1 Inv 4 Exp 5 MM 1H 1 Inv
Our scheme 1 Exp 1 MM
Merchant Juang’s scheme [14] 0
4 Exp 5 MM 1H 1 Inv 4 Exp 2 MM 1H
Our scheme 0
4 Exp 2 MM 1H
50
Zulkharnain and A. Sher
6 Conclusion The smart card-electronic cash scheme proposed in this paper utilizes a blind signature to maintain consumer anonymity while spending electronic cash. The proposed scheme also allows consumers to request their own electronic cash denominations, which not only solves the problem of double spending, but also provides complete date and interest information, thus making the smart card system secure. Considering that smart cards are gaining popularity, the design also incorporates a hash function, which can help decreasing consumer’s needs for calculation. Such a system is compatible for smart cards with low computing abilities and thus making electronic services increasingly popular.
References 1. Chaum, D.: Blind Signatures for Untraceable Payments. In: McCurley, K.S., Ziegler, C.D. (eds.) Advances in Cryptology 1981 - 1997. LNCS, vol. 1440, pp. 199–203. Springer, Heidelberg (1999) 2. Brands, S.: Untraceable Off-Line Cash in Wallets with Observers. In: Stinson, D.R. (ed.) CRYPTO 1993. LNCS, vol. 773, pp. 302–318. Springer, Heidelberg (1994) 3. Camenisch, J., Piveteau, J.M., Stadler, M.: An Efficient Fair Payment System Protecting Privacy. In: Gollmann, D. (ed.) ESORICS 1994. LNCS, vol. 875, pp. 207–215. Springer, Heidelberg (1994) 4. Chaum, D., Fiat, A., Naor, M.: Untraceable Electronic Cash. In: Goldwasser, S. (ed.) CRYPTO 1988. LNCS, vol. 403, pp. 319–327. Springer, Heidelberg (1990) 5. Chaum, D., Pedersen, T.P.: Wallet Databases with Observers. In: Brickell, E.F. (ed.) CRYPTO 1992. LNCS, vol. 740, pp. 89–105. Springer, Heidelberg (1993) 6. Ferguson, N.: Single Term Off-Line Coins. In: Helleseth, T. (ed.) EUROCRYPT 1993. LNCS, vol. 765, pp. 318–328. Springer, Heidelberg (1994) 7. Okamoto, T., Ohta, K.: Universal Electronic Cash. In: Feigenbaum, J. (ed.) CRYPTO 1991. LNCS, vol. 576, pp. 324–337. Springer, Heidelberg (1992) 8. He, D., Chen, J., Hu, J.: An ID-based client authentication with key agreement protocol for mobile client-server environment on ECC with provable security. Information Fusion (2011) 9. Abe, M., Okamoto, T.: Provably Secure Partially Blind Signatures. In: Bellare, M. (ed.) CRYPTO 2000. LNCS, vol. 1880, pp. 271–286. Springer, Heidelberg (2000) 10. Fan, C.I., Lei, C.L.: Low-computation partially blind signatures for electronic cash, IEICE Transactions on Fundamentals of Electronics. Communications and Computer Sciences E81-A (5), 940–949 (1998) 11. Yang, F.Y., Jan, J.K.: A Secure Scheme for Restrictive Partially Blind Signatures. In: The Sixth International Conference on Information Integration and Web-based Applications & Services IIWAS 2004, Jakarta Indonesia, pp. 541–548 (2004) 12. Chang, C.C., Lai, Y.P.: A flexible Date-attachment Scheme on E-cash. Computers & Security 22(2), 160–166 (2003) 13. Fan, C.I., Chen, W.K., Yeh, Y.S.: Date Attachable Electronic Cash. Computer Communications 23(4), 425–428 (2000) 14. Juang, W.S.: D-Cash: A Flexible Pre-paid E-cash Scheme for Date-attachment. Electronic Commerce Research and Applications 6(1), 74–80 (2007)
Smart Card Based on Hash Function
51
15. Keating, W.S.: Performance analysis of AES candidates on the 6805 CPU core, AES Round 2 public comment, April 15 (1999) 16. Yang, C.H.: Performance Evaluation of AES/DES/Camellia on the 6805 and H8/300 CPUs. In: 2001 Symposium on Cryptography and Information Security (SCIS 2001), pp. 727–730 (2001) 17. Krawczyk, H., Rabin, T.: Chameleon signatures. In: Symposium on Network and Distributed Systems Security (NDSS 2000), pp. 143–154 (2000) 18. Shamir, A., Tauman, Y.: Improved online/Offline signature schemes. In: Kilian, J. (ed.) CRYPTO 2001. LNCS, vol. 2139, pp. 355–367. Springer, Heidelberg (2001) 19. Yang, F.Y.: Improvement on a trapdoor hash function. International Journal of Network Security 9(1), 17–21 (2009) 20. Yang, F.Y.: Efficient trapdoor hash function for digital signatures. Chaoyang Journal 12, 351–357 (2007) 21. Okamoto, T., Tada, M., Miyaji, A.: Efficient ‘on the Fly’ Signature Schemes Based on Integer Factoring. In: Pandu Rangan, C., Ding, C. (eds.) INDOCRYPT 2001. LNCS, vol. 2247, pp. 275–286. Springer, Heidelberg (2001) 22. Pointcheval, D.: The Composite Discrete Logarithm and Secure Authentication. In: Imai, H., Zheng, Y. (eds.) PKC 2000. LNCS, vol. 1751, pp. 113–128. Springer, Heidelberg (2000) 23. Poupard, D., Stern, J.: On the fly signatures based on factoring. In: Proceedings of the 6th ACM Conference on computer and communications security (CCS), pp. 48–57 (1999) 24. Asokan, N., Shoup, V., Waidner, M.: Optimistic fair exchange of digital signatures. IEEE Journal on Selected Areas in Communications 18(4), 593–610 (2000) 25. Yang, J.H., Chang, C.C.: An efficient fair electronic payment system based upon non-signature authenticated encryption scheme. International Journal of Innovative Computing, Information and Control 5(11A), 3861–3874 (2009)
RDF Data Models in Oracle Zdenka Telnarova University of Ostrava, 30. dubna 22, 701 03 Ostrava, Czech Republic
[email protected]
Abstract. This contribution deals with the RDF data model and its implementation in the Oracle system. After reminding the RDF and RDFS models the reader is informed about the possibilities how to define these models in the Oracle system, how to input data into the database, how to define the rule base and to fulfill it with further rules. The indexes, data types, the usage of the RDF and RDFS bases for a modified command "select" which provides information from RDF data and rules follow. Finally, a demonstration example is given. Keywords: RDF, RDFS, ORACLE, RDF data type in Oracle, rule, query.
1 Motivation The idea of a semantic web was presented for the first time in May 2001. Tim Berners-Lee, the author of the current web technology and the director of the Consortium W3C together with other co-authors set it into operation in the article [1] published in the Scientific American. The semantic web means not only to use a seemingly infinite quantity of data on the web but to complete it with its meanings. Current relation databases contain a great quantity of data which may be meaningfully interpreted and thus provide their users with the needed information. Data in relation databases is mapped into mutually interconnected relations which reveal their semantic meanings. The semantic web offers means using universal semantic information to connect two different engines employing different data sources and to exchange and process them effectively. Thus, for the implementation of the semantic web, semantic metadata has to be added to data sources. This will be the only way for the engines to be able to process the data effectively. In case the data is completed with semantic metadata associated with its corresponding data, the computer can deduce further data or statements, e.g. it can understand what a data source is about and how it is related to another data source. XML (eXtensible Markup Language) paved the way by means of adding metadata in form of tags readable for humans and describing the data. For example, an XML document may include information about the author of the web site, relevant key words, etc. XML offers syntactic interoperability, not semantic interoperability. For instance, let’s have one element denoted as 12.000 and another A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 52–64, 2011. © Springer-Verlag Berlin Heidelberg 2011
RDF Data Models in Oracle
53
one denoted as 12.000 ; there is no mode for the computer to find out that it is, in fact, the same "item". The semantic web helps to solve this problem by introducing semantic metadata which is comprehensible both for a human and the engine. One possible step making data comprehensible for the engine is to transform this data into a unified format so that the field of type "income" always has the same format and always includes the same type of information. Considering the huge quantity of data from various data sources, it is obvious that this method of "data standardization" cannot eliminate the concerned problem. Another possible step towards the semantic web requires the data from various domains to be sorted according to its properties and mutual relations to other data. These problems are solved by means of the RDF, RDFS and OWL technologies which are based on conceptualization of data available on the internet and aimed at knowledge representation, its sharing and reuse.
2 Data Model of the Semantic Web Each logic language differentiates syntax and semantics. Syntax defines how to form statements and formulas, while semantics deals with their meanings and/or interpretations. Both syntax and semantics may be considered as levels of abstraction. Syntax is a concrete level while semantics is an abstract level. RDF brings an interlayer into this consideration designated as abstract syntax. RDF is a language designed for description of metadata, especially for information saved on the Web. RDF has a simple data structure which may be modeled effectively using directed graphs. To describe the metadata in a network data model based on RDF, we use triples consisting of two components representing nodes of graphs and the third component is the directed edge describing the relationship between the nodes. The triples are saved in a logical network. Thus, the RDF statements are expressed in the following triples: {subject or source, predicate or property, object or value}. Each triple is a complete and unique fact concerning the specified domain and its can be represented by an edge in a directed graph. RDF is a standard for representing data, especially for representing metadata on web sources. We may make use of the simplicity of RDF for interoperability between applications while their format which may be read by an engine enables an automatic web sources process. The simple idea of the RDF technology is based on the description of existing data on the web by means of URI (see 2.1.2) by the help of which every source is defined. URI is used both for identification of resources and construction of propositions about these resources. The RDF triples enable the engine to form logical propositions based on relationships between subjects and objects. As RDF uses URI to identify resources, each resource is connected to a unique side on the web. RDF describes syntax and not the semantics of the resources. To define the semantics, we need RDFS and OWL.
54
Z. Telnarova
2.1 General RDF Data Model For the possibility to fill in automatically readable semantic information concerning the content of the documents, it is necessary to formulate the information in generally shared concepts based on properties and mutual relationships, and not on key words as it was usual in the past. “The world” which is to be modeled by formal means must be seen through abstractions of exempted relevant concepts and their mutual relations. The foregoer of the conceptually oriented paradigm in the representation of knowledge is the E-R or object paradigm which is nowadays considered conventional. However, there is one basic difference. Namely, modeling on the conceptual level in an E-R model or in an object oriented model is based on the definition of the entity set or class by means of properties which these entity sets or classes have to share. Thus, it is a principle which may be called the class-oriented principle. In terms of the “conceptual” paradigm the basic concept is the property, while the set of entities/classes is characterized by generally shared basic attributes and mutually differing in other properties. In this property-oriented case, the emphasis is especially laid on the generally shared properties which represent the basic components of the description of the modeling reality. 2.2 RDFS RDF Schema (RDFS) serves to describe concepts used in RDF data and contains type constrains on objects and values in triples. In case we have a scheme, we can validate data in RDF with respect to this scheme. The RDFS language starts from the object oriented approach, it is based on classes and ISA-hierarchies of these classes with set semantics. The following categories are at the highest level of concepts: rdfs:Resource, rdfs:Class and rdfs:Property. Classes in RDFS are not defined by means of properties and have no associations to other classes. The properties themselves are objects and may be reinherited for several times. The ISA-hierarchy is modeled in RDF Schema by means of the following given property types: instanceOf, subClassOf: instanceOf defines the relationship between the resource and element of Class, the subclassOf defines the relationship between two elements of Class and is supposed to be transitive. Constraint is a subclass of Property and has two basic instances – range and domain applied to property types having class as a value. They define the value domain or the domain of property types. The RDF model may be considered a graph or a set of triples. The RDF scheme (RDFS) defines classes (subjects and objects) and properties and concerns the data semantics. A class is a set of resources and conforms to the designation of a type or category of another representation. The rdf property is used to create a relationship in the class where the resource is an element. Thus, a certain type of rdfs:Class is determined to all classes. Properties are also resources and members of the rdf Property class. This class may be reduced by the help of an additional semantics for properties rdfs: range, which reduces the range of properties.
RDF Data Models in Oracle
55
RDFS is used for arranging dictionaries describing groups of related RDF resources and relationships among such groups of resources. The dictionary defines admissible properties which may be matched with the RDF resources within the given domain. RDFS enables to create classes of resources which share common properties. We can simply say that an RDF model represents data, while an RDFS model represents metadata. When we use – similarly to the case of RDF - as building stones RDFS triples, then the resources are defined as instances of classes. A class is also a resource and each class may be a subclass of another class. Such hierarchy of semantic information enables an engine to define the sense of resources based on their properties and classes. Thus, RDFS is a simple language for defining dictionaries which express relationships between resources. Much more expressive and richer language to define dictionaries for semantic web is the OWL language. OWL defines types of relationships which may be used in an RDF model and indicates their hierarchy and relationships between various resources. It is a language for defining ontology consisting of taxonomy and a set of deductive rules by the help of which an engine can derive logical consequences.
3 Contribution of Semantic Data Model The term data “modeling” first appeared thirty years ago in relation to the development of relational databases. Moving forward, data modeling has become an analysis tool of data semantics and its structure, usually relating to a specific problem. At present, the concept of data semantics is in the centre of our interest. Knowledge engineers have accepted that, for subsequently developed systems, formalization of semantic side of knowledge is essential (in terms of intercommunication using program agents). Both semantic data models and data models are dependent on conceptualization, therefore they both have to respect the structure and the rules of the domain. Abstraction is always influenced by an application aspect. Metamodel and language that are defined on its basis, therefore have to be built on a common core of integration of possible abstractions. In contrast to the data model oriented on solving a problem in a domain, a semantic data model should be as general as possible and also as independent as possible with respect to its application. In this context it is necessary to always keep in mind the following fundamental principles: • Semantic data model is in fact a knowledge base representing generally accepted semantics of a given domain. In contrast to a data model, a semantic data model itself is relatively independent of any special application in the domain since, it primarily consists of generic (i.e. generally usable) knowledge derived from various applications related to the domain. Similarly to data modeling that made information of a given domain (contained in the data) independent of the applications using the data, a semantic data model provides a possibility of existence of the basic relevant domain knowledge independent of applications working with this knowledge. Semantic data model represents knowledge which formally specifies shared logical theories of the domains considered.
56
Z. Telnarova
• Data model represents a structure and integrity of data elements of a specific application. In this case it is not expected that conceptualization (and the subsequently created data model dictionary) would be shared by other applications. Data semantics is usually derived from the shared concept of data between the designers and users of the model. The model is updated as new requests are solved. Both semantic data models and data models are dependent on conceptualization; therefore, they have to respect semantic structure and rules of the respective domain. In contrast to a model oriented on problem solving, a semantic data model should be general and application-independent. There may be a number of different conceptualizations used in a domain. The differences between them would primarily appear at lower levels of abstraction, i.e. towards concrete specializations of the domain in the given context of the application. Reusability of semantic data model presumes the same conceptualization. The difference between the data model and the semantic data model of the same domain primarily lies in their level of generality. In order to distinguish a data model from a semantic data model, it is necessary to distinguish dual purpose concepts: 1. Purpose of modeling and sharing knowledge in software engineering, where data models typically specify the structure and integrity of the created data files. (In this case creating a data model, e.g. for a company, it depends on specific needs of the company. Data semantics are usually derived from the shared concept of data between the designers and users of the model. The model is updated as new requests are solved.) 2. Purpose of reusability. (e.g. in the open world of a semantic web, ontology represents knowledge which formally specifies shared logical theories of a given application domain.) Since there are two basics aspects, i.e. integration and reusability, our objective is to focus on the principle of building a semantic data model from data in relational databases.
4 RDF Network Data Model in Oracle As we mention above, contribution of the semantic approach to database modeling is evident. The question is how to implement semantic approach into the standard database management systems. Our article concentrates on one of the most frequent system Oracle, summarizes familiarity with the topic and thoroughly describes (including a practical example) the activity in creating semantic data model and its instances. Our approach is based on [8], [9], [10]. All RDF triples are divided and stored in Oracle as input data of tables in a system scheme (MDSYS scheme). The RDF triple is treated as one database object. All subjects and objects of triples are mapped into network nodes and properties are mapped into edges possessing a subject as a starting node and an object as an ending node. An RDF model accepts empty nodes, URI, simple literals and type literals. Before specifying the notions of simple and standard literal more precisely, we will introduce the definition of a data type.
RDF Data Models in Oracle
57
Definition. A data type shall be signed as D = (L, H, lh, uriD), where uriD ⊂ URIREF is a set of URI references, L is a set of Unicode chains, the so-called lexical space. Set H is a set of values of a given type. We call the L → H representation a lexically-value representation. Such representation is not simple, for each space element of values we may have more elements of a lexical space. We call such elements the lexical representation of values. Literals are constants together with a facultative specification of a data type or language. They designate values as numbers, chains, and data. A literal may be also an object of an RDS triple, but not a subject or a predicate. A simple literal is a chain of Unicode characters in brackets after which the language tag may follow, e.g. „a1…an“@cz is a simple literal, where a1, a2, …an are signs in Unicode and @cz is a tag for the Czech language. A type literal is a chain in brackets after which URI reference of a data type separated by ^^ must follow. A type literal designates an element from the space of values of a data type which we attain when applying the lexical-value representation to the literal chain. Oracle database has type names URIType for storing the URI (HttpUri, DBUri a XDBUri) instances. These types are used to preserve the names of nodes and edges in the RDF model. Oracle RDF data model satisfies the following requirements: • a subject must be URI or an empty node • a property must be URI • an object may be of any type (URI, empty node, literal). Value NULL is not supported. 4.1 Metadata for RDF Models Information on all RDF models created in the Oracle database is stored in the MDSYS.RDF_MODEL$ system table. The model may be established using a builtin procedure SDO_RDF.CREATE_RDF_MODEL. When creating the model it is necessary to specify the name of the model and the structure of the table in which the RDF data will be kept. The system generates the ID of a model automatically. The administration of the created models and the corresponding model is organized by the Oracle system. The user can change neither the model structure nor the corresponding tables. Namespaces The MDSYS_RDF_NAMESPACE$ system table contains information concerning namespaces (NAMESPACES) which are used in RDF XML documents and makes the documents more readable. In the Oracle RDF model the namespaces are stored directly with their subjects, properties and objects. Oracle adds the namespace into the MDSYS.RDF_NAMESPACE$ table automatically as soon as the space is created by activating the , SDO_RDF.ADD_NAMESPACES procedure. Subjects and objects The RDF subjects and objects are mapped into nodes of the network data model. As already mentioned, a subject may be a URI or empty node, an object may be an empty
58
Z. Telnarova
node or a literal. Both simple and type literals are used. Empty nodes are used if unknown objects or subjects are to be represented. However, empty nodes are also used when the n-ary relationship between a subject or an object exists (the case of containers). The RDF properties are mapped into edges and represent a complete RDF triple. If an RDF triple is input into a database, it is checked whether the subject, property and objects already exist in the database. In case they already exist, the new triple cannot be input into the database. In case they do not exist, three lines are input into the RDF_VALUE$ table. 4.2 Materialization of RDF Statement Materialization of RDF statement enables to use these statements as subjects or objects in other RDF statements. Let’s have the following statements: 1. 2. 3.
Person A is a good candidate for a working position. Person B says that person A is a good candidate for a working position. Person C denies that person A is a good candidate for a working position.
To be able to apply statement 1 while specifying statement 2 and statement 3, statement 1 has to be materialized, i.e. statement 1 must exist in the database e.g. in the following form: Thus, statements made by persons B and C may be saved in the following way: 4.3 Rules and Database of Rules A rule is an object in Oracle RDF which is used for deducing new objects from RDF data. A rule is identified by a name and its composition is as follows: IF a part for antecedents Facultative filter for other conditions THEN a part for a consequent For instance, a rule stating that an author of a course is also its tutor may be written in the following way: (‘tutor_rule‘,‘ (?x :authorOf ?y)‘, NULL, ‘(?x :tutorOf ?y)‘, SDO_RDF_Aliases (MDSYS.RDF_Alias(‘‘,‘http://some.org/test/‘)))
RDF Data Models in Oracle
59
where: tutor_rule is the designation of the rule x,y are resources authorOf, tutorOf are properties (‘tutor_rule‘, is the designation of the rule ‘ (?x :authorOf ?y)‘, antecedent NULL, filter ‘(?x :tutorOf ?y)‘, consequent SDO_RDF_Aliases MDSYS.RDF_Alias (‘‘,‘http://some.org/test/‘))) [5] The database of rules is an object containing rules. Oracle supports the RDFS database of rules according to the definition W3C (www.w3.org/TR/rdf-mt/). The database of rules may be created in Oracle by activating the built-in procedure SDO_RDF.CREATE_RDF_NETWORK. In addition,a database of rules defined by a user may also be created using the SDO_RDF_INFERENCE.CREATE_RULEBASE procedure. For each database of rules, a table for storing the rules into the database and the corresponding view is created by means of which it is possible to delete and modify the rules in the database of rules. The following example illustrates how the database of rules „family“ is created and the rule "grandparent" is inserted into this database. 1. Creating the database of rules defined by the user applying the built-in. SDO_RDF_INFERENCE.CREATE_RULEBASE EXECUTE SDO_RDF_INFERENCE.CREATE_RULEBASE (‘family‘) procedure; 2. Inserting the rule for the definition „grandparent“. INSERT INTO MDSYS.RDFR_family VALUES (‘grandparent_rule‘,‘(?x : parent ?y) (?y :parent ?z)‘, NULL,‘(?x :grandparent ?z)‘,SDO_RDF_Aliases (MDSYS.RDF_Alias(‘‘,‘http://www.example.org/family/‘))); 4.4 Indexes and Rules If we want to recall rules in queries (or if you like, SELECT contains a built-in SDO_RDF_MATCH procedure which indicates a reference into the database of rules) an index has to be created on the corresponding rules. An index is a database object containing calculated triples which can be derived applying specified rules from the database of rules to a specified model, or - if you want – set of models. To create an index to a set of rules, the built-in SDO_RDF_INFERENCE.CREATE_ RULES_INDEX procedure may be employed. The following example creates an index (designed as ‘rdfs_rix_family‘) to rules specified in the database of rules "family“ . Begin SDO_RDF_INFERENCE.CREATE RULES_INDEX( ‘rdfs_rix_family‘, SDO_RDF_Models(‘family‘), SDO_RDF_Rulebases (‘RDFS‘, ‘family‘)); End;/
60
Z. Telnarova
4.5 RDF Data Types, Constructors and Methods RDF data triples are, as given above, represented by means of SDO_RDF_TRIPLE object data types. Such triples are persistently stored in the database tables, while the data type for a triple to be stored is the SDO_RDF_TRIPLE_S object data type (S stands for persistent storage). The SDO_RDF_TRIPLE type is used to show the RDF triples, while SDO_RDF_TRIPLE_S type is used for storing triples into database tables. 4.6 Methods of the SDO_RDF_TRIPLE_S The SDO_RDF_TRIPLE_S object date type disposes of the following four methods for obtaining triples or their parts: GET_TRIPLE() RETURNS SDO_RDF_TRIPLE GET_SUBJECT() RETURNS VARCHAR2 GET_PROPERTY() RETURNS VARCHAR2 GET_OBJECT() RETURNS VARCHAR2 The following example shows an article from the table of articles where id = 1 SELECT X.TRIPLE.GET_TRIPLE () AS TRIPLE FROM ARTICLE_RDF_DATA X WHERE X.id=1; Depending on the concrete data, the answer to this question will be, e.g.: TRIPLE(SUBJECT, PROPERTY, OBJECT) SDO_RDF_TRIPLE (‘http://www.articles.cz/article1‘, ‘http://www.articles.cz/article_name‘, ‘Introduction to …) 4.7 Query over the RDF Data To get RDF data, the Oracle disposes of the SDO_RDF_MATCH function possessing the following arguments: SDO_RDF_MATCH (
Query Model Rule_base Alias Filtr Index_status RETURN ANYDATASET;
VARCHAR2, SDO_RDF_MODELS, SDO_RDF_RULEBASES, SDO_RDF_ALIASES, VARCHAR2, VARCHAR2)
Argument Query the SDO_RDF_MATCH function is required. It is either a literal or concatenation of literals with one or more triple samples with one or more triple examples usually containing variables. All variables contained in the enquiry must be only free. Each triple is an atom enclosed in inverted commas.
RDF Data Models in Oracle
61
E.g. ‘(?x, :father ?y)‘ is one triple which may be recorded as atom father(x,y) in the predicate logic of the first order. Variables x and y are free variables because they are not within the reach of any quantifier. Argument enquiry may contain one or more such triples. Thus, as follows from the RDF model, individual triple components represent subject, property and object. Argument Model identifies an RDF model, it is a SDO_RDF_MODELS data type, which is defined as follows: TABLE OF VARCHAR2(25). Argument Rule_base identifies a base of rules; the rules are to be used when evaluating the enquiry. SDO_FDR_DATARULES is a data type, defined as TABLE OF VARCHAR2(25). Argument Alias identifies one or more namespaces which will be used when evaluating the enquiry. Argument Filtr, in case it does not assume the NULL value , may be a chain in the form of WHERE clause without the WHERE key word. An example of a query which shows all pairs of fathers and sons in the RDF model Premyslovci. select x,y from table (SDO_RDF_MATCH('(?x :grandParentOf ?y)',SDO_RDF_MODELS('Premyslovci'),SDO_RDF_RULEBASES('R DFS','Premyslovci_rb'),SDO_RDF_ALIASES(SDO_RDF_ALIAS('' ,'http://www.osu.cz/premyslovci/')),null,'incomplete')) ; The example supposes that there exists: The RDF model Premyslovci Base of rules Premyslovci_rb, where at least the “?x :grandParentOf ?y” rule exists Namespace 'http://www.osu.cz/premyslovci/'. 4.8 Import and Export of RDF Data To download data into the Oracle RDF model we may choose from the following two possibilities: either Java API or SQL command INSERT. Java API may be used for exporting. In the following part, we are going to describe downloading of RDF data by means of the INSERT command using constructor SDO_RDF_TRIPLE_S. Command syntaxes: INSERT INTO table VALUES (SDO_RDF_TRIPLE_S (‚model’,’subject’,’ property’, ‘object’)) Before downloading, the following structures have to be created (the sequence of the given structures has to be respected). Table space RDF network RDF data RDF model
CREATE TABLESPACE EXECUTE SDO_RDF.CREATE_RDF_NETWORK CREATE TABLE EXECUTE SDO_RDF.CREATE_RDF_MODEL
62
Z. Telnarova
5 Example The following example demonstrates creating the RDF model and all other needed constructors as well as importing the data into it. Creating the base of rules and downloading the rules into a database is also illustrated. 5.1 Creating Table Space for the RDF System Tables CREATE TABLESPACE rdf_tblspace DATAFILE 'rdf_Premyslovci.dat' SIZE 1024M REUSE AUTOEXTEND ON NEXT 256M MAXSIZE UNLIMITED SEGMENT SPACE MANAGEMENT AUTO; 5.2 Creating the RDF Network EXECUTE SDO_RDF.CREATE_RDF_NETWORK('rdf_Premyslovci'); 5.3 Creating the Table for Downloading RDF Data CREATE TABLE PREMYSLOVCI_RDF_DATA (id NUMBER, triple SDO_RDF_TRIPLE_S); 5.4 Creating the RDF Model EXECUTE SDO_RDF.CREATE_RDF_MODEL('PREMYSLOVCI', 'PREMYSLOVCI_RDF_DATA', 'TRIPLE'); 5.5 Downloading the Triples -- Borivoj is the father of Spytihnev. INSERT INTO PREMYSLOVCI_RDF_DATA VALUES (1, SDO_RDF_TRIPLE_S('PREMYSLOVCI', 'http://www.osu.cz/premyslovci/Borivoj', 'http://www.osu.cz/premyslovci/fatherOf','http://www.os u.cz/premyslovci/Spytihnev')); -- Borivoj is the father of Vratislav I. INSERT INTO PREMYSLOVCI_RDF_DATA VALUES (1, SDO_RDF_TRIPLE_S('PREMYSLOVCI', 'http://www.osu.cz/premyslovci/Borivoj', 'http://www.osu.cz/premyslovci/fatherOf','http://www.os u.cz/premyslovci/Vratislav_I')); -- Ludmila is the mother of Spytihnev. INSERT INTO PREMYSLOVCI_RDF_DATA VALUES (3,
RDF Data Models in Oracle
63
SDO_RDF_TRIPLE_S('PREMYSLOVCI', 'http://www.osu.cz/premyslovci/Ludmila', 'http://www.osu.cz/premyslovci/motherOf','http://www.os u.cz/premyslovci/Spytihnev')); 5.6 Creating the Rule Base RDFS over RDF Data Begin SDO_RDF_INFERENCE.CREATE_RULES_INDEX('rdfs_rix_PREMYSLO VCI',SDO_RDF_Models('PREMYSLOVCI'), SDO_RDF_Rulebases('RDFS')); End;
6 Data Integration as an Example of Semantic Data Model Practical Usage An example of semantic data model practical usage can be found in the area of company applications integration. On one hand, companies dispose of sets of data sources from various applications, on the other hand of sets of services using those sources. The individual applications mostly do not stem from a common data model, which does not make it possible to reveal semantic differences or similarities (e.g. “Jan Novák” and “Novák Jan” is not considered to be the same person). Semantic data integration is a part of the concept of company applications integration which focuses on data exchange among applications with respect to its meaning. It is based on a common semantic data model which describes the structure and meaning of all data used by the company. It is possible to create a common data model for a whole range of activities, i.e. not to limit to one company or organization. To illustrate, there are two concrete examples of semantic data model practical usage: • telecom operators repair land and cellular telephone networks and they provide internet service every day. Each of the activities has been developed in a separate system (above a different data model). A common semantic data model would enable to integrate the data from the existing systems. • trading companies have various types of customers. Their data is stored in various systems. However, the companies want to provide their services via a single portal.
7 Conclusion The advantage of a semantic data model and its concrete form of an RDF/RDFS model in Oracle is dealt with in this paper, primarily “tidying-up” in the conceptual domain, i.e. unambiguous assigning of the meaning to the data which is used by information systems. Data integration from various sources based on a common semantic data model and aimed at its usage to provide services on a single portal is a building element in semantic web creation. Using deduction in semantic data models brings another
64
Z. Telnarova
doubtless advantage. It derives new statements (derived concepts) on the bases of statements explicitly stored in databases and a rule database which also makes a part of a database. Such a concept does not explicitly have to be stored in a database and it is independent of application as well – it is not derived in an application, but it is a part of the database. Knowledge represented by a semantic data model and by a database of inference rules then becomes independent of applications and it is usable by various services (applications) as a common background. We often speak of socalled knowledge independence.
References 1. Berners-Lee,T.: Uniform Resource Identifier (URI), Network Working Group (2005), http://tools.ietf.org/html/rfc3986 [on-line] (2009) 2. Berners-Lee, T., Hendler, J., Lassila, O.: The Semantic Web. Scientific American (May 2001), http://www.scientificamerican.com (2009) 3. Biskup, J.: Achievements Of Relational Database Schema Design Tudory Revisited. In: Thalheim, B. (ed.) Semantics in Databases 1995. LNCS, vol. 1358, Springer, Heidelberg (1998) 4. Chiang, R., Barron, T., Storey, V.: Reverse engineering of relational databases: extraction of an EER model from a relational database. Journal of Data and Knowledge Engineering 12(2), 107–142 (1994) 5. Krötzsch, M., Vradečic, D., Völkel, M., Haller, H., Studer, R.: Semantic Wikipedia. Journal of Web Semantics 5, 251–261 (2007) 6. Kalfoglou, Y., Schorlemmer, W.M.: IF-map: An ontology-mapping method based on information-flow theory. In: Spaccapietra, S., March, S., Aberer, K. (eds.) Journal on Data Semantics I. LNCS, vol. 2800, pp. 98–127. Springer, Heidelberg (2003) 7. MatulÍk, P., Pitner, T.: Sémantický web a jeho technologie (2009), http://www.ics.muni.cz/zpravodaj/articles/296.html 8. ORACLE® Database Semantic Technologies Developer’s Guide 11g Release1 (2008), http://www.oracle.com/pls/db111/portal.all_books 9. ORACLE® Spatial Topology and Network Data Models 10g Release 2 (2008), http://download-uk.oracle.com/docs/cd/B19306_01/appdev.102/ b14256/toc.htm 10. Polez, X.: Developing Semantic Web Applications using Oracle Database RDF Data Model (2009), http://www.oracle.com/technology/tech/semantic_technologies/ pdf/oow2006_semantics_061128.pdf 11. Reiter, R.: A logic for default reasoning. Artificial Intelligence 13, 81–132 (1980)
Order Processing in Supply Chain Management with Developing an Information System Model: An Automotive Manufacturing Case Study Mohammad Reza Khoei, Misam Kashefi, Pezhman Ghadimi, Amir Hossein Azadnia, Mat Rebi Abdul Rani, and Morteza Lalmazloumian Department of Manufacturing and Industrial Engineering, Faculty of Mechanical Engineering, Universiti Teknologi Malaysia 81310 Skudai, Johor, Malaysia {rkmohammad4,kbmisam2,gpezhman2}@live.utm.my, {azadnia.ie,mortezamazloumian}@gmail.com,
[email protected]
Abstract. Nowadays, high competitive market needs fast, effective, high responsiveness, online interactive, 24 hours 7 days availability and easy to follow up order processing. Consequently, there is a need for a model in which interdisciplinary approaches for understanding the range of Supply Chain Management (SCM) Information System (IS) capabilities are provided. In this study, an integrated model of SCM IS was developed that is supported by empirical evidence specific to SCM IS implementations. The developed model integrates and enriches theories of competitive strategy, supply chain management and inter-organizational information systems. Then, a case study of an automotive manufacturing industry was conducted to demonstrate the proficiency of the proposed model. As a result, better understanding of capabilities of implemented supply chain management information systems and expected future capabilities could be identified by practitioners and decision makers. Finally, findings of this study are listed together with some future works. Keywords: Supply Chain Management, Information System modeling, SCM IS, Order Processing.
1 Introduction Since early 1990s, information system has been applied in order to coordinate the flow of materials together with information inside the company; this process was named supply chain management [1]. During these years, supply chains have been reengineered using information system such as electronic data interchange (EDI) and point-of-sale (POS) systems [2]. Intensified competition in worldwide markets resulted in an increasing demand of supply chain management (SCM) and information systems (IS) activities within companies [3]. Basically, information integration plays an important role in managing a supply chain process between its stages. Consequently, organizing and sharing supply chain information resources can lead to facilitate the progress of this integration [4]. So, Enhanced supply chain management highly depends on the use information system ideas followed by reconstructing business practices and strategic plans within the whole organization A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 65–76, 2011. © Springer-Verlag Berlin Heidelberg 2011
66
M.R. Khoei et al.
[5]. Subsequently, information system is applied by companies in order to develop the supply chain process [6]. Electronic commerce as a very fast rate developing issue has been enabled by progressing in information system [7]. Enhancing the information sharing level between the supply chain stages is as important as controlling order fulfillment process in a flawless manner [8]. Supply chain management often requires the integration of different types of flows within and outside organizational relationships and management of inside the whole supply chain organization. Moreover, improved supply chain performance is related to Intercompany integration and coordination via information system [1]. In many organizations, it is becoming essential to eliminate or ignore sources of turbulence and volatility. Information and, significantly, agile information systems have been recognized as being an important issue to approach agility in the supply chain [9]. Supply Chain Management Information System acting as increasingly vital basis in the ability of companies to decrease costs and raise the responsiveness of their supply chain. However, there is a very few research articles [10,11] that show the influence of information system in supply chain management. Fundamentally, information system is like a backbone system for supply chain management. In this research, an integrated model of SCM IS capabilities was developed which is supported by empirical evidence specific to SCM IS implementations. After that, a SCM IS prototype was designed. The advantages of developed model are to integrate and enrich theories of competitive strategy, supply chain management, and interorganizational information systems. The developed model can fill the existing gap of many manufacturing and non manufacturing companies regarding the complex information flow among supply chain members.
2 Literature Review The success of many companies are involved with the supply chain management information systems [12,13] but received inadequate consideration in experiential information system study [14]. The profits and abilities of various kinds of supply chain management information systems such as EDI [15,16], electronic market [17,18] or extended enterprise resource scheduling [19] systems have been investigated by few researches. Also, there are few experimentally extracted models appropriate for examining the scope of supply chain management information systems options [20]. An organizational skill is the aptitude of an organization to attain its objectives by leveraging its different resources [21]. Supply chain management information systems abilities are organizational abilities [22] to form the desired competitive policies [23]. Stephen Hays Russel (n.d) lists four types of information systems in sustaining supply chain operations: Enterprise resource planning (ERP) software, Electronic data interchange (EDI), Electronic product code technologies and Supply chain analytics [24]. Supply chain management and information systems are firmly attached. Consequently, execution of supply chain strategies will be extremely complicated without supporting of information systems [25]. Supply chain management information systems are the information systems between companies that employ information and communication technology to arrange information inside and among the participator of a supply chain like the clients, sellers, providers, and distributors included in the utilization and supply of a special product or service [16].
Order Processing in SCM with Developing an IS Model
67
Companies have identified the advantages of using supply chain management information systems to synchronies information between the customers and producers of a supply chain since the Electronic Data Interchange (EDI) discovered [16].
Fig. 1. Structure of the supply chain management information system [32]
As shown in Figure 1, the central components of the structure are business processes and the decision-making system [25]. Factors and applications of the supply chain information system can be requested while implementation of the business processes. Information flows between factors and applications are as well determined by method of an exchange of messages [26]. The production information exchange categories are: product definition information, production capability information, production schedule information and production performance information, this includes feedback information needed to respond to the business system request to make product [27]. Also, there are seven major types of organizational units included in the supply chain: a supply chain headquarters, parts suppliers, warehouse, retailers, distributor, a final assembly plant, and a transportation network [28]. It is long time that researchers have recommended and discussed models for investigate the appropriateness of competitive policies with different high stage information systems concepts [29]. In this paper, a system was developed which shows the vital role of information system in supply chain management. It is illustrated that how well SCM IS facilitate different organizational potential in a company. Moreover, it is shown that decreasing the complexity of estimating different supply chain management information system is dependent on SCM IS. The rest of this paper is arranged as follows. In section 3, the methodology of this research is shown. This is followed by section 4 which explained the case study and results. Finally, conclusion is presented.
3 Methodology The proposed methodology, illustrated in Figure 2, used in this paper can be used as a road map to improve the data flow between members in a supply chain system. For
68
M.R. Khoei et al.
this purpose, a database was developed in which after a certain period of revising and improving the new system a decrease in terms of costs and data flow time could be expected to occur. Microsoft Access and Ms Project software were utilized to design an efficient information system for SCMIS which shows the flow of information. The proposed method consists of three main phases as follows: - Phase 1: Data collection - Phase 2: Developing the database - Phase 3: Implementation of the database
Fig. 2. Proposed methodology
4 Case Study and Results In this section, the case study along with the obtained results is presented. This task is done by integrating the description of different phases and steps of proposed method
Order Processing in SCM with Developing an IS Model
69
together with conducting the case study for better demonstration purposes for the readers. Vehicle Axle Manufacturing Company (VAM) was selected to be as case study in order to evaluate the proficiency of the method. This company was established in 1986 and it is located in Nazar Abad, an industrial city, in Iran. In this research, the cover of rear axle for light vehicles was selected as the case product. Phase 1. Data collection. The first step involves in designing a model for information system in supply chain in order to define a real supply chain with its parameters. Then, an input and the output measurement were carried out so that the perceptual models were shaped. After that, the quantitative step starts; this step consists of working on technical problems such as progression and evaluation of mathematical models, simulation models and control theory techniques. Figure 3 shows the flow of information in the factory before developing the database.
Fig. 3. Current flow of information in the factory
In order to run and model SCMIS, three different levels are involved. These levels consist of tactical, operational and strategic. The focus of this research is on strategic modeling of a supply chain management information system. Moreover, Strategic decisions are long term decisions are interrelated to the company’s strategy. Furthermore, they are usually engaged with most of the supply chain members. As a matter of fact, one of the basic necessities for supply chain implementation is visibility of appropriate information in the exact time and in the exact arrangement. In inventory control strategy, these elements might be used at the level of inventory safety, reorder point, stock level of finished goods, raw material and middle parts and inventory location. When it comes to purchase and logistics, significant factors are supplier lead time, supply lot size, supplier capacity, and purchase time. Significant order information consists of parameters such as due date, preference, start and end information and order pattern. Strategic information constitutes order control policies and dispatch policies. In the design modeling step of the supply chain, the first mission is to find the most significant variables like inputs and outputs and the effect of these variables on the supply chain system. To develop the database, these data are collected: 1.
Production data: Production rate, part number, product number, product price, product type, part type.
70
M.R. Khoei et al.
2. 3. 4.
Customer data: Customer order, due date, order date, order number. Supplier data: Part description, supplier list, part price. Information flow: Information within supplier, manufacture and customer.
Phase 2. Developing the database. This database was created with Microsoft Access and it is utilized to improve the flow of information between suppliers, manufacturer and customers. This database should be installed in a server inside the organization with regard to the point that the only way to access the database is by operators of the organization. Customers can only enter their data by internet. Moreover, customers can order online at any time they want via using internet. After that, the order data will automatically goes to the database and after less than 8 hours their order will be processed in the organization. The users of the database are the customers, manufacturing company and the suppliers of the selected product. The users of this database need to have Microsoft Access 2007, Microsoft Outlook 2007 and Microsoft Info path 2007 in their computers. The database is designed to help the customers dealing with their order requests step by step. By following these steps, customers are able to request their order at any time that they want. i) Design the Entity relationship diagram (IDEF1X). IDEF1X is a data modeling language for the developing which relates data models [30]. It is used to produce graphical information model which represents the structure and relation of information within an environment or system. IDEF1X of the company is shown in Figure 4.
Fig. 4. IDEF1X diagram of the company
Order Processing in SCM with Developing an IS Model
71
ii) Design the database relationship. Figure 5 shows the relationship between supply chain members.
Fig. 5. Data relationship Microsoft Access
iii) Developing the database and flow of information in the factory. In this figure, developed flow of information in the factory can be seen. In the developed system, the factory use EDI for the flow of information between supply chain members so moving the data between supply chain members became fast which resulted in decreased order time.
Fig. 6. Developed flow of information in the factory
vi) SCMIS Database. In this database, customer can key in his order information online and all the entered order information goes direct to the factory server. In the
72
M.R. Khoei et al.
next step, at the end of the day factory operator should key in the amount of daily production and rejection, product released to the costumer, raw material received from supplier to the factory, received raw material for each type of raw material separately and the raw material for additional order. The last part is to send the report to the customers and make them aware of the estimated release date for their order. Also, factory operator can send the report of raw material order to each supplier by clicking on the supplier order report. Consequently, suppliers would be aware of the detail of raw material order. Also estimated delivery time for costumer’s order would be known. Phase 3. Implementation of the Database. Figure 7 shows the main page (Home Page) of the database, daily production plan is shown in the figure.
Fig. 7. Home page of the database
In Figure 8, the daily amount of order that should be released to a specific customer is shown. Also, total amount of product that the factory owes to specific customer is found.
Fig. 8. Product order by customer page in database
Order Processing in SCM with Developing an IS Model
73
In Figure 9, the amount of raw material which should be received as a daily basis can be observed. This numbers are automatically calculated and shown on this box.
Fig. 9. Raw material order and information page in database
The three important elements in calculating the amount of additional order are shown in Figure 10. These three elements are daily raw material rejection, daily raw material inventory and daily raw material ROP.
Fig. 10. ROP and inventory information page in database
Comparison between old system and developed one. As it shown in Table 1, the end result of this study is that the company has decreased the cost of production by using the database within one month. The comparison between the old system and the newly developed system can be seen in table below in order to illustrate the proficiency of the proposed model. All presented information for old system is taken from the historical records of the case company. Also, information regarding the new system is obtained from the outputs of the proposed SCM IS model.
74
M.R. Khoei et al. Table 1. Comparison between old system and new system Element Product in inventory
Old System 400
New System 44
Unit Number
Respond Time
56
8
Hour
Transport Cost
55000000
10000000
IRR
Production Lain Stop
1.5
Almost Zero
Hours
Cost Of Production Lain Stop
5000000
Almost Zero
IRR
This study has five major findings which are discussed as follows: i)
ii)
iii)
iv)
v)
Ordering cycle time is shortened: The order cycle time in the newly developed system decreased in comparison to the old system. This is because of that the time spent on paper work in the newly developed system is removed. Supply members connect with each other via the internet. Product inventory decreased: Prior to the implementation of the database, the factory in this study keep their stock for seven days to avoid out-of-stock problems in inventory. This led to high cost for inventory and maintenance. The implementation of database enables the factory to reduce the number of days from seven to one, which results in reduction of cost for inventory and maintenance. Transportation cost decreased: The case factory used to spend considerably huge amount of money on transportation. They had to send their customer’s order in different parts because they couldn’t meet the release time of the order. So, they used to send the order in different points of time until the order was completely sent out. The newly developed system allows the production of total order to be finished at specific point of time so that when the factory sends the product to the customer, their cost on transportation and constant delivery of product can be saved. The amount of production line stop decrease to almost zero: In the newly developed system, factory receives the required raw material from suppliers on time and in the right amount, but in the old system due to lack of information flow between suppliers and the factory, usually there was a lacking of raw material in the production line which that makes the production line stop. The cost of production line stop become almost zero: In the newly developed system when there is no production line being stopped; as a result, there is no extra cost for production line stop.
5 Conclusion The goal of this study was to develop an integrated model of SCMIS capabilities that is supported by empirical evidence specific to SCMIS implementations. The model developed integrates and enriches theories of competitive strategy, supply chain management, and inter-organizational information systems. According to a forementioned
Order Processing in SCM with Developing an IS Model
75
literature, supply chain management information systems are progressively important to the prosperity of many companies. The advantage of SCMIS among others is that it synchronizes information among the customers, producers, distributors, and other members in a supply chain. Despite the fact that SCMIS has many advantages to offer, there are only a few experimentally extracted appropriate models that examine the scope of supply chain management information systems options. In addition, if company wants to develop its business, manager also can control all of branches using network system. In conclusion, implementing Information system in supply chain management can help to manage system efficiently and effectively because it can provide information and communication properly. The contributions of this research activity are as follows: i. SCM IS model in order to facilitate the information flow within the members of supply chain. ii. Reduced amount of ordering cycle time. iii. Reduced logistic costs within the whole supply chain stages. In this particular case, computer system only covers activities within manufacturing system. Networking only covers relationships between main office, each branch, customer and supplier. Future researcher can use PHP web programming language and visual basic to facilitate the flow of information in the whole supply chain area.
References 1. Barut, M., Faisst, W., Kanet, J.: Measuring supply chain coupling: an information system perspective. European Journal of Purchasing & Supply Management 8, 161–171 (2002) 2. Raghunathan, S.: Impact of demand correlation on the value of and incentives for information sharing in a supply chain. European Journal of Operational Research 146, 634–649 (2003) 3. Bayraktar, E., Koh, S.C.L., Tatoglu, E., Zaim, H.: Analysis of the impact of information systems and supply chain management practices on operational performance: Evidence from manufacturing SMEs in Turkey. Int. J. Production Economics 122(1), 133–149 (2009) 4. Chandra, C., Tumanyan, A.: Organization and problem ontology for supply chain information support system. Data & Knowledge Engineering 61, 263–280 (2007) 5. Humphreys, P.K., Lai, M.K., Sculli, D.: An inter-organizational information system for supply chain management. Int. J. Production Economics 70(3), 245–255 (2001) 6. Wu, F., Yeniurt, F., Kim, S., Cavasugil, D.S.T.: The impact of information technology on supply chain capabilities and firm performance: A resource-based view. Industrial Marketing Management 35, 493–504 (2006) 7. Ovalle, R.O., Crespo, M.A.: The effectiveness of using e-collaboration tools in the supply chain: an assessment study with system dynamics. Journal of Purchasing & Supply Management 9, 151–163 (2003) 8. Martınez-Olvera, C.: Entropy as an assessment tool of supply chain information sharing. European Journal of Operational Research 185, 405–417 (2008) 9. White, A., Daniel, E.M., Mohdzain, M.: The role of emergent information technologies and systems in enabling supply chain agility. International Journal of Information Management 25, 396–410 (2005) 10. Guo, Z., Fang, F., Whinston, A.B.: Supply chain information sharing in a macro prediction market. Decision Support Systems 42(3), 1944–1958 (2006)
76
M.R. Khoei et al.
11. McLaren, T.S., Head, M.M., Yufei, Y.: Supply chain management information systems capabilities. An exploratory study of electronics manufacturers. Information Systems and e-Business Management 2, 207–222 (2004) 12. Chopra, S., Meindl, P.: Supply Chain Management: Strategy, Planning, and Operation. Prentice Hall, Inc., Upper Saddle River (2001) 13. Kumar, K.: Technologies for Supporting Supply Chain Management. Communications of the ACM 44(6), 58–61 (2001) 14. Subramani, M.: How do suppliers benefit from information technology use in supply chain relationships? MIS Quarterly 28(1), 45–73 (2004) 15. Malone, W.T., Crowston, K., Lee, J., Pentland, B., Dellarocas, C., Wyner, G., Quimby, J., Osborn, S.C., Bernstein, A., Herman, G., Klein, M., O’Donnell, E.: Tools for inventing organizations: Towards a handbook of organizational processes. Management Science 45(3), 425–443 (1999) 16. Mukhopadhyay, W., Kekre, S., Kalathur, S.: Business value of information technology: A study of electronic data interchanges. MIS Quart. 19(2), 137–156 (1995) 17. Dagenais, T., Gautschi, D.: Net Markets Driving Success in the B2B Networked Economy. McGraw-Hill Ryerson Ltd., Toronto (2002) 18. Kaplan, S., Sawhney, M.: E-Hubs: The New B2B Marketplaces. Harvard Business Review, 97–103 (2000) 19. Green, F.B.: Managing the Unmanageable: Integrating the Supply Chain with New Developments in Software. Supply Chain Management: An International Journal 6(5), 208–211 (2001) 20. Reddy, R., Reddy, S.: Supply Chains to Virtual Integration. McGraw-Hill, New York (2001) 21. Ulrich, D., Lake, D.: Organizational Capability: Competing from the Inside Out. John Wiley & Son’s, Inc., New York (1990) 22. Allen, B.R., Boynton, A.C.: Information Architecture: In Search of Efficient Flexibility. MIS Quarterly, 435–445 (1991) 23. Henderson, J.C., Venkatraman, N., Oldach, S.: Aligning Business and IT Strategies. In: Luftman, J.N. (ed.) Competing in the Information Age: Strategic Alignment in Practice, pp. 21–41. Oxford University Press, New York (1991) 24. Chauncey, B.B.: Transportation of Troops and Material, Kansas City: Hudson Publishing, 125. For an expanded discussion on the origin of the term logistics and for various definitions, see Stephen Hays Russell. The Growing World of Logistics, Air Force Journal of Logistics XXIV(4), 15–19 (1905) 25. Chandra, C., Grabis, J.: Supply Chain Configuration – Concepts, Solutions and Applications. Springer Science, Business Media, New York (2007) 26. Curbera, F., Goland, Y., Klein, J., Leymann, F., Roller, D., Thatte, S., Weerawarana, S.: Business Process Execution Language for Web Services, Version 1.0.Specification, BEA Systems, IBM Corp., Microsoft Corp. (2002) 27. Gerhard, G., Ranjan, G.: Practical E-Manufacturing and Supply Chain Management. Newnes, United Kingdom (2004) 28. Lee, T., McLean, C., Umeda, S.: A preliminary Information model for a supply chain simulation. National Institute of Standards and Technology, Gaithersburg, MD 20899-8260 USA (2008) 29. Ciborra, C.U.: A critical review of the literature. In: Ciborra, C.U. (ed.) From Control to Drift: The Dynamics of Corporate Information Infrastructures, pp. 20–33. Oxford University Press, Oxford (2000) 30. Li, Q., Chen, Y-L.: Modeling and Analysis of Enterprise and Information Systems, Part Three, pp. 140–155 (2009)
Spatial Information Databases Integration Model Mustafa Man1, Mohd. Shafry Mohd Rahim2, Mohammad Zaidi Zakaria3, and Wan Aezwani Wan Abu Bakar4 1
Department of Computer Science, Faculty Science & Technology, Universiti Malaysia Terengganu (UMT), 21030 Kuala Terengganu,Terengganu, Malaysia
[email protected] 2 Department of Computer Graphics & Multimedia, Faculty of Computer Science & Information Systems Universiti Teknologi Malaysia (UTM), 81310 UTM Skudai, Johor Bahru, Malaysia
[email protected] 3 Department of Fisheries Science, Faculty of Agrotechnology & Food Science, Universiti Malaysia Terengganu (UMT), 21030 Kuala Terengganu,Terengganu, Malaysia
[email protected] 4 Pre Sea Marine Engineering Department, Training & Education, Malaysian Maritime Academy (ALAM), 21030 Kuala Terengganu,Terengganu, Malaysia
[email protected]
Abstract. The integration process of the various information in various database types require a thorough understanding to carry out data extraction process. The data extraction process here is in terms of the scheme and the structure for scattered and distributed locations. The Spatial Information Databases Integration Model (SIDIM) is a model which covered the integration processes such as pre-integration, scheme comparison, algorithm development process and intermediary software (middleware) and post-integration. Emphasis are administered in algorithm development by using hybrid approach based on CLARANS approach's, combination with abstract visualization and Catch Per Unit Effort (CPUE) to enable to achieve processed data or information that is located in different database type in order to obtain data in quick, trusted and reliable manner. SIDIM will become a new engine to process information in various database types without changing any of existing organization system. To verify the model credibility, case studies which are related to fishing industry in Malaysia and artificial reef project are being made as a foundation for SIDIM efficiency testing. Keywords: SIDIM, Integration, Database, CLARANS Model, Abstract Visualization, Catch per unit effort (CPUE).
1 Introduction Information, communication and technology (ICT) era is much put into major influence in human life. Every day, human are now flooded with various digital data types through web sites, message acceptance through telecommunications equipment such as mobile and other printed media. Delivery process, access, storage and digital A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 77–90, 2011. © Springer-Verlag Berlin Heidelberg 2011
78
M. Man et al.
data mining perpetrated in rhythm that quickly make world alleged borderless. The increasing of variability and voluminous of digital data can create difficulties in analyzing and obtaining a useful information. In fishery database system, presently there are huge numbers of digital data that are available and separated elsewhere and these data are increasing in trend from time to time. Data that are analyzed through merging and integration can give rise to useful information especially for future planning and management purposes. The problem is how these various kinds of digital databases could be merged, integrated and automatically visualized in the form of individual’s desired information. It is a matter of how to group or unify all these different sources of database at various locations into one place. Recently, many research on information-seeking were more diverted on merging and unifying the spatial and non-spatial databases. The method of information searching is very much crucial among consumer. The consumer may requires information on some location and the updates on the change of that location are the major concern besides continuing to make a comparison through a visualization method which is based on map. The combination of both spatial and non spatial data is indeed important and crucial [1]. There are several challenges that need to be faced in the process of merging and integrating the database from different sources and locations. It is required that these data, i.e., data from different sources to be in priority and can be selected prior to continuing the integration process. The objectives of our research are to; 1. 2.
3.
Produce a model that can enable the process of integrating spatial and nonspatial data from different types of database format How the model that proposed is affordable to carry out an integration process among various information on spatial and non spatial data on line which differ in its database structure and environment? Proposed a model which can improve the integration processing time?
Spatial Information Database Integration Model (SIDIM) is one of our prototypes which can integrate spatial and non-spatial data from various databases format and location. In principle, our model in this study is based on location coordinate technique in order to merge the predicted results based on location. The integration approach from various databases were used to give an impact analysis of information especially among spatial and non spatial data. The algorithm is developed using location integration technique in order to fasten the extraction process of spatial and non spatial data.
2 Related Work Research on data integration is one of the most important element and is a ‘hot issues’ in the research of spatial data. Previous researches were more concentrated on developing a common form of information that is easier for the user to manage the database without consuming longer time. Integration is defined as a merger of various information from various sources that can provide benefits from the aspects of collection of the best information processing time, resource conservation and data-sharing to various purposes.
Spatial Information Databases Integration Model
79
Information which will be combined or integrated must be staying in a state of analogous (uniform), but if the state of the information which will be consolidated is unequal, thus a form or modification process has to be executed on that particular information structure without damaging the original information. To place the information in various databases and at different or scattered location may contribute to difficulty in performing integration process. Thus a research to integrate ‘every single piece into one’ is to be initiated as a kickoff start. According to [2], there are still no overall and concrete solutions to data integration over almost three decades. In his work, he has used the following methods for integrating the databases; 1.
2. 3. 4.
5.
6.
Manual Integration – resolving the interface of different and various demand language, consumer need to know in detail the location, data presentation in logical and semantic order. Same interface – consumer were prepared with uniform interface to facilitate information-seeking which requires the use of one navigator. Integration through application – application which enables information source to give a uniform result to the consumer through more workable tools. Integration through middleware (intermediary software) – intermediary software use back function which generally can solve the problem of integrating data such as intermediary software users SQL. Different intermediary software is used to combine so that all the software will become aligned. Standard Data Access – logical data integration were achieved globally although staying in physical located at different location and this was achieved through virtual. Reserves the same storage utilization – Integrating the physical data were done through reserving the same storage partnership. Local storage is able to process various information that had been integrated and can process much faster.
The history of integration method started in early 80’s which is known as MULTIDATABASE, followed by a mediator called “GARLIC”. Later on, INFOSLEUTH is introduced and were updated into a method of integration based on ontology which is known as OBSORVER. The integration method were then developed using peer to peer known as HYPERION. The journey of integration method end nowadays using web based integration method which is known as ACTIVE XML [3]. Interoperability is the key solution to solve the issue in integration.. It allows heterogeneous data to communicate or interact from one application to another application by replying with the intended information. Interoperability is a complex feature because the data source obtained is not in database format, i.e., in email form or file attached only and external data source obtained through internet access in which the data is either unstructured or semi-structured which needs more understanding in its semantic. Technology and management system in database are always subject to change. Online access has becoming very popular among users because it is faster and quicker. And today, data models for spatial and temporal data are most in need.
80
M. Man et al.
Spatial semantic can display about any information which can be presented in various forms, such as, solid representation: space which is divided into small pieces. Spatial feature for each entity is presented based on small scale set for each entity or boundary representation (wire frame models): Spatial feature is presented based on line segment or boundary segment [4].
3 Research Methodology SIDIM is an integrated spatial and non spatial database designed to enable access and efficiency evaluation of artificial reefs based on the growth of phytoplankton and zooplankton for each dedicated artificial reefs area. SIDIM is considered as a method, model or new idea to ensure the effectiveness and efficiency of artifcial reefs development project as a place for fish population study at marine park. Since the high cost incurred to evaluate the efficiency of artificial reef by hiring a special scuba task force, the development of SIDIM can help to solve the problem of analysing and assessing the dynamic fish population. SIDIM enables the integration of more than one database either from same or different format. Research methodology consists of systematic process which is illustrated in Fig. 1; Start
Problem Definition
Related Work
Development of an Integration Model
SIDIM Developments of integration algorithm for multi-type of databases in centralize and distributed server.
No
Testing
Yes Validation and Verification
End
Fig. 1. The flow chart for Integration Process using Multi Database Types
Spatial Information Databases Integration Model
81
Fig. 2 shows four steps that need to be taken prior to integration, that is i. Pre-integration: This is a process in order to assess the types of database environment similar to ORACLE, such as, MYSQL, or MS ACCESS and from other databases. ii. Scheme Comparison: This is a process whereby the Scheme or structure for each databases is compared in order to facilitate integration process. In the early stage, this process is made manually. iii. Intermediary software development (middleware): This is an application parts that were developed for making the process of integration of database workable. iv. Post-Integration: This is the process of Integration assessment which is conducted using credibility process and “interoperability” aspect. The focus of this paper is to describe the algorithm development which enables multi database types in a distributed environment to be integrated into another database through online. This is done by using a three-tier architecture. Fig. 3 depicted an overall SIDIM architecture in a distributed environment. Pre-integration
Schema comparison
Development of algorithm
Integration: Development of a Middleware
Post-Integration
Fig. 2. The Flow Chart for Information Integration Process for SIDIM Model
Fig. 3. Three-tier Architecture for the development of SIDIM Model
82
M. Man et al.
A. Pre-integration Processes 1st step As mentioned previously, the process of pre-integration requires a study to identify the present database environment. For example, we have developed two application format, such as, Artificial Reef Positioning System (ARPOS), funded under eSciencesFund (MOSTI) research grant in 2006 [5] which used MYSQL format and (WiFISH) database in MSSQL format. The system development uses Hypertext Preprocessor (PHP) software which is accessible through wired internet and wireless internet [6]. Both of the databases are owned by different agencies namely, i.e., Department of Fisheries (DOF) and Lembaga Kemajuan Ikan Malaysia (LKIM), respectively and were located at different location. B. Scheme Comparison 2nd step The results from scheme comparison show that these two databases can be integrated because they are from the same location, i.e., the catch from the WiFISH databases are from the same location (position/ground) of artificial reef. So therefore, the location based technique is a core and as a fundamental to determine the effectiveness of the artificial reef project development at a certain location within the timeline. Furthermore, at similar location, the assessment of catch per unit effort (CPUE) can be evaluated by integrating one more database, the fishing effort database. The fishing effort can be defined as number of fishing vessels, number of day fishing or trips, number of crews etc. However, these fishing efforts are basically depend on types of fishing gears and fishing power. The searching time to the fishing ground also make a sense to effect the CPUE of a certain boat. A lot of database needs to be integrated in order to obtain a reliable CPUE estimation or abundance of fish at certain area. In order to integrate multiple databases, we developed an algorithm based on the assumption below; Catch location (CL) is equivalent with artificial reefs development (AR). Assessment factor is dependent on catch yield number (CT) for each type of fish (FT) and comparison with artificial reef type (RT) which included. A formula can be set up here. CL ≡ AR where if found CT high ≡ FT is effective which is depending on catch date (DT) made. Vice versa if CT low, then FT to be ineffective. Integration theory and model are made based on set and theorem proving to perform scheme comparison among distributed multi database types [7]. The formal specification through Z-eves scheme was used to verify the specification of a developed SIDIM model. Fig. 4 shows a Z scheme development and verification based on theorem proving. Based on the proven scheme, another new algorithm which is called location integration technique was formulated to evaluate the efficiency level of the artificial reefs development project. All of integration processes and scheme comparisons are done by the developed middleware.
Spatial Information Databases Integration Model
Fig. 4. Scheme Proving Using Z-eves Tools
83
84
M. Man et al.
Since these data is located and taken from different database and in different format, then the development of algorithm for the application software (middleware) was a pre-requisite. The algorithm is developed based on CLARANS algorithm and identified Abstract Representation are distributed and gather in multi database types environment through integration process. The overall process is as depicted in Fig. 5 as follows : Step 1: CLARANS Approach Classification or clustering all the spatial data and non-spatial data for Artificial Reef Distribution Database, Fish Landing Databases and Vessel Profile database.
Step 2: Abstract Representation Visualization Approach Combining all the data on spatial and non-spatial into one new database followed by visualizing through online mapping format (GIS Format).
Step 3: Query Processing Information for all the databases are done by using SQL query and visualizes according to the user specification through online map (GIS format).
Fig. 5. Algorithm Development Process to integrate spatial and non-spatial data in distributed environment for heterogeneous SIDIM Model
The CLARANS algorithm were found suitable and fit to be used in order to evaluate the integration process of spatial and non-spatial data from multi database types [8]. In the second step, that is, when the clustering process has successfully done, then the next process is abstract representation approach for the integration estimation. This estimation is conducted to determine the efficiency of artificial reefs development in fulfilling the number of fish catches obtained at the predetermined cluster. The relationship known as Spatial Semantic, e.g. ABOVE, NEAR, IS-NEXTTO, and BEHIND are used to map and display every entity in various database types. The third step, the process of displaying the output of integrated databases in a visual form is conducted based on map. This method will enable the searching of information concerning the position of artificial reef which is easily assesible using Geographical Information System (GIS), and then compare and super imposed with the fish catch obtained at artificial reef to visualized the results. Under this activity, evaluation will be made to test for efficiency level and reliability of developed middleware in integrating process. The evaluation is carried out at the system user level. The process time is recorded to obtain and justify the information processing speed and interoperability is fulfilled [9].
Spatial Information Databases Integration Model
85
Other than processing time factor, evaluation of catch per unit effort or index of abundance (CPUE) was encountered in order to assess the whole project’s efficiency for each dedicated artificial reef location. CPUE is defined as the level and efficiency degree between the catch in ton (t) and the ability of catch (frequency of fish catch) is being conducted. This method is based on variance and average CPUE [10]. CPUE with non zero catch is modeled in the following formula: ln
, , ,
, , ,
(1)
where µ is the variable, αi is the year factor, βj is the month factor, is the zone factor, tl is the rate factor, and , , , is the random error component. As a means to integrate travelling trip without catch, the writers have modeled the success probability of fish catch where it relates with the efficiency degree between the travelling trip with non-zero catch and the total of jack mackerel fish catch trips. The model followed binomial distribution and ‘Logic’ function which used to relate P with its estimator. (2) is the zone where µ is the variable, αi is the year factor, βj is the month factor, factor, tl is the rate factor, and , , , is the random error component. Then the yearly value of CPUE is calculated in tone / total travelling trip that relates to statement below : (3) Besides, if ‘p’ is defined as efficiency evaluator, then the yearly value of CPUE will relate to the calculation as follows: (4) Assumption is made where an increase in CPUE trips with the catches can be influenced by the changes of boat’s efficiency and the catch resources which represents either the fish catch is a success or otherwise. The result of integration process can determine the fish types with the catch area. Through this, we can identify artificial reefs type for the production of phytoplankton and zoo plankton which in fact is the food source for marine life.
4 SIDIM Testing and Evaluation Prior to formal specification the SIDIM model were found to be valid. The coding is done using an open source .PHP programming language in order to develop a middleware application. Below is one of the module that have been developed for coding and integrating three multi-types of databases which was included in SIDIM model. The output for this module is as shown in Fig. 6 using XML.
86
M. Man et al.
//check if exist in SIDIM database $select_SIDIM = mysql_query("select * from SIDIM where date = '$date' and boat_no = '$boat_not' and fish_code = '$fish_code'"); $row = mysql_fetch_assoc($select_SIDIM); $id = $row['id']; $sdate = $row['date']; $sboat_no = $row['boat_no']; $sfish_code = $row['fish_code']; $shasil = $row['hasil']; $stypes = $row['types'];
Spatial Information Databases Integration Model
87
$slocation = $row['location']; $szone = $row['zone']; if(($sdate != $date) && ($sboat_no != $boat_no) && ($sfish_code != $fish_code)){ $insert = mysql_query("INSERT INTO SIDIM (date,boat_no,tangkapan,types,location,zone,fish_code) VALUE ('$mysql_date','$mysql_boat_no','$mysql_catch','$ac cess_types','$access_location','$oracle_zon','$mysql_fish _code')",$conn); //if data already exist if(($sdate == $date) && ($sboat_no == $boat_no) && ($sfish_code == $fish_code)){ echo "Data Already Exist"; }
Fig. 6. Output of databases Integration Using SIDIM
The process of evaluating and testing for SIDIM model was conducted by comparing between manual data and a SIDIM processed data. The testing had been focused on the ’interoperability’ feature in spatial and non-spatial data processing among the different types of database within a centralized server and followed by comparing with spatial and non-spatial data processing in a different types of database within a distributed server. The result obtained which is shown in Table 1 were compared and detrermined whether the two different servers (centralized or distributed) having a similar results or vice versa. This comparison will ultimately confirm whether each algorithm deployed at each particular steps (as shown in Fig. 3) has achieved the objectives or otherwise. After the result of data integration is obtained, then the next process is to visualize the result by adopting a clustering process using CLARANS approach. The Google Map software together with the abstract visualization element were generated as to show the result of an integration process. The visualized information is depicted in Fig. 7.
88
M. Man et al.
From the visualized information, it confirms that the efficiency evaluation for the fish catches in a certain artificial reef location can be achieved as well as the estimation of the fish catch obtained for the following years can be done. Fig. 8 and Fig. 9 illustrate the comparison of the fish species and the calculation had been done using CPUE model. Table 1. An interoperability testing between different types of database in a centralized environment
Database Types
Types of Web Server
MYSQL ORACLE MS Access
Apache IIS IIS
Interoperability Testing with Multi-types of Databases MYSQL √ √ √
ORACLE √ √ √
MS Access √ √ √
Fig. 7. Visualized Information as a result from integration of a different database types
Result of integration in both databases obtained could be analyzed directly by intermediary software and comparison is acquired by bar graph as shown in Fig. 8 and Fig. 9. The result reveals that the catch yield in February 2008 and 2009 has a condition of almost similar trend. Conclusion may be formed that catch yield in forwarding month onwards can be made as a measurement to ensure the growth effectiveness of the artificial reef or otherwise. Due to this, future intermediary software developed would enable the integration process of these data can be carried out efficiently and at faster rate and can be achieved through website only. Comparison may be made considering the result on average distribution of fish catches by monthly originated from January to December 2008 and 2009. Whilst year 2010 will be a determinant on how effective the methods are.
Spatial Information Databases Integration Model
89
Fig. 8. Graph of total catch of Kerisi of Fish Catches in February 2008 at Bidong Island artificial reef
Fig. 9. Graph of total catch of Kerisi of Fish Catches in February 2009 at Bidong Island Artificial reef
5 Summary and Discussion SIDIM is a model design which is reusable in various types of database integration problem. Due to this, if this method is successfully deployed and implemented, then the determination of effectiveness for artificial reef development project can be valued easily without requiring high cost and can give huge impact to fishing industry in Malaysia. According to pass experiences, many theorems have been through a long and repetitious proving process. But, if the proving is done manually by humans, the possibility of mistake made is very high. Using Z/EVES, not only this possibility can be reduced, the proving can be done fast and reliable. One of our future works shall deal with complete and precise specification and validation for multiple databases in distributed systems environment and also this model could be tested in any domain if related to spatial information databases integration industries.
90
M. Man et al.
Acknowledgement. Appreciation conveyed to Malaysian Fisheries Development Authority (LKIM), Fisheries Research Institute, Malaysia, Fisheries Department of Malaysia and Ministry of Science, Technology and Innovation Malaysia (MOSTI) for contribution to the above data and information and ARPOS project financing under eSciencesFund research grant. Last but not least, appreciation also goes to researchers in this research group from UMT, UTM and ALAM.
References [1] Parent, C., Spaccapietra, S.: Issues and approaches of database integration. CACM 41(5), 166–178 (1998) [2] Patrick, Z., Klaus, R.: Three decades of data integration-All problems solved? In: IFIP Congress Topical Sessions 2004, pp. 3–12 (2004) [3] Parent, C., Spaccapietra, S., Dupont, Y.: Model independant assertions for integration of heterogeneous schemas. Very Large Database Journal 1, 81–126 (1992) [4] Elmasri, R., Navathe, S.: Fundamental of Database System, 5th edn. Addison Wesley (2006) ISBN: 0321369572 [5] Mustafa, M., et al.: GIS spatial data visualization tools for artificial reefs distribution. In: Proceeding of The 3rd International Conference on Mathematics and Statistics (ICoMS-3), pp. 25–38 (2008) [6] Mustafa, M., Yazid, M.S.M., Aezwani, W.A.B.W., Zaidi, Z.M.: SPI: Productive Software for Fish Landing Data Collection by Using Wireless Mobile Technology. In: Proceeding 1st Regional Conference On Human Resource Development (RESERD 2008), Primula Hotel, Kuala Terengganu, January 14-15, pp. 45–53 (2008) [7] Bowen, J.: Formal Specification and Documentation using Z: A Case Study Approach. Thomson Publishing (2003) [8] Raymond, T.N., Jiawei, H.: CLARANS: A Method for clustering objects for Spatial Data Mining. IEEE Transactions on Knowledge and Data Engineering 14(5), 1003–1015 (2002) [9] Ismail, W., Joseph, K.S.: Future IT trends for GIS/Spatial Information Management. Scientific Research and Essay 5(10), 1025–1032 (2010) [10] Stephans, A., MacCall, A.: A multispecies approach to sub setting logbook data for purposes of estimating CPUE. Journal of Fisheries Research, 299–310 (2004)
A Framework of Hybrid Semantic Speech Query via Stemmer for Quran Documents Results Mohd Amin MohdYunus, Roziati Zainuddin, and Noorhidawati Abdullah Faculty of Computer Science and Information Technology, University of Malaya, 50603, Kuala Lumpur, Malaysia
[email protected], {roziati,noorhidawati}@um.edu.my
Abstract. The single query has a lack of words reference to provide more related words for documents result retrieval. The study covers those available words in the dictionary which related to the single word. Therefore it also means that single word is assigned for many words from single query as multirelevant queries in the dictionary. Hence, the single query is speech query to be converted to words in the query. Cross language information retrieval (CLIR) system investigates the Quran system results performance using looping referred semantic words and stemmers are known as stemming looping semantic query (SLSQ). The query however is based on the speech to be input and converted into text. Therefore, this study is conducted with the purpose to introduce the framework which covers speech, semantic and stemmers approach against the queries and vice versa. Furthermore, it is also conducted to investigate the performance SLSQ based on total retrieve and relevant to maximize words than single word. It also found that looping semantic words technique in dictionary contributes to better approach in using looping referred words in the dictionary for its better performance for Quran document results. Keywords: Stemming approach, looping referred words, speech query.
1 Introduction Hybrid semantic referred looping words in dictionary are not only related to semantic itself but how to link the whole related words in dictionary according to index of every line until the last line of words at the end of dictionary. The capability of query function to utilize the whole relevant words which means there is no line of related words to be missed in the link list process for searching very related documents. Hence, the most relevant documents are derived of utilizing similar meanings of the whole words in the dictionary to improve the retrieval performance in CLIR. Therefore, the dictionary plays the main role to provide more relevant words from the single word to be called as looping referred words spoken words with stemming semantic query (SLSQ) than semantic query only. Both of them are different in terms of retrieval process, collective semantic words and link to index. The simple introduction about the looping referred semantic words is retrieving more links list compared to the semantic itself which only have one link list. A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 91–101, 2011. © Springer-Verlag Berlin Heidelberg 2011
92
M.A. Yunus, R. Zainuddin, and N. Abdullah
One link can have few words to be used for searching process. Thus it can say that the looping referred semantic words is more intelligent compared to the semantic itself to provide semantic words as many as provided in the dictionary as a collective semantic words. Hence many semantic frameworks have been modified and investigated for applying the better CLIR performance retrieval results which come from the each word carries many relevant looping referred words from dictionary to the query. However, speech recognition is first process to be going through and converted to text or words [7]. In order to look for the looping referred words, the first text or words are then going to the next process to list the links which carry few unique words in the dictionary. It such one carries many, and then those words are processed again and again if similar word in the dictionary exists and find the link list of the words. It depends on the similarity meanings as looping referred semantic words dictionary [14]. Looping referred words or multi-relevant words is more effective even though it can be applied to the music and audio semantics that can be actually represented within the system retrieval structure [1]. The existing CLIR applications display generally whole documents as result to meet the query given in the process and provide relevant results as relevant judgment. Therefore, ontology plays a key role for adding a semantic dimension between the user’s query and the data sources [2],[6]. Yet it is still to be improved by using the link list process and loop to search more relevant words in dictionary to be in queue in order to achieve high retrieval performance. Some attempts for using ontology in search engines can be found in the literature [4]. One difficulty is to find relevant domain ontology, and combine them [3]. Ontology can be also exploited in result retrieval systems during query analysis [8],[9]. In conclusion, the system should be flexible in terms of information delivered, from documents to exact answers [10]. In order to apply semantic query, it can be successfully improved to employ the whole words in dictionary which are more significant to retrieve more relevant documents as a whole relevant documents result. The whole words are known as looping referred words from the single word as the link list of the words results. The repetition additional semantic words are collected in the whole dictionary database through the speech recognition and then stemmed for improving al-Qur’an information results [5],[11],[13]. In this study, looping semantic algorithm is the process of looking for matched words in order to get the rest matched words and again until the whole dictionary has been checked. It is the study of improving the information retrieval in utilizing all matched words after matching again and again in the dictionary database. In the dictionary database, the semantic or synonym words can be added or deleted and then the result of SLSQ is different depending on the words related to addition and deletion which have been made.
2 Related Work SLSQ is based on combination of automatic speech recognition, looping additional semantic words and stemmers as hybrid processes in order to investigate its performance according to the percentages of recall and precision. The framework
A Framework of Hybrid Semantic Speech Query via Stemmer for Quran Documents
93
designed is for improving al-Qur’an searching process through previous researches conducted based on separated methods such as lexical semantic approach, stemming algorithm and speech recognition. The study is referred based on the novelty of the SLSQ framework to improve relevant information retrieved based on combination or hybrid methods and processes. This work proposes the use of contextual information provided by web search engine queries for improving text recognition performance. Then it firstly describes a framework for converting information type from speech recognition into text [16]. It is therefore the main task of this framework can generate more synonym words in spite of speech recognition and stemming approach. In the synthesis stage, these phonetic targets constrain the dictionary search for the sequence of diphones that maximizes similarity to the input test data in the search field, subject to the acoustic domain [17]. This paper introduces a novel approach to the task of lexical translation between languages in order to retrieve multilingual relevant documents. This approach have been introduced by Mausam et al. (2010) which build a massive translation graph, automatically constructed from over 630 machine-readable dictionaries and Wiktionaries [18]. They use a transductive learning approach to automatically construct a dictionary, and then refine it by improving its test set coverage while reducing its over-fitting tendency. In addition, they incorporate frequency information to discriminate overlapping matching words [19]. A basic issue for achieving that goal is the appropriate management of the process quality. In this sense, they are interested in researching how Semantic Web technologies may improve the quality of the educative process [20].
3 Theories of SLSQ Formulations Let L as total words of looping words process through the system based on total words (wn) in the search field typed by the users. Then each word in total of words are matched for getting the rest matched or semantic words (ex) and then each word in total semantic word are matched in whole the dictionary database which consists of total looping semantic words (Es) which retrieve al-Qur’an ayats or verses (mz). Thus the total of looping words formulated is as follows formula
L =
n
w =1
x
w ∈ e ∈ e =1
s
E =1
E ⊆
z
m
(1)
m =1
In mathematical words, L is total looping semantic words which consist of synonym or semantic words and original query words typed by the users. L also is referred as matched words and to be matched again and again until the whole dictionary database will have been traced word by word. The L is collected to be matched with those words in the al-Qur’an collection for retrieving relevant ayats or verses according to L. In matrix equation is assumed as L is as follows. It defines that the prediction each word (a, b, c and d) has carried available relationship synonym words (p, q, r and t). Hence ap+br, cp+dr, aq+bt and cq+dt are the total results of the words have carried
94
M.A. Yunus, R. Zainuddin, and N. Abdullah
the sequence synonym words which are added with each original word in the query. Assume the query has four words which is a, b, c and d. Then SLSQ results of expansion of total synonym words in the dictionary that is ap+br, cp+dr, aq+bt and cq+dt. It means that the total result of relevant synonym words depend on second matrix which lead to unlimited number of rows and column.
a L = c
b p d r
q ap + br = t cp + dr
aq + bt cq + dt
(2)
The last assumption is two brackets which the first bracket which carries the query while the second bracket carries its synonym words. It is therefore the dictionary has provided the matched words and then it looks for the sequence of the synonym words. In this equation, (a+b+c+d) is a query which has four words while (p+q+r+t) are their synonym words. Hence ap+aq+ar+at+bp+bq+br+bt+cp+cq+cr+ct+dp+dq+dr+dt is the SLSQ result with the expansion of two brackets whereby 16 SLSQ words.
L = = ap + bp + cp + dp
(a
+ b + c + d )( p + q + r + t + aq + ar + at
+ bq + br + bt + cq + cr + ct + dq + dr + dt
) (3)
4 System Procedure Figure 1 shows the workflow of Stemming Lopping Referred Semantic Speech Query (SLSQ) which is started with spoken words from query number 1 [15]. The query term can be Arabic, Malay or English. Those words in the query is process to find the whole relevant words as looping referred words semantic words or links list which carry relevant words in the respecting line or index in the whole dictionary. Then searching process categorizes looping referred words as query into two which are keyword and queryword. If the query is keywords, the results retrieved according to word by word results and redundant document names existed in the results. While querywords, results retrieved according to the whole words and no redundant or unique document names retrieved rankly. Then the looping referred semantic words query (SLQ) is conflated with stemming algorithm (SLSQ) after this query is translated in either Arabic [13], Malay stemming Malay [5] or English [11] stemming algorithm according to the query keyed-in. Stemming algorithm removes the suffix, infix, and prefix of each word in the query to be root word in order to retrieve more relevant documents in the results before looping referred semantic words in the query of each language. The last step is
A Framework of Hybrid Semantic Speech Query via Stemmer for Quran Documents
95
Fig. 1. The Framework of SLSQ
retrieval process of SLSQ of spoken words. It retrieves multi-language documents from Arabic, Malay and English. The looping semantic query (LSQ) is the sequence of looping semantic words while they are stemmed for getting the root word of LSQ words for collecting SLSQ words. The explanation of LSQ is shown in Figure 2 to match the synonym words with the available words in the whole dictionary. For example the query has a word and then its synonym and semantic words are b, c, d, e, f and g in a line in a file. In order to get the LSQ words, there are an approach to match a with all words and all lines with words as described in the figure with the arrows with colors according to each word with unique color whereby each word can match with all words in the dictionary such as a word. When it finish, the next synonym word such as b word takes part to be matched until the synonyms words at the first line in the file finish to be matched. All words are collected for providing the relevant and significant results especially in IR.
96
M.A. Yunus, R. Zainuddin, and N. Abdullah
Fig. 2. The SLSQ Process
5 Results and Discussion Table 2 is referred from Yunus [15, 21-24] thesis for query number 1 to prove that the stemming semantic spoken query in CLIR significantly retrieve and provide more and more relevant and related results according to the available words provided in semantic files for three languages. When the spoken words are automatically displayed in search field in Q, those words generate automatically their semantic words to be stemmed into root words. Table 3 shows that stemming looping referred semantic words in speech query evaluation for query number 1 in Arabic, Malay and English language. It indicated that when the Q is key-in, then SLSQ is produced to search more relevant results in the sense of effective searching process performance. Then, searching process is done according to the type of process of results required.
A Framework of Hybrid Semantic Speech Query via Stemmer for Quran Documents
97
All documents are saved in “.txt” format file for UTF-8, ASCII or EBCDIC text. For searching process, word by word matching is used in the process. The matching words refer to the words similarity between query and documents in retrieving process. The query submitted to the system is also represented by translated query that is used to search the related files. Each specific language has been tested and evaluated in terms of recall and precision for semantic and non-semantic. All results depend on stemming process for both query and documents in the experiments. Hence the looping referred words stemming semantic query (SLSQ) from spoken words which is the original query is Q. it means that spoken words of Q yields more significant words from the dictionary become looping referred words or multirelevant words or query. Those are single query and semantic query through the process to display two different results according to specific language. Then those looping semantic words are stemmed to be SLSQ. In this context, those experiments are involved three languages which are Arabic, English and Malay. The comparison between the result in Table 2 [15, 21-24] and 3 is SLSQ yields more synonym words compared SSSQ due to the design of framework to be hybrid processes in terms of speech recognition, looping process of matching semantic and stemming algorithm. It is therefore SLSQ has improved al-Qur’an information retrieval in order to apply SLSQ framework according to hybrid process and design which have been developed. When the system accepts the same query, the results are different between SSSQ and SLSQ. SLSQ yields a lot semantic or synonym words compared to SSSQ. It means that SLSQ has dynamic approach to utilize the words in dictionary database. All results of the query(ies) are referred to the natural language queries of Malay [5] and then translated into Arabic and English [15] queries in this study. Every query is tested to evaluate each result which is matched with manual result as total relevant documents (TRE) for respective query [5]. The evaluation technique is used for precision and recall results [12]. Table 1 shows the formula to calculate the percentage of precision and recall. The semantic speech query is quite significant and easier to be processed from the dictionary that consists of synonyms words and retrieves better results of the most relevant documents. It leads to help to search more and more documents in other languages from the spoken semantic words. These examples also include stemming words and matching the words in every document in collection in order to retrieve the most relevant document required from the query given. It is called as CLIR that focuses on search specific language if given query with the same language. It proves that the spoken semantic query in CLIR is significant to retrieve and provide more and more relevant and related results according to the available words provided in semantic files for three languages. The comparison between Table 4 and 5 shows that the improvement of performance figures and percentages in Table 4 [15, 21-24]. The total retrieved documents (TRT) and total retrieved relevant documents (TRR) are increasing. For example TRT of K (word-by-word) is increasing from 7008 to 23783 and Q (queryword) is remaining the same total in Malay. Then English and Arabic TRT also increase in TRT and TRR in K. Thus Recall percentages of TRR increase. Total relevant documents (TRE) are benchmark for documents retrieval result that consist
98
M.A. Yunus, R. Zainuddin, and N. Abdullah
the similar documents in TRE [5]. All the results are derived from SLSQ in Table 3 which is yielding the synonym words of spoken words from lexical dictionary which is better than SSSQ. It means that the stemming looping referred words semantic spoken query results have retrieved more relevant and significant the stemming looping referred words semantic spoken query documents retrieval results as shown in Table 5. TRE refers to Fatimah (1995) benchmark which is 19 of total relevant documents as her example while the query processed through the system can retrieve more relevant documents compared to her mentioned standard total documents. SLSQ results in Table 5 shows the total retrieved documents which are more recall documents compared to her benchmark whereby Malay K has 23783, Malay Q has 149, English K has 1318, English Q has 104, Arabic K has 3025 and Arabic Q has 379. These values are higher than benchmark which has compared to the current values retrieved by the system. When the system retrieval values to be compared to benchmark, these lead the precision values are lower than 50% for three languages due to the Fatimah (1995) benchmark are not retrieving the same total retrieved documents which have been retrievably approved in this study. If the Fatimah benchmark has the same total with the SLSQ retrieval values, these lead the precision may carry closely 100%. Base on the comparisons however, it means that SLSQ effectively retrieves more relevant documents compared to Fatimah (1995) as well as SSSQ values. Table 1. Recall And Precision Formula
Table 2. Stemming Semantic Of Spoken Query Results Language Q Malay Kelahiran nabi isa English Prophet jesus birth
Arabic
ϲΒϨϟ ϰδϴϋ
SSSQ KELAHIRAN NABI ISA TERHASIL UTUSAN ALLAH LELAKI RASUL BIRTH JESUS PROPHET DNA ABORTION AFFILIATION ALLELE ANCESTRY ANIMATION ARISTOCRACY BABYHOOD BEAR BEGINNING PRINCE OF PEACE LORD SAVIOUR SAVIOR GOOD SHEPHERD ASTROLOGER AUGUR CLAIRVOYANT DAYDREAMER DIVINER DREAMER ENTHUSIAST ESCAPIST FORECASTER
ΪϟϮϣ έϮϬϤΠϟ Ϟπϓ ϡΪϘΗ ΓΪΎϓ ΔΤϠμϣ ΔϴϠπϓ Γΰϴϣ ϰδϴϋ ϲΒϨϟ ΪϟϮϣ Ϳ ΏήτϤϟ έϮϬϤΟ ΐόη έϮϬϤΟ
A Framework of Hybrid Semantic Speech Query via Stemmer for Quran Documents
99
Table 3. Stemming Looping Referred Words Semantic Of Spoken Query Results Language Malay
Q Kelahiran isa
SLSQ nabi KELAHIRAN NABI ISA TERHASIL KEJADIAN KEWUJUDAN KEHIDUPAN PERSENYAWAAN UTUSAN ALLAH PESURUH ALLAH RASUL LELAKI PENYAMPAI PENYEBAR PENDAKWAH LELAKI NABI RASUL MARYAM INJIL ISRAEL jesus BIRTH PROPHET JESUS DNA ABORTION AFFILIATION ALLELE ANCESTRY ANIMATION ARISTOCRACY BABYHOOD BEAR BEGINNING FERTILITY RATE FECUNDITY FERTILITY RATE NATALITY PROLIFICACY RANKNESS RICHNESS PREGNANCY ABUNDANCE APPROPRIATENESS AUSPICIOUSNESS BABYHOOD BEGINNINGS BIGNESS BIRTH BOUNTIFULNESS BROODING CHILDHOOD CONVENIENCE COPIOUSNESS COVERING CRADLE DELIVERY NATIVITY NASCENCY NASCENCE CASSANDRA DRUID ASTROLOGER AUGUR CLAIRVOYANT DAYDREAMER DIVINER DREAMER ENTHUSIAST ESCAPIST FORECASTER MESSAGER PREACHER MAN CHRIST PRINCE OF PEACE LORD SAVIOUR SAVIOR GOOD SHEPHERD ISRAEL MARY ALLAH
English
Prophet birth
Arabic
ﻣﻮﻟﺪ اﻟﻨﺒﻲ ﻋﻴﺴﻰ ﻣﻴﺰة أﻓﻀﻠﻴﺔ ﻣﺼﻠﺤﺔ ﻓﺎﺋﺪة ﺗﻘﺪم وﻻدة ﻓﻄﺮة ﻣﻮﻟﺪ اﻟﻨﺒﻲ ﻋﻴﺴﻰ ﻓﻀﻞ ﻣﻨﻔﻌﺔ ﺣﺴﻦ اﻟﻮﻻدة ﻣﻴﻼد أﺻﻞ ﻣﻨﺸﺄ ﺷﺮﻳﻒ اﻻﺻﻞ ﻧﺎﺷﺊ اﻟﺠﻤﻬﻮر ﺟﻤﻬﻮر ﺷﻌﺐ ﺟﻤﻬﻮر اﻟﻤﻄﺮب ﺣﻀﻮر ﻣﺴﺘﻤﻊ رﺳﻮل ﻗﺎﺋﺪ ﻣﻠﻬﻢ ﺷﺎﻋﺮ ﻣﻠﻬﻢ ﻧﺬﻳﺮ اﻟﺸﺆم ﻧﺬﻳﺮ اﻟﺴﻮء ﷲ
Table 4. Stemming Semantic Spoken Query Documents Results Evaluation Language
TRT
TRE
Recall(%)
Precision(%)
K
Q
K
Q
K
30
7
157.89
36.84
0.43
4.70
19
5
4
26.32
21.05
0.45
3.85
19
12
6
63.16
31.58
0.49
1.58
K
Q
Malay
7008
149
19
English
1101
104
Arabic
2455
379
TRR
Q
Table 5. Stemming Looping Referred Words Of Semantic Spoken Query Documents Retrieval Results Language
TRT
TRE
K
Q
TRR
Recall(%)
Precision(%)
K
Q
K
Q
K
Q
Malay
23783
149
19
94
7
494.74
36.84
0.40
4.70
English
1318
104
19
7
4
36.84
21.05
0.53
3.85
Arabic
3025
379
19
13
6
68.42
31.58
0.43
1.58
100
M.A. Yunus, R. Zainuddin, and N. Abdullah
6 Conclusion and Future Work All results depend on looping referred words semantic query is embedded with stemming process for both query and documents in the experiments. Looping referred words semantic words cover and employ whole related words in the dictionary from single word as a query. In the comparison between looping referred semantic words and semantic itself, looping referred words or many links list query successfully improve the semantic query. Hence Table 3 also shows the stemming method to be conflated with looping referred words semantic results from spoken words (SLSQ) has more relevant and significant synonym words in order to retrieve more retrieved relevant and significant documents results in Table 5. Acknowledgement. This research has been funded by the University of Malaya, under the grant number (PS210/2009B) and full-scholarship from the University of Malaya. Thus I would like to forward our deepest thanks to Prof. Dr. Roziati Zainuddin and Dr. Noorhidawati Abdullah from the Faculty of Computer Science and Information Technology for their endless assistance, technical advice and cooperation.
References 1. Alberto, P.: Multi-model music content description and retrieval using IEEE 1599 XML standard. Journal of Multimedia 4(1) (2009) 2. Aufaure, M.A., Grand, B.L., Soto, M., Bennacer, N.: Metadata- and Ontology- Based Semantic Web Mining. In: Taniar, D., Wenny Rahayu, J. (eds.) Web Semantics and Ontology, ch. 9, p. 406, pp. 259–296. Idea Group Publishing, (March 2006) ISBN: 1591409055 3. Baazaoui-Zghal, H., Aufaure, M.A., Mustapha Ben, N.: A model-driven approach of ontological components for on-line semantic web information retrieval. Journal on Web Engineering, Special Issue on Engineering the Semantic Web 6(4), 309–336 (2007) 4. Baazaoui-Zghal, H., Aufaure, M.A., Soussi, R.: Towards an on- line semantic information retrieval system based on fuzzy ontologies. Journal of Digital Information Management 6(5), 375–385 (2008) 5. Fatimah, A.: A Malay language document retrieval system an experiment approach and analysis. Tesis Ijazah Doktor Falsafah Universiti Kebangsaan Malaysia (1995) 6. Gruber, T.: Toward Principles for the Design of Ontologies Used for Knowledge Sharing. International Journal of Human-Computer Studies, Special Issue on Formal Ontology in Conceptual Analysis and Knowledge Representation (1993); Guarino, N., Poli, R. (eds.) 7. Javier, T., Dong, W., Joe, F., Simon, K., Jose, C.: A comparison of grapheme and phoneme-based units for Spanish spoken term detection. Speech Communication, 980–991 (2008) 8. Lopez, V., Motta, E., Uren, V.S.: PowerAqua: Fishing the Semantic Web. In: Sure, Y., Domingue, J. (eds.) ESWC 2006. LNCS, vol. 4011, pp. 393–410. Springer, Heidelberg (2006) 9. Lopez, V., Uren, V., Motta, E., Pasin, M.: AquaLog: An Ontology-Driven Question Answering System for Organizational Semantic Intranets. Journal of Web Semantics 5(2), 72–105 (2007)
A Framework of Hybrid Semantic Speech Query via Stemmer for Quran Documents
101
10. Pasi, G.: Flexible Information Retrieval: Some Research Trends. Mathware and Soft Computing IX(9-1), 107–121 (2002) 11. Porter, M.F.: An Algorithm for Suffix Stripping. Program 14(3), 130–137 (1980) 12. Salton, G., Mcgill, M.J.: Introduction to Modern Information Retrieval. McGraw-Hill, New York (1983) 13. Shereen, K.: Arabic Stemmer (2002), http://zeus.cs.pacificu.edu/shereen/ 14. Špela, V.: Corpora in Translation: A Slovene Perspective. The Journal of Specialised Translation (10) (2008) 15. Yunus, M.A.M.: Short Query Translation: A Dictionary-Based Approach to Cross Language Information Retrieval. Master of Computer Science, Thesis, Universiti Teknologi MARA, Malaysia (2008) 16. Donoser, M., Wagner, S., Bischof, H.: Context Information from Search Engines for Document Recognition. Pattern Recognition Letters 31, 750–754 (2010) 17. Hueber, T., Benaroya, E.L., Chollet, G., Denby, B., Dreyfus, G., Stone, M.: Development of A Silent Speech Interface Driven by Ultrasound and Optical Images of The Tongue and Lips. Speech Communication 52, 288–300 (2010) 18. Mausam, Soderland, S., Etzioni, O., Weld, D.S., Reiter, K., Skinner, M., Sammer, M., Bilmes, J.: Panlingual Lexical Translation via Probabilistic Inference. Artificial Intelligence 174, 619–637 (2010) 19. Tsai, R.T.H.: Chinese Text Segmentation: A Looping Referred Words Approach Using Transductive Learning and Statistical Association Measures. Expert Systems with Applications 37, 3553–3560 (2010) 20. Castellanos-Nieves, D., González-Martínez, R., Soler-Segovia, C., Robles, M.C., Hernández-Franco, J., Prendes-Espinosa, M.P., Fernández-Breis, J.T.: Semantic WebBased System for Managing the Educative Curriculum. Procedia - Social and Behavioral Sciences 2, 521–526 (2010) 21. Yunus, M.A., Zainuddin, R., Abdullah, N.: Semantic Query for Quran Documents Results. In: 2010 IEEE Conference on Open Systems (ICOS), pp. 1–5 (2010) 22. Yunus, M.A., Zainuddin, R., Abdullah, N.: Semantic Speech Query via Stemmer for Quran Documents Results. In: 2011 International Conference on Electronic Devices, Systems and Applications (ICEDSA), pp. 17–21 (2011) 23. Yunus, M.A., Zainuddin, R., Abdullah, N.: Visualizing Quran Documents Results by Stemming Semantic Speech Query. In: 2010 International Conference on User Science and Engineering (i-USEr), pp. 209–213 (2010) 24. Yunus, M.A., Zainuddin, R., Abdullah, N.: Semantic Query with Stemmer for Quran Documents Results. In: 2010 IEEE Conference on Open Systems (ICOS), pp. 40–44 (2010)
A Semantic Layer for a Peer-to-Peer Based on a Distributed Hash Table Mohammed Ammari, Dalila Chiadmi, and Laila Benhlima University Mohammed V Agdal Rabat, Mohammadia School of Engineers, Labo SIR Rabat, Morocco
[email protected], {chiadmi,benhlima}@emi.ac.ma
Abstract. Peer to Peer Distributed Hash Table (DHT) facilitates access to specific data when their identifiers or keys are known accurately in advance. However, in practice, users searching for resources stored in P2P systems possess only partial information for identifying these resources (e.g. keywords, metadata, queries in XQuery...). In this paper, we propose the use of multidimensional data analysis to ensure semantic indexing of metadata. The aim is to confer the semantic aspect to P2P DHT in order to respond to broad queries. Our system takes the form of a semantic layer that can be superimposed on top of any P2P DHT infrastructure .This layer is subdivided as areas through a Hilbert curve which has the merit of a good preservation of the locality and semantic affinity. Digital libraries are presented as a formal context to illustrate our system. Keywords: P2P DHT, Semantic classification, multidimensional data analysis, Correspondence analysis, Hilbert curve, metadata, ontology.
1 Introduction Nowadays, the integration in P2P has a cardinal interest for its flexibility and its application in various fields (massive computing distributed in the form of Gridcomputing, massive data storage, instant messaging, videoconferencing, Voice over IP Multi-player ...). In this article, the focus is on structured P2P. Structured systems organize the space of nodes in a data structure well defined. They also organize the distribution of resources on these nodes. They thus allow overcoming the obstacles inherent to unstructured P2P model. Indeed, the unstructured P2P suffers from several disadvantages (high consumption of bandwidth, no concept of peer quality, no estimate for the duration of query execution...). There are several examples of structured P2P systems: P-Grid [1] uses a data structure of type tree, Chord [2] uses a ring like structure, and CAN [2] uses a structure based on a virtual two-dimensional Cartesian space divided into zones. These structures provide both efficiency and security research. Most structured systems use a DHT (Distributed Hash Tables) for the management and allocation of resources on these nodes (e.g. Chord [2], CAN [3], Pastry [4], and Tapestry [5]). A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 102–114, 2011. © Springer-Verlag Berlin Heidelberg 2011
A Semantic Layer for a Peer-to-Peer Based on a Distributed Hash Table
103
An exact matching key for access to resources is a major constraint to DHT P2P system: we need to know the exact key (identifier) of a resource to locate the node responsible for storage. However, in reality, users often have only partial information for identifying these resources, and tend to submit broad queries (e.g. all the articles written by a particular author). P2P DHT also suffers from an intrinsic problem of semantics. Indeed, the application of the hash function on semantically related information gives independent keys that will be stored later in independent nodes (declustering). In this paper, we propose to refine the DHT P2P systems with semantics based on metadata and ontology. We used practical techniques in order to locate resources using partial information. Our mechanisms are based on multidimensional data analysis that processes data collected most often in rectangular arrays of very large size. It aims to reveal the structure and make an approximation by other tables that have lower dimensions in order to provide a simpler interpretation. Our system can be overlaid on top of any infrastructure P2P DHT. Therefore, it could benefit from a broad range of benefits offered by DHT (e.g. load balancing, replication, scalability, flexibility, autonomy) We place ourselves in the context of digital libraries to illustrate our system, and highlight the techniques proposed. In the following section we discuss some related work. Then we specify the formal context and present some definitions. In the fourth section we present our solution before closing on future job prospects.
2 Related Work P2P recent work is marked by the use of ontology and semantics in order to overcome the problems inherent to this mode [6] [7] [8]. Our work can be classified in the field of indexing metadata in order to confer the semantic aspect to P2P DHT. Joseph and Hoshiai [9] have listed the different strategies underlying metadata P2P research. These include among others the TFIDF (Term Frequency Inverse Document Frequency). It is an approach to information retrieval of Salton and Yang's [10] for measuring the degree of representativeness of a word in a document. The TFIDF can be calculated automatically from the text of a document. There are several variations of the basic approach of TFIDF usually combined with the use of a VSM (Vector Space Model), or by means of LSI (Latent Semantic Indexing) [11] which allows overcoming the problems related to VSM, such as synonymy, polysemy, or errors in documents. In the article of Schlosser et al. [12] HyperCup [12] [13] structure (non dht-based) was replaced by a semantic relationship between nodes. Thus nodes offering the same services are brought together in the same Cluster Concept. That said, the dimension of the hypercube depends on the number of basic concepts in the network. But the downside is that two semantically related nodes together in the same cluster concept are not necessarily close in the underlying physical topology. This is reflected in a decline in performance compared to the original HyperCup [12] [13]. In the article of Sunaga et al. [14], metadata is used to improve the routing of requests. But the work suffers from the growth of indexes size supporting concepts, forced to meet the scalability of the system. The paper [15] presents the design and implementation of Bibster, a solution fully implemented, open source, built on top of the JXTA platform. It is a P2P system of
104
M. Ammari, D. Chiadmi, and L. Benhlima
exchange of bibliographic data among researchers. It uses ontology for data storage, formulation and routing queries, and for presentation of responses. But this system is being optimized due to the attractive alternatives of different underlying modules. Rostami et al. [16] presented a hybrid approach between indexing and semantic annotation. They propose the use of ontology in order to limit the size of indexes without loss of information on indexing. This proposal reduces network traffic significantly, and improves the routing of requests and system maintenance. Our system has the merit of gather semantically related metadata into a single semantic category, composed of nodes close in the underlying physical topology (P2P DHT). This is reflected in a significant gain in performance. Another major advantage of our system is its adoption of a Hilbert curve to support the emergence of new semantic classes more specialized in order to satisfy its scalability while maintaining the best possible location and semantic affinity.
3 Formal Context and Definitions Our system takes the form of a semantic layer that can be superimposed on top of any DHT P2P infrastructure. This layer is subdivided as areas. Each zone will be managed via a node (or more nodes serving as backup) which has excellent characteristics in terms of storage capacity, CPU, etc. The system will manage the bibliographic records, which are presented in light forms of metadata. However, storage of the full text of a document will be delegated to the infrastructure underlying P2P DHT. A Hilbert curve is a continuous fractal space-filling curve first described by the German mathematician David Hilbert in 1891. It is demonstrated that the trajectory is dense in the square of departure, therefore equal to the square. The Hilbert curve has been proposed in the context of multidimensional databases because it has a better behavior preserving the locality.
Fig. 1. First, second and third iteration of the Hilbert curve
A Semantic Layer for a Peer-to-Peer Based on a Distributed Hash Table
105
Factor analysis is an exploratory method analyzing the contingency tables, developed primarily by J. -P., Benzecri during the period 1970-1990. It is a set of multivariate statistical methods whose primary purpose is to define the structure of the correlations between many variables, determining a set of common dimensions called factors. Any method of factor analysis has for Input:
{
}
ii)
Ν( I ) = X i ∈ R P , i ∈ I a system of mass attributed to the point X i
iii)
A metric that calculates the distances between
i)
a scatter plot
Xi
For Output: i) ii)
axes of inertia Points X i with their coordinates on the axes.
iii)
A number of additional indicators, known as indicators of decision support.
Ν(I ) looks like a rugby ball. The scatter plot has orthogonal axes of inertia u1 , u 2 , u 3 (see Figure 2). Perform a factor analysis is to find these axes. It is demonstrated that u1 is the eigenvector corresponding to the largest In ℜ , the scatter plot 3
eigenvalue of the matrix of inertia V = X MX where X is the matrix of initial data centered by the barycentre G, and M is the diagonal matrix of weights vectors. t
Fig. 2. The axes of inertia of the scatter plot
106
M. Ammari, D. Chiadmi, and L. Benhlima
If we project the cloud of points Ν(I ) on the factorial plane defined by the first two axes of inertia, we obtain n projected points, this time visible (see Figure 3). This image will be even more faithful to reality than the percentage of inertia explained by the two axes will be greater. Variables that have strong contributions on an axis are those with the highest coordinates on this axis. The proximity between two sources of data means their correlation, having said that there is a semantic affinity between them. Indeed
(
) (
d 2 X i , X i' = 2 1 − rii' When
rii '
)
representing the correlation coefficient between X i and Xi' is equal to
(
cov X i , X i '
)
( )
var ( X i ) var X i ' It is important to note that the covariance between
X i and Xi
'
is a number to evaluate
the independence between these two variables. The contrast between two data sources indicates a negative correlation. Perpendicularity means independence. For example groups I and II, and III and IV are opposed to each other. Data sources in the same group are similar to each other, and dissimilar of data sources owned by other groups. That said they are characterized by strong intragroup homogeneity, and high intergroup heterogeneity.
Fig. 3. Projection of the scatter plot on the first factorial plane
A Semantic Layer for a Peer-to-Peer Based on a Distributed Hash Table
107
4 Solution Overview Our system takes the form of a semantic layer that can be superimposed on top of any DHT P2P infrastructure. This layer is subdivided as areas through a Hilbert curve which has the merit of a good preservation of the locality. We will make a correspondence between the semantic domains extracted from a global ontology and areas of Hilbert (see Figure 4).
Fig. 4. Correspondence between the global ontology and areas of Hilbert
The system is divided in the form of semantic categories. A category is a semantic grouping of records with high internal cohesion and low external coupling. Each semantic category is responsible for managing semantically related records under its jurisdiction. It will be managed through a node (or more nodes serving as backup) which has excellent characteristics in terms of storage capacity, CPU, etc. To avoid creating bottlenecks, we will use the classification in the various semantic categories. Indeed a division of each semantic category in more specialized subcategories is essential, which will give increased scalability to our system, and allow it to manage complexity by structuring. The classification within a semantic category will be conducted through the correspondence analysis. At the end of this operation of
108
M. Ammari, D. Chiadmi, and L. Benhlima
classification, the order of the iteration of Hilbert curve will be increased by a notch in order to support the new emerging classes (see Figure 5). This process will be repeated recursively in a progressively fine-tuning of the ontology to satisfy the scalability of the system.
Fig. 5. Evolution of the Hilbert curve with the refinement of the ontology
We have seen that each semantic category will act as a containment vessel, where the various semantically related documents will be embedded in. Indeed, each semantic category will manage a matrix of contingency ensuring correspondence between I the set of data sources and J the set of terms related to the semantic domain, extracted from an ontology constructed from the data schemas. The contingency matrix is defined by
F = ( f ij )(i , j )∈[1, n ]*[1, p ]
the frequency of occurrence in the entire population of the couple extracted term
where
f ij is
(i, j ) formed by
T j , j ∈ J and the data source S i , i ∈ I .
To prevent the genesis of bottlenecks, and overcome the problem of storing a huge and hollow contingency matrix, we will use the classification of data sources through the multidimensional data analysis. In our case we use the correspondence analysis. We now have the triplet which allows us to perform a factor analysis:
{
i) a scatter plot Ν ( I ) = X i ∈ R n , i ∈ I
}
Simplex distribution is the set of points X of R n which has the coordinates defined by:
x j ≥ 0 satisfying
Each data source
n
x j =1
j
xj
=1 .
S i is associated with the point X i in the n-dimensional space,
which has the following coordinates
A Semantic Layer for a Peer-to-Peer Based on a Distributed Hash Table
f ij
x ij = f j / i = With
f j/i
And
f i.
the frequency of the term T
j
109
f i. Si
in the data source
Si in the entire population.
the frequency of the data source
We prove that: p
p
x
ij
j =1
=
p
j =1
f j/i =
p
f ij
j =1
f i.
=
f ij
j =1
f i.
=1
So each data source is located on the simplex of distribution R n
ii)
A mass distribution
X i -- f i P
X i we affect the mass f i . =
At each point
f ij
j =1
Therefore, we find the size information concerning the data source. This information has been lost in the transition from the source S i to the point Xi
iii)
The distance between
x ij =
f ij f i.
Xi = (xi1, xi2 ,...,xip )
We reduce by
d With
(
)
Distance to measure the similarity between points. Indeed, to calculate
2
(X
t
and
Xi' = xi' 1 , xi' 2 ,...,xi' p
G = ( f .1 , f .2 ,..., f . p ) using the distance χ i, X
i'
)=
1 f. j
j∈ J
f . j represents the frequency of the term T
j
f i' j f ij − f i. f i'.
t
with
2
in the entire population.
The correspondence analysis is a method widely used to analyze the crosstabulations, including contingency table crossing two qualitative variables. We place ourselves in the context of digital libraries to illustrate our proposals, and highlight classification techniques based on multidimensional data analysis. We choose "Computer Science" as a semantic category. We consider that we have 10 data sources, I = S1, S2 , S3, S4 , S5 , S6 , S7 , S8 , S9 , S10 and we restrict the relevant candidate
{
}
terms to 14, J = {ROM, BUS, ASP, PHP, C, Oracle, MySQL, Drivers, CPU, C + +, Linux, RAM, Compilation JAVA}. The frequency of keyword occurrence in a data source is evaluated according to Likert scale from 1 to 9 (see table1). The simulation of results is performed using the software XLSTAT 2011.
110
M. Ammari, D. Chiadmi, and L. Benhlima Table 1. Contingency matrix ensuring the correspondence between sources and terms
The interest of the Correspondence Analysis is to provide a graphical representation of scatter plots I and J (dual cloud) in a space of lower dimension, allowing highlight the hidden correspondences that the numerical calculation does not immediately reveal. Traditionally, the representation is done in two dimensions by plotting the first factorial plane. The quality of the graphical representation can be evaluated by the eigenvalues histogram. If the sum of the first two eigenvalues represents a large portion of the total inertia, the quality of graphics is considered good.
Fig. 6. Histogram of eigenvalues
A Semantic Layer for a Peer-to-Peer Based on a Distributed Hash Table
111
In our case, the quality is good since the first two eigenvalues account for 73.83% of the total variance (see Figure 6). The main advantage of correspondence analysis is simultaneous graphical representation of the terms of two variables. When the quality is good (73.83% in our case) the graphical representation can easily interpret the data. The proximity between two points-lines (data sources) (such as S10 and S5) reflects a similar profile of key words (see Figure 7). The proximity between two pointscolumns (keywords) (such as C, C + +, Java ...) reflects a similar profile of data sources. The simultaneous representation of points-lines and points-columns is used to identify the variables responsible for some proximities. This can be exploited to search efficiently. Thus the relevant data sources will be close - as defined in the distance
χ2
- of the keywords used in a search.
Fig. 7. Simultaneous graphical representation of data sources and keywords
We can note that the sources S8, S10, S5 and S2 rather cover the area of Software (databases, programming languages, ...) while the data sources S4, S9, S1, S6, S3, S7 tends to cover the area hardware (CPU, bus, RAM, ROM, ...). That said the semantic category “Computer Science” will be divided into two semantic categories, more specialized and smaller (“Software” and “Hardware”).
112
M. Ammari, D. Chiadmi, and L. Benhlima
The integration of new data sources is made as follows: The information element to insert spreads gradually to the way of an epidemic in a human group. This element can spread to different semantic categories through an epidemic protocol [17], [18]. Any new data source to insert will be considered as a fictitious source S fictive in each semantic category (see table2). Table 2. Insertion of the fictive data source in the matrix of contingency
We know that the variables which are best represented in the factorial plan are those which have the highest coordinates. That said, for example, the weakness of the contribution of the fictitious source
S fictive
to form factors (representing the
underlying dimensions, which when interpreted, describes the data in summary form) leads us to refute the assumption that S fictive belongs to the semantic category (see Figure 8).
Fig. 8. Projection of the fictitious source in the first factorial plane
A Semantic Layer for a Peer-to-Peer Based on a Distributed Hash Table
113
The data source to insert can belong to one or more semantic categories. For example a record in the field of biochemistry may belong to semantic categories “Chemistry” and “Biology”. This will allow our system to manage the overlap between the different areas. The search for relevant data sources within a semantic category will be made according to the algorithm below: i) Receiving a request in the form of XQuery. ii) Extraction of terms from the predicate. iii) Semantic enrichment of terms extracted from the predicate, with semantically related terms from an ontology constructed from the data schemas of semantic category. iv) Construction of the sub matrix of contingency v) Discharge of sources
S i checking : ∀j ∈ J
Where
υ
F = ( f ij )(i , j )∈[1, n ]*[1, p ]
pertinence
f ij f i.
<
υ
pertinence
represents the relevance threshold parameter set by
the system administrator.
vi) Ranking of selected data source, in order of relevance.
5 Conclusion and Perspective In this paper, we used multidimensional data analysis to ensure semantic indexing of metadata. The aim is to confer the semantic aspect to P2P DHT in order to respond to broad queries. We opted for a Hilbert curve to support the emergence of new more specialized classes. However, we estimate that our work can be improved if we assign a Hilbert curve for each semantic category. Indeed this will ensure the independence of the evolution of a semantic category compared to evolutions in other semantic categories. The implementation of the solution through the J2EE platform is an imminent prospect of our work.
References 1. Aberer, K., Cudré-Mauroux, P., Datta, A., Despotovic, Z., Hauswirth, M., Punceva, M., Schmidt, R.: P-Grid: A Self-organizing Structured P2P System Distributed Information Systems Laboratory École Polytechnique Fédérale de Lausanne, EPFL (2003) 2. Stoica, I., Morris, R., Karger, D., Kaashoek, M., Balakrishnan, H.: Chord: A scalable peerto-peer lookup service for internet applications. In: Proc. ACM SIGCOMM (2001) 3. Ratnasamy, S., Handley, M., Karp, R., Shenker, S.: A scalable content-addressable network. In: Proc. ACM SIGCOMM (2001)
114
M. Ammari, D. Chiadmi, and L. Benhlima
4. Rowstron, A., Druschel, P.: Pastry: Scalable, Decentralized Object Location, and Routing for Large-Scale Peer-to-Peer Systems. In: Guerraoui, R. (ed.) Middleware 2001. LNCS, vol. 2218, pp. 329–350. Springer, Heidelberg (2001) 5. Zhao, B.Y., Kubiatowicz, J., Joseph, A.D.: Tapestry: An infrastructure for fault tolerant wide-area location and routing. Technical Report UCB/CSD-01-1141, University of California, Berkeley (April 2001) 6. Broekstra, J., Ehrig, M., Haase, P., van Harmelen, F., Kampman, A., Sabou, M., Siebes, R., Staab, S., Stuckenschmidt, H., Tempich, C.: A metadata model for semantics-based peer-to-peer systems. In: Proceedings of the WWW 2003 Workshop on Semantics in Peerto-Peer and Grid Computing (2003) 7. Nejdl, W., Wolf, B., Qu, C., Decker, S., Sintek, M., Naeve, A., Nilsson, M., Palmér, M., Risch, T.: Edutella: A P2P networking infrastructure based on rdf. In: Proceedings to the Eleventh International World Wide Web Conference (2002) 8. Castano, A., Ferrara, S., Montanelli, S., Pagani, E., Rossi, G.: Ontology-addressable contents in p2p networks. In: Proceedings of the WWW 2003 Workshop on Semantics in Peer-to-Peer and Grid Computing (2003) 9. Joseph, S., Hoshiai, T.: Decentralized meta-data strategies: Effective peer-to-peer search. IEICE Transactions on Communications E86-B (6) (2003) 10. Salton, G., Yang, C.: On the specification of term values in automatic indexing. J. Doc. 29, 351–372 (1973) 11. Deerwester, S., Dumais, S.T., Furnas, G.W., Landauer, T.K., Harshman, R.: Indexing by Latent Semantic indexing (1990) 12. Schlosser, M., Sintek, M., Decker, S., Nejdl, W.: Hypercup -hypercubes, ontologies and efficient search on p2p networks. In: International Workshop on Agents and Peer-to-Peer Computing, Bologna, Italy (July 2002) 13. Schlosser, M., Sintek, M., Decker, S., Nejdl, W.: Hypercup shaping up peer-to-peer networks. Technical Report, Stanford University (2001) 14. Sunaga, H., Ueda, K., Iwata, T., Kikuma, K., Takemoto, M.: P2p applications using the semantic information oriented network. In: P2P 2004: Proceedings of the Fourth International Conference on Peer-to-Peer Computing. IEEE Computer Society, Washington, DC, USA (2004) 15. Broekstra, J., Haase, P., Harmelen, F.V., Menken, M., Mika, P., Schnizler, B., Siebes, R.: Bibster-a semantics-based bibliographic peer-to-peer systems. In: The Second Workshop on Semantics in Peer-to-Peer and Grid Computing (2005) 16. Rostami, H., Habibi, J., Livani, E.: Semantic routing of search queries in P2P networks J. Parallel Distrib. Comput. (2008) 17. Bertier, M., Busnel, Y., Kermarrec, A.: Rumeurs, populations et communautés: équivalence uniquement sociologique? Protocole de population versus protocoles épidémiques 12èmes Rencontres Francophones sur les Aspects Algorithmiques de Télécommunications (AlgoTel) (2010) 18. Felber, P., Kermarrec, A.-M., Leonini, L., Rivière, E., Voulgaris, S.: PULP: Un protocole épidémique hybride AlgoTel (2009)
Design and Development of a Ubiquitous Cancer Care System: A Collaborative Communication among Cancer Community Ariffin Abdul Mutalib1 and Norlaily Hashim2 1
School of Multimedia Technology Computing, School of Computing College of Arts and Sciences, Universiti Utara Malaysia, 06010 UUM Sintok, Kedah, Malaysia
[email protected] 2
Abstract. Communication among cancer community (specifically patients, parents, and medical practitioners) has not been pervasively-supported. In response to that, this paper reports on a study that ensures communication among cancer community is supported with a ubiquitous communication model. The model is proposed to provide an alternative for the community to communicate; discussing among patients, parents, and medical practitioners. In particular, the system is important because the treatments for cancer mostly take a long time, and the entities in the community are normally closed, which leads to frequent communications. Furthermore, most of topics communicated among cancer community are only understood by themselves. Hence, the paper aims at discussing the design and development of respective prototype. An appropriate model is formed first; then the prototype development follows. The prototype is discussed at length in this paper. At the end, initial feedbacks from the cancer community prove that the designed prototype works well in supporting the communication among cancer community in a ubiquitous environment. Keywords: Ubiquitous, web-based application, cancer community, communication.
1 Background of Study Undeniably, the rise of social technology such as Web sites brings a lot of benefits to users. It provides capabilities and opportunities for users to share their experiences, to expose their tacit knowledge, to discuss a wide range of topics, and to seek for support [1], in a ubiquitous environment. According to[2], social interaction web sites such as Blogger, Facebook, Friendster, MySpace, and Twitter have radically changed user interactions styles on the World Wide Web (WWW) from a static, one way, and consumption model to a dynamic, multi-way, and participation model [3], in making communication pervasive. In addition, Walther and Boyd [4] found that social support communication traditionally is considered to be the exchange of verbal and nonverbal message such as expression, information, or recommendation. Nevertheless, traditional social support recently has been transformed into computer-mediated communication A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 115–134, 2011. © Springer-Verlag Berlin Heidelberg 2011
116
A.A. Mutalib and N. Hashim
(CMC). It involves in the large social network among people who not know each other and it has eliminated the face-to-face communication. While traditional is referred to a particular person in seeking support, online interaction support enables a particular person goes to a virtual space where no one knows any one. The concept of online social networking is prospectively potential in supporting cancer community in communicating among themselves [5]. A study conducted by Schweizer et al. [6] shows that most of the issues of social relationship split amongst cancer patients are caused by the burden of the disease. Therefore, new relationships have to be established. In addition to fellowships and self-help support groups, virtual cancer communities have the capability to provide an environment in which the relationships can be established. Nevertheless, other advantages of such system include: (1) the internet allows patients to seek support from the person who have similar experiences, (2) patients can use the internet and be connected to the people at any time regardless geographical or physical limitations, and (3) the internet allows patients interact with persons who have different experience about the disease. Hence, a model for social interaction has to be designed and developed for the cancer community. This study proposes that the model is called Cancer Care System (CCS). The main objective of this study is to ensure communication among cancer community is well-supported. To achieve that, three sub-objectives are outlined: (1) to determine the components of CCS, (2) to design and develop a prototype of CCS, and (3) to test the CCS with the end users in order to investigate their satisfaction on the prototype. This paper intends to discuss about objectives 1 and 2 and their results, leaving objective 3 for future discussion.
2 Related Works This section focuses on some literatures that are significant to this study. It starts with online social interaction, followed by online social support for cancer community. 2.1 Online Social Interaction The main characteristics of online interaction are isolation from face-to-face communication which includes broader scope of people who can connect with and the anonymity which text-based communication provides [7]. According to Caplan and Turner [8], the potential for greater anonymity in online social interaction enables people who are unwilling to seek support through face-to-face changing to communicate with others in a straightforward and honest way about their conditions, to be more self-disclosure, and to be more direct in asking personal questions about others. Walther [9] proposes that CMC may be more effective than traditional face-to-face behavior for interpersonal endeavors that involve social risk. According to Caplan [10], one of the features of CMC is that online social interaction often involves less personal and social risk than face-to-face conversation. Thus, a safe and secured environment is easy to be established in which most participants are willing to discuss deeply regarding personal and upsetting topics. In addition, Caplan [10] also argues that those who are lonely and involved in online social interaction feel safer, more efficient, more comfortable, and more confident with the virtual relationship rather than the traditional face-to-face social activities. Also, people suffering emotional
Design and Development of a Ubiquitous Cancer Care System
117
distress from stigmatized conditions such as cancer, HIV, and physical disabilities might benefit from the reduced social risk associated with online support. In addition, it is challenging for them to seek face-to-face social support because people will try to conceal their actual stigmatized conditions from others [8]. According to Walther and Boyd [4], since the virtual relationships established through online support is weaker than face-to-face relationships, then the conversations of stigmatized issues are likely to be perceived as less threatening than face-to-face mode. Thus, individuals suffering from stigmatized conditions might benefit from online social support, because they have unique opportunity to discuss any personal and sensitive topics without concerning the inherent interpersonal costs of disclosing potentially embarrassing information among each other [8]. 2.2 Online Social Support for Cancer Community Eysenbach [11] claims that provision of information over cancer patients has helped the patients gaining controls, reduce distresses and nervousness, promote participation and self-care, generate feelings of safety and security, create realistic expectations, and improve observance. Patients who are satisfied with the accuracy and sufficiency of the given information might feel happy with their level of participation in the overall process of decision making. Online communication is a coping strategy employed by patients and their caregivers throughout the cancer experiences [5]. In conjunction, Burrows et al. [12] found that the sharing and obtaining information through online community can empower cancer patients and their caregivers and give the forward momentums in their life. In another study, Wright [13] found that cancer patients were more motivated to communicate interpersonally online compared with their families and friends who never suffer from the stigmatized condition similar with themselves. There are three key reasons making online support overcomes faceto-face support that are pointed out by Sundquist et al. [14] i.e. flexibility, overcoming barriers to community, and cost effective.
3 Methods This section describes the methods used in this study. It starts off with defining the requirements of CCS, followed by design and development of the CCS where prototyping approach is used. Finally data collection and analysis follow. Figure 1 Phase
Activity
Requirement Definition
•Document study •Comparative analysis •Interview
Design and Development
•Prototyping
User Testing
•Data collection •Data analysis
Outcome
1 Component of Cancer Care System
2 Prototype of Cancer Care System
3
Fig. 1. Summary of the Processes
Level of user’s satisfaction
118
A.A. Mutalib and N. Hashim
illustrates the activities and the outcomes for each phase. Activities in phases 1 and 2 are explained deeply in the following sections. Phase 3 is not included in this paper. 3.1 Requirement Definition There were three fact finding techniques used to gather requirements for this study; interview, document study, and comparative analysis. In this study, unstructured interviews with open-ended question were conducted with cancer community members who consist of patients, parents, and medical practitioners. The main purpose of the interviews was to gather and identify the requirements for CCS. The interview questions were focused on the problems or limitations the cancer community faces on daily basis regarding the interaction among community members, the expected features in the system, and social support tools they expect in the system. Document review consists of reading materials in journals, magazine, newspaper, books, and online resources. Other than that, a comparative analysis over five cancer information system was carried out. This is because the characteristics of the systems could be homogenous and could be generalized. Thereby, components of CCS are inherited from those systems. This paper does not intend to discuss about the comparative analysis. The objectives of the comparative study are to obtain in-depth understanding about CCS and to gather the components for CCS. 3.2 Design and Development In this study, the development of CCS utilized prototyping approach. In this study, the prototyping process consists of four steps which is adapted from Laudon and Laudon [15] as shown in Figure 2. Identify basic requirements
Develop initial prototype
Use the prototype
Yes
User satisfy ? No
Evaluate as operation prototype
Revise and en hance the prototype
Fig. 2. Prototyping Process
Design and Development of a Ubiquitous Cancer Care System
119
As seen in Figure 2, the activities are divided into identifying basic requirements of CCS, develop the initial prototype, use the prototype, and testing. For the purpose of this paper, when the prototype is ready, it was let for the cancer community to experience it. Then, they can have personal feedbacks on either they prefer it in supports of their ubiquitous communication. Meanwhile, the remaining steps are not within the scope of this paper.
4 Design and Development This section elaborates at length on the design and development of the CCS. It starts with the design followed with the CCS after being developed. 4.1 The Design of CCS This section illustrates the design of the CCS including the architecture design, prototype modules and their functionalities, database design, and UML diagrams, which are almost similar to works by Wahab et al. [16], Kadir et al. [17], Wahab et al. [18], and Wahab et al. [19]. 4.1.1 Architecture Design CCS is a web-based system. Thus, its architecture comprises a web server, database, terminals, and networking. The CCS is installed and released on a web server which is connected to the network. Users and system administrator can access the CCS online and all the operations are controlled by the web server. In a nutshell, the architecture design of CCS is shown in Figure 3.
Fig. 3. Architecture Design of CCS
4.1.2 The Prototype Modules The CCS consists of two major modules namely Cancer Community and Administrator as shown in Figure 4. Each module is further explained in the following subsections.
120
A.A. Mutalib and N. Hashim
Cancer Care System CCS
Administrator
Cancer Community
Fig. 4. Cancer Care System Main Modules
4.1.2.1 Cancer Community Module. Users of this module are cancer patients, parents, and medical practitioners. This module further consists of four sub modules which are Chat, Manage Message, Manage Bulletin, and Manage Forum as illustrated in Figure 5. Each sub module of Cancer Community consists of several functionalities and operations as listed in Table 1.
Cancer Community
Chat
Manage Message
Manage Bulletin
Manage Forum
Fig. 5. Cancer Community Module Table 1. List of Functionalities for Cancer Community Sub Module
Sub Module Chat Manage Message Manage Bulletin Manage Forum
Functionalities • User can make text-based conversation with other users by sending the message through the chat room. • User can compose a message and send it to a particular user. • User can reply the incoming message. • User can search and view the info that posted on the bulletin board. • User can manage own application for bulletin board. • User can enter other users’ forum post and post message on it. • User can create own post through the community forum.
Design and Development of a Ubiquitous Cancer Care System
121
4.1.2.2 Administrator Module. Users of this module are personnel who manage the content of the CCS. This module consists of five sub modules; Manage User Account, Manage Bulletin, Manage Chat List, Message, and Manage Forum. Figure 6 illustrates the modules. Each sub module for the Administrator contains several functionalities and operations as listed in Table 2.
Administrator
Manage User Account
Manage Chat List
Manage Bulletin
Manage Forum
Message
Fig. 6. Administrator Module Table 2. List of Functionalities for Administrator Sub Module
Sub Module Functionalities Manage User • System administrator can create user account for cancer Account community members. • System administrator can block the status of a particular user. • System administrator can drop a particular user account form the system database. Manage • System administrator can manage all the bulletin applications Bulletin that are sent by user. Manage Chat • System administrator can view the conversation records in the List chat room. • System administrator can clear the chat list. Message • System administrator can compose message and send it to any particular user. Manage • System administrator can enter other users’ forum post and post Forum message on it. • System administrator can drop a particular post from the forum. • System administrator can create own post through the community forum.
122
A.A. Mutalib and N. Hashim
4.1.3 Database Design Figure 7 illustrates the database design of CCS. It contains six tables in which each table includes a number of important information that hold values to represent the main data stored in the database. Besides that, it also shows the relationships between tables as well as the degree of each relationship.
manage
admin_tb
user_tb
1..*
adminName email adminId (PK) password aKey status 1..1
1..1
1..1
1..*
clear send 1..*
1..*
1..1 manage
manage
1..1
chatMessage_tb chatMsgId (PK) Msg sendBy (FK) sendDT
1..1 manage
send
userName icNo gender dob category cancerType email userId (PK) password aKey status regisDT
1..1
1..1 manage 1..*
message_tb msgId (PK) msgTitle msg sendBy (FK) receive By (FK) sendDT 1..*
1..*
1..*
bulletin_tb bulletinId (PK) postDT expDate bulletinTitle bulletinInfo postBy (FK) updateDT status
1..*
forum_tb forumId(PK) forumTitle forumMsg postBy (FK) postDT updateBy (FK) updateDT status
manage
1..*
* PK = Primary Key FK = Foreign Key
Fig. 7. Database Design
4.1.4 UML Design Three diagrams (Figures 8, 9, and 10) represent the classes in CCS, as well as use cases for the cancer community module and administrator module.
Design and Development of a Ubiquitous Cancer Care System
User userName icNo gender dob category cancerType email userId password aKey status regisDT getLoginInfo() 1..* 1
LoginControl verifyLoginInfo() postLoginInfo() 1 1..*
LoginInterface
Message
ChatMessage
msgId msgTitle msg sendBy receieveBy sendDT
chatMsgId msg sendBy sendDT getChatRecords() storeChatInfo()
getMsgRecords() storeMsgInfo() storeRepliedMsgInfo()
1..* 1
ChatControl
1..* 1
MessageControl
Bulletin bulletinId postDT expDate bulletinTitle bulletinInfo postBy updateDT status
forumId forumTitle forumMsg postBy postDT updateBy updaetDT status
getBulletinRecords() storeBulletinInfo() updateBulletinInfo()
getForumRecords() storePostinfo() storeRepliedPostInfo()
1..* 1
retrieveChaList() chat() 1
retrieveMsg() sendMsg() replyMsg()
1..*
ChatInterface
1 1..*
BulletinControl searchBulletin() createBulletin() editBulletin()
displayChatList() displayNotice()
MessageInterface
1 1..*
displayMsg() displayNotice()
Forum
1..* 1
ForumControl retrieveForumList() createPost() replyPost() 1 1..*
BulletinInterface
ForumInterface
displayBulletinInfo() displayNotice()
displayForumList() displayNotice()
MainPage display()
Fig. 8. Class diagram
123
124
A.A. Mutalib and N. Hashim
Cancer Care System
Chat
Send Message
Manage Message
Reply Message
User Manage Bulletin
View Bulletin
Login
Create Bulletin
Manage Forum
Edit Bulletin
Reply Post
Create Post
Fig. 9. Class diagram – cancer community module Cancer Care System
Clear Chat Lis t
Send Mes s age
Admin
Manage Bulletin View Bulletin
Login Create Bulletin
Manage Forum
Edit Bulletin
Create Pos t
Reply Post
Delete Pos t
Manage Us er
Edit Us er Records Delete Us er Records
Fig. 10. Class diagram – administrator module
Design and Development of a Ubiquitous Cancer Care System
125
4.2 The CCS Based on the design as explained in the previous section the CCS was developed. In this section, the CCS is explained, in which the cancer community module is explained first followed with administrator module. 4.2.1 Cancer Community Module Figure 11 shows the login page of the CCS. The CCS only allows registered users to access. Therefore, for new users, they must first register (Figure 12). Once the user has successfully registered, he or she is allowed to login into the CCS by using registered user ID and password. If the combination is not valid, error message will appear as illustrated in Figure 13.
Fig. 11. The Cancer Care System Login Main Page
Fig. 12. User’s Registration Form
Fig. 13. Error Message if Login Process is Not Permitted
After a user has successfully logged into the CCS, it will display the user home page as shown in Figure 14. As mentioned earlier, the cancer community module is incorporated with several features which are chat room, message box, bulletin board, community forum, member directory, community links, and change password. Each
126
A.A. Mutalib and N. Hashim
feature has its own functionalities and operations. There are two additional features which are ‘ask expert’ and ‘ask parent’. User can compose message and send it to all the users that included in either parent user group or medical practitioners group. Through the chat room as illustrated in Figure 15, users are able to make text-based conversations with other users by sending the messages to others.
Fig. 14. User Home Page
Fig. 15. Chat Room Page
In addition, users are also able to compose messages and send them to a particular registered user as shown in Figure 16 and the message recipient can reply the message as can be seen in Figure 17.
Design and Development of a Ubiquitous Cancer Care System
Fig. 16. Compose Message
127
Fig. 17. Reply Message
Figure 18 shows the main screen for the bulletin board, in which users are able to view more detailed about a particular approved bulletin by searching it with the provided bulletin ID that are shown on the bulletin board.
Fig. 18. Bulletin Board Page
128
A.A. Mutalib and N. Hashim
Also, users can manage their own bulletin records such as creating a new entry and waiting for approval from the system administrator in order to post the bulletin onto the CCS. They can also edit the entry which is still ‘pending’ or when it is rejected by the system administrator. Users are able to make changes on the rejected bulletin and send it back into the CCS in order to get the approval (Figure 19). Next, Figure 20 shows the screen for Community Forum. Users are able to establish a forum site in this community forum and the posted forum will be listed on the page after users have sent all the required information.
Fig. 19. User Manage Bulletin
Fig. 20. Community Forum Page
Besides that, users can access the listed forum site by clicking on the ‘Enter’ button at the end of the list. After that, users will be navigated to the selected forum site and user can post message on it as seen in Figure 21. For Member Directory as illustrated in Figure 22, users can make an advanced search by keyword using the combination of name, user ID, status, or cancer type. If the search query matches with any user records in the database, the results list will be displayed as shown in Figure 23. Users can explore the details of that particular member by clicking on the “Detail’ button on the list. Completed personal information will appear as in Figure 24.
Design and Development of a Ubiquitous Cancer Care System
Fig. 21. User Post Message in a Particular Forum Site
Fig. 22. Member Directory Page
Fig. 23. The Searching result
Fig. 24. User’s Detail
129
130
A.A. Mutalib and N. Hashim
Figure 25 depicts the screen for Community Links that provides links to other related web sites. Users can navigate to the listed cancer community sites on the World Wide Web (WWW) by clicking on the provided links. Registered user also can make changes on her/his recent password from time to time for security purpose (see Figure 26).
Fig. 25. Community Links Page
Fig. 26. Change Password Page
4.2.2 Administrator Module Figure 27 shows the administrator login page which is separated from the user login page and the main user of this module is system administrator. Features included in this module includes manage user account, manage bulletin, manage chat list, message, and manage forum.
Fig. 27. Admin Login Page
After the administrators log into the system, they are able to manage all user accounts which create a new user account for cancer community member, can block a particular user’s account status in order to prevent the user from logging into the system, and drop user records from the database. Besides that, administrators can also insert an authentication key (AKey) for security purpose. This Akey will be informed to the user when he/she forgets the password to log into the system. The user will insert the AKey as well as user ID in order to retrieve the password.
Design and Development of a Ubiquitous Cancer Care System
131
Figure 28 shows the screen for Manage Bulletin Board. In the main page of the manage bulletin board features, the administrators can view all the records of bulletin board which are in ‘approved’ and ‘pending’ status. Administrator can access the particular pending application created by users and decide the status of the application. The status varies among approved, rejected, or obsolete. Only the approved bulletin will be posted onto the bulletin board.
Fig. 28. Manage Bulletin Board
In addition, in order to manage all operations in the chat room, the administrator can view all the conversation records of the chat room and are also able to clear the chat list manually at anytime in order to prevent overloading in the database as shown in Figure 29.
Fig. 29. Manage Chat List
The system administrators are also allowed to compose messages and send them to any particular registered user. This type of message is just for notification purpose, so the user who receives this message is not able to reply on it as shown in Figure 30.
132
A.A. Mutalib and N. Hashim
Fig. 30. Message Page
Figure 31 visualizes the screen for Manage Community Forum. The system administrator is able to create a new forum site, post message on users’ forum sites, and can drop unwanted forum site from the list of community forum.
Fig. 31. Manage Community Forum
5 Conclusion and Discussion Based on the research objectives that are stated earlier, this paper aims at (1) determining the components of CCS, and (2) designing and developing a prototype of CCS. Both have been explained extensively in previous sections. Initial feedbacks from the cancer community (6 patients, 2 parents, and 2 medical practitioners) have been obtained. Visits to University Malaya Medical Center were very successful because the patients, parents, and medical practitioners were very cooperative. When experiencing the CCS, they were observed and informally interviewed. This study found that the patients were excited using the system, especially those who are familiar with computer systems. In fact, they were also excited telling their friends about the proposed system, demonstrating the CCS to other patients. In relation, studies such as by Mutalib and Shiratuddin [20] and
Design and Development of a Ubiquitous Cancer Care System
133
Mutalib, Alwi, and Salam [21] have proven that children have been using computers in their daily life. Parents were found happy with the idea of the system. Thorough the prototype, they got the ideas on how they could communicate among each other, with their children, and more importantly with the practitioners at any time anywhere. It is important to acknowledge that cancer treatments are very painful, and stressful. Not only patients feel them, but also the caregivers, which in most cases are the parents. Furthermore, talking to normal people is not advisable because they are not able to understand the pain. Hence, the idea of the CCS that connects the cancer community is very helpful for the community to discuss special matter regarding their children’s treatment and progress. On the other hand, the medical practitioners when experiencing the prototype found that it will help them a lot in following cases under their observation. In normal practice, they have to make phone calls, which are costly, and sometimes are not answered. Also, parents from remote areas can refer to them using the system without having to travel. This is particularly important because hospitals that provide cancer treatments are limited (only in major cities), which for most people are quite ‘difficult’ to access under financial and time constraints. Hence, in a nutshell, the idea of CCS is very positive among the cancer community, in supports of their collaborative communication. In response to making communication among the community ubiquitous, the idea of CCS is clearly inline with. The system is always ready for access at anytime from anywhere. The availability of the CCS is at par with the community’s ability to effort for the technologies. In regards to this, a big percentage of parents appeared in the ward during the visits have personal laptops with them. This supports the needs for deploying the CCS, which entirely runs only on computers without any extra equipment. For future development and expansion of this study, the CCS should be installed and tested in an actual web server which is connected to the network. Besides that, the consistency and reliability of the system should be emphasized in order to ensure the users are able to access the system properly. Then, actual systematic testing on the system in terms of usefulness and ease of use particularly on the interface [22] could be carried out. It should involve bigger samples for patients, parents, and medical practitioners.
References [1] Girgensohn, A., Lee, A.: Making Web Sites be Places for Social Interaction. In: CSCW 2002: Proceedings of the 2002 ACM Conference on Computer Supported Cooperative Work, pp. 136–145 (2002) [2] John, A., Seligmann, D.D., Adamic, L., Davis, M., Nack, F., Shamma, D.A.: The Future of Online Social Interactions: What to Expect in 2020. In: WWW 2008: Proceeding of the 17th international conference on World Wide Web, pp. 1255–1256 (2008) [3] Brym, R.J., Lie, J.: Sociology: Your Compass for a New World, 3rd edn. Wadsworth Publishing, USA (2007) [4] Walther, J.B., Boyd, S.: Attraction to Computer-Mediated Social Support. In: Lin, C.A., Atkin, D. (eds.) Communication Technology and Society: Audience Adoption and Uses, pp. 153–188. Hampton Press, Cresskill (2002) [5] Fernsler, J.I., Manchester, L.J.: Evaluation of a Computer-Based Cancer Support Network. Cancer Practice 5(1), 46–51 (1997)
134
A.A. Mutalib and N. Hashim
[6] Schweizer, K.J., Krcmar, H., Leimeister, J.M.: The Role of Virtual Communities for the Social Network of Cancer Patients. In: Proceeding of the 12th America Conference on Information System, Acapulco, Mexico, August 04-06 (2006) [7] Johnson, D.G.: Ethics Online: Shaping Social Behavior Online Takes More Than New Laws and Modified Edicts. Communications of the ACM 40(1), 60–65 (1997) [8] Caplan, S.E., Turner, J.S.: Bringing Theory to Research on Computer-Mediated Comforting Communication. Computers in Human Behavior 23(2), 985–998 (2007) [9] Walther, J.B.: Computer-Mediated Communication: Impersonal, Interpersonal, and Hyperpersonal Interaction. Communication Research 23, 3–43 (1996) [10] Caplan, S.E.: Preference for Online Social Interaction: a Theory of Problematic Internet Use and Psychosocial Well-Being. Communication Research 30, 625–648 (2003) [11] Eysenbach, G.: The Impact of the Internet on Cancer Outcomes. CA Cancer J. Clin. 53, 356–371 (2003) [12] Burrows, R., Nettleton, S., Pleave, N., Loader, B., Muncer, S.: Virtual Community Care? Social Policy and the Emergence of Computer-Mediated Social Support. Information, Communication and Society 3, 95–121 (2000) [13] Wright, K.: Social Support within an On-line Cancer Community: an Assessment of Emotional Support, Perceptions of Advantages and Disadvantages, and Motives for using the Community from a Communication Perspective. Journal of Applied Communication Research 30(3), 195–209 (2002) [14] Sundquist, K., Caceres, R.A., White, A., McLean, A., Kennedy, M., Fitz-Gerald, L.: Cancer Connections: Development of an Online Cancer Carers Community. In: Proceeding of the Carers NSW Conference, Novotel, Brighton-le-Sands, Sydney, June 11-12 (2009) [15] Laudon, K.C., Laudon, J.P.: Management Information Systems: Organization and Technology in the Networked Enterprise, 6th edn. Prentice Hall, New Jersey (2000) [16] Wahab, M.H.A., Lee, O., Kadir, H.A., Johari, A., Mutalib, A.A., Mohsin, M.F., Sidek, R.M., Noraziah, A.: WAS-GN: Web-based appointment system with GSM network. International Journal of Advancements in Computing Technology 3(5), 315–333 (2011) [17] Kadir, H.A., Wahab, M.H.A., Tukiran, Z., Mutalib, A.A.: Tracking student movement using active RFID. In: Proceedings of the 9th WSEAS International Conference on Applications of Computer Engineering, ACE 2010, pp. 41–45 (2010) [18] Wahab, M.H.A., Kadir, H.A., Tukiran, Z., Tomari, M.R., Mutalib, A.A., Mohsin, M.F.M., Idrus, M.N.E.M.: Web-based laboratory equipment monitoring system using RFID. In: Proceedings on The 2010 International Conference on Intelligent and Advanced Systems, ICIAS 2010, art. no. 5716177 (2010) [19] Wahab, M.H.A., Mutalib, A.A., Kadir, H.A., Mohsin, M.F.M.: Design and development of portable RFID for attendance system. In: Proceedings on The 2010 International Conference on Information Retrieval and Knowledge Management: Exploring the Invisible World, CAMP 2010, pp. 173–178, art. no. 5466925 (2010a) [20] Mutalib, A.A., Shiratuddin, N.: Electronic learning media (ELM): Reality learning media (RLM) vs video. Proceedings of World Academy of Science, Engineering and Technology 38, 1419–1426 (2009) [21] Mutalib, A.A., Alwi, A., Salam, S.N.A.: The acceptance of E-book readers among Malaysian children. In: Proceedings of 2006 International Conference on Computing and Informatics, ICOCI, art. no. 5276513 (2006) [22] Hix, D., Hartson, H.R.: Developing User Interface: Ensuring Usability through Product and Process. John Wiley & Son, Inc., New York (1993)
ICA-WSS: A QOS-Aware Imperialist Competitive Algorithm for Web Service Selection Shahriar Lotfi1 and Khalil Mowlani2 1
Department Of Computer Science, University Of Tabriz, Tabriz, Iran
[email protected] 2 Department Of Computer Science, University Of Tabriz, Tabriz, Iran
[email protected]
Abstract. One of the most important works in technology of web service is web service composition. In the recent years numerous studies to develop methods to build composite web services is carried out but making composite web service yet is a complex issue. In this paper a QOS-aware method based on imperialist competition algorithm for web services selection is proposed. In this method called ICA-WSS, web service selection is done based on QOS parameters. ICA-WSS takes an abstract process and a set of all candidate web services and tries to find an optimal combination based on QOS parameters. In comparison to QQDSGA without considering QOE parameters, ICA-WSS finds better combination in term of value of objective function. Experimental results endorse that ICA-WSS is an efficient algorithm with good convergence for solving problem of web service composition. Keywords: web service, web service selection, web service composition, imperialist competitive algorithm.
1 Introduction In recent years, rapid development of using Internet and its undeniable influence in the field of trade and business, has led to web services become very good choice for make business application. Environment of the Web is a suitable place for make distributed applications and Web services can be used as appropriate entities. Web service is a software system is defined for machine-machine interactive on a network [1]. Web services can communicate with each other through http protocol and using SOAP messages. In fact, a web service is a software system specified by a URL and its public interfaces and binding’s methods are described by XML documents [2]. There are more details about the basic concepts of web services technology in [2, 3]. Because Web services use XML to exchange messages, do not belong to any particular platform and this is a great advantage that allows creating large-scale distributed software systems. Web services provide sharing functionality among different providers and allow them to build heterogeneous systems using messagebased communication [4]. There are some problems on the use of web services. The A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 135–148, 2011. © Springer-Verlag Berlin Heidelberg 2011
136
S. Lotfi and K. Mowlani
one of most important problem is limitation of web service’s functionality. A web service is a small and has little ability and not adequate to respond to a complicated request. When request is complicate, several web services should be used to respond to the request. Putting together several Web services and integrate them to one unit work, called web service composition and new integrated web service is called composite web service. A composite service has specific functions that can be divided into some component functions. These component functions can be accomplished by some component services respectively [5]. In this paper a QOS-aware algorithm proposed for Web services selection. Our algorithm based on imperialist competitive algorithm. Selecting Web services will be based on QOS criteria. The composite web service should be optimal in term of QOS parameters. This paper consists of 8 sections. After introduction in first section, in second section, the problem of web service composition is explained. Section 3 includes comments about imperialist competitive algorithm. Section 4 gives related work about web service selection and composition. In section 5, the problem of web service composition is formulated. Section 6 introduces proposed method. Section 7 gives simulation results and experiments. Finally section 8 includes conclusion and future works.
2 Web Service Composition Problem Process of making composite web service includes four steps. In first step, the user’s request is broken to several tasks so that each task can be performed by a single service. Tasks and relations between them form an abstract process. Each abstract process is relevant to a workflow. In second step, finding appropriate web services for each task is done. This step is called web service discovery. In the end of web service discovery, each task has a set of candidate web services. In third step, selection web service among candidate web services is done. In this step, the best candidate web service for each task is selected and composite web service is made. Finally, in fourth step, composite web service is executed. A web service generally includes two types of parameters; functional parameters and nonfunctional parameters. The functional parameters are used only in the web service discovery. All candidate web services for specific task have the same functional parameters. Non-functional parameters are used in web service selection step. In selection step, a web service with better non-functional parameters is chosen. Non-functional parameters are quality of service standards that vary for different services. There are several non-functional parameters for Web services such as cost, availability, response time and so on. There are more details about the non-functional parameters in [6]. Despite of all efforts, the construction of a composite web services has remained as very complex process and may not be performed manually [7]. There are some problems about web service composition in [8].The problem’s space of web service composition is exponential and making a composite web service requires a lot of calculations. So web service composition is a NP-Hard problem [9, 10] and requires a lot of time.
ICA-WSS: A QOS-Aware Imperialist Competitive Algorithm for Web Service Selection
137
3 Imperialist Competitive Algorithm Imperialist competitive algorithm is new type of evolutionary algorithms based on political-social process of imperialism phenomenon. Unlike other evolutionary algorithms such as GA, PSO and SA, which are inspired from natural evolution, ICA inspired from society’s improvements. In brief, algorithm begins with Npop random initial solutions. Each solution is called a country. Countries can be considered as a chromosome in GA algorithm or particle in the PSO algorithm. Nimp of the best countries are selected as the imperialists. Other countries (Ncol) are considered as a colony. Based on imperialist power, each imperialist takes specific number of colonies at random and then based on specific mechanism, colonies move toward their relevant imperialists. With formation of initial empires, competition between them is begun. Gradually, the weak imperialists losses their ability to compete and will be removed from the imperial competition. The end of algorithm, only one empire will remain. All colonies will be under the control of this unique empire. The last empire is the optimal solution. There are more details about ICA in [11]. Since ICA is used for solving various optimization problems that has led to good results. For example in [12, 13 and 14] ICA is used.
4 Related Works In recent years, many algorithms based on integer programming (IP), mixed integer linear programming (MILP), multi-dimension multi-choice 0–1 knapsack problem (MMKP), Markov decision programming (MDP), genetic algorithm (GA), particle swarm optimization (PSO) and so on, have been presented to solve web service composition problem [15]. Howard and Kerschberg In [16] has proposed method based on semantic description of Web services to build composite services. Their proposed method does not consider QOS parameters and making of composite web service only rely on semantic. In [17], a linear logic-based method for automatic composition of web services has been used. In this proposed method, semantic descriptions based on OWL translate to logical expressions then logical expressions are used to construct a composite service. In [18], an IP-based method for select web service for service composition has been proposed. In [18], computational complexity is exponential and the scalability of the method is low. In [5], an improved genetic algorithm has been proposed that consider QOS parameters for selecting web services. This method called CoDiGA, improves value of fitness function for optimal solution rather than normal genetic algorithm. The disadvantage of this method is there is no mechanism for adaptability to different conditions. In [19], Peng, Liu and Hu have proposed a multi objective QOS-aware method for selecting web service based on ant colony algorithm. Their proposed method that calls MOACO, only considers sequential work flow of abstract description of composite web service. Researchers in [20] have proposed an algorithm based on QOS and QOE parameters. This algorithm that briefly called QQDSGA, is hybrid of genetic algorithm and simulated annealing algorithm. QOE parameters consider to feedback user’s satisfactions. However the semantic descriptions of web services are omitted in this algorithm. In [21], a heuristic algorithm has been proposed for making composite web
138
S. Lotfi and K. Mowlani
services. This method is QOS- based. Authors claimed that their method is better than HMCOP [10, 24]. Evaluation of a composite service in this case is performed easily. This simplicity caused some disadvantage such as loss of flexibility of the final composite service. In [22] a genetic algorithm is proposed that use rough set theory to improve performance of standard genetic algorithm. The disadvantage of this method is that users can not set their own parameters and determine their level of satisfaction to the whole of composition process. In [23] a penalty-based genetic algorithm for making a composite service that is optimized based on the QOS parameters has been used. Dependencies and collisions between web services are considered. Non-feasible solutions do not remove suddenly. This method like [22] does not consider user constraints.
5 Problem Formulations In this section, first, some related concepts and symbols are defined. Then, QOS parameters are formulated. Finally, an evaluate function for composite web service is proposed. 5.1 Symbols Table 1 shows symbols and their definitions that are used in this study. Table 1. Symbols and Their Definitions Symbol WS t CWS ABSP ALLCWSS COWS
Definition a concrete web service a task that is performed by a concrete web service a candidate web service for a specific task an abstract process a set that contains all candidate web service for all tasks a composite web service that includes several WS
5.2 QOS Formulation There is currently no standardized description framework to include all aspects of web service’s non-functional characteristics [25]. For modeling QOS parameters for web service, combination of researches in [9, 20, 21, 25, 26] is used. Table 2 shows selected QOS parameters in this paper. Web service selection based on QOS parameters is a multi objective problem. In multi objective problems often there are some inconsistencies among parameters. For avoid these inconsistencies, QOS parameters is normalized. For normalizing QOS parameters, combinational information of [20, 21, 27] is used. Main focus is on [22]. because in section 7, ICA-WSS is compared to [20]. (1) And (2) are used for normalize QOS parameters.
ICA-WSS: A QOS-Aware Imperialist Competitive Algorithm for Web Service Selection
139
Table 2. QOS Parameters that are used in This Paper Parameter’s Name
Symbol
Definition and Attributes
Execution cost
price (WSi)
Response time
time (WSi)
availability
av (WSi)
Reliability
rl (WSi)
represents the money or cost that a service requester has to pay for invoking WSi [20] the time required for invoke service WSi and execute it is the degree to which service WSi is operational and accessible when it is required for use [25] refers to the service provider’s ability to successfully deliver requested service functionality [25]
Nx(WSi)
(1)
Nx(WSi)
(2)
is maximum value for x , x is minimum value for x, x ws is value of x parameter x for web service i, Nx(wsi) is normalized value of parameter x for web service i. (1) is used for normalize cost-based parameters such as time and price but (2) is used for benefit-based parameters such as availability and reliability parameters. A workflow is a graph is relevant to an abstract process and consists of all tasks and how the tasks communicate to each other. Workflow includes four types of communications [9, 18, 26]. These four types are: sequential, parallel, switch and cycle. For evaluate QOS parameters for a workflow, QOS parameters of all web services should be combined. Evaluation of QOS for workflow in all types of communications is shown in table 3and table 4. This evaluation extracted from [9, 19, 21, 26]. Table 3. Evaluation of Cost-based QOS Parameters for Workflow in Different Structures communication type
allprice
alltime
Sequence Switch Parallel Cycle
max
|
1 ,if performs c times
,if c times
performs
140
S. Lotfi and K. Mowlani
Table 4. Evaluation of Benefit-based QOS Parameters for Workflow in Different Structures communication type
allav
allrl
Sequence Switch Parallel ,if
Cycle
perform c
times
,if
perform c
times
5.3 Evaluation of Composite Web Service Our proposed method is a maximization problem. For evaluation composite web service, each QOS parameters take a weight based on its importance. Objective function for evaluate a composite web service is defined as (3). F2obj1= w1*allprice+w2*alltime+w3*allav+w4*allr w1+w2+w3+w4=1.
(3)
In (3), QOE parameters have not been considered in objective function. Objective function that is presented in (4), considers both QOS and QOE parameters. Consideration of QOE parameters same as [20]. F1obj=p* F1obj +q*QOE, p+q=1
(4)
Value of QOE parameters same as [20]: excellent=0.8, good=0.7, average=0.6, bad=0.5 and terrible=0.4.
6 ICA-WSS Selecting web service for making composite web service is a discrete problem. In this section we explain our method. Difference between ICA-WSS and standard ICA [11], is in implementation of assimilation operator. 6.1 Initialization In this step, first, parameters of algorithm are specified. then, initial countries are created. Each country is considered as 1 n array. Each cell of array relevant to a task so that index of the cell is task’s number and value of the cell is web service’s number that selected to perform task. Figure 1 shows details about coding of countries. The web service’s number is an integer number that obtained from combination of index of its row and its columns in ALLCWSS matrix. For example if a service is in 3th row and 4th column of ALLCWSS, its number will be 34.
ICA-WSS: A QOS-Aware Imperialist Competitive Algorithm for Web Service Selection
Task 2
Task 1
11
22
141
Task n
….
nm
(a) 11
21
…
n1
12
22
…
n2
2m
…
…
…
… 1m
…
nm
(b) Fig. 1. Coding a Country a) Final Presentation of a Country and b) ALLCOWSS Set
Because ICA based on cost, after creating Npop initial random countries, cost of all countries is obtained. Cost of country is obtained by (5). cost(countryi)=MAXobj-Fobj(countryi)
(5)
MAXobj is a maximum value that objective function can take. Nimp of best countries (countries with less cost) are selected as initial imperialists. Then normalized cost for each country is calculated by (6). Other countries considered as colony. C =MAXcost – cost(Impi)
(6)
MAXcost is a maximum cost of all imperialists and Ci is normalized cost of ith imperialist. For specify number of colonies of each imperialist, first, normalized relative power of imperialist calculated by (7). Then, the number of colonies for imperialists are obtained by (8). Pi=
C N
∑
C
(7)
Pi is normalized relative power of ith imperialist. NCoi=round(Pi * Ncol)
(8)
NCoi is number of colonies that ith imperialist possess and round(k) give larger integer less and equal than k. Each imperialist take its colonies in random manner. After this, initial imperialist formed and evolution begins.
142
S. Lotfi and K. Mowlani
6.2 Assimilation For implement an assimilation operator for this discrete problem, first, a 1 n mask array is produced. The value of array’s cells selected from range 11(least number among web service’ numbers) to nm (maximum number among web service’s number) in random manner. Then, for each colony, value of all cells is compared to relevant cells in produced mask array. If value in ith cell of colony less than value of ith cell in mask array, value of ith cell in relevant imperialist replaces with value of ith cell in colony. Figure 2 shows an assimilation example. imperialist:
15
24
33
47
colony:
18
12
54
43
mask:
16
19
33
49
result:
18
24
54
47
Fig. 2. An Example for Showing Assimilation Operator
6.3 Exchange the Position of Imperialist and Colony After applying assimilation operator, some colonies maybe become better from their imperialist. In this case, the position of imperialist and colony is exchanged. 6.4 Calculate Total Cost and Total Power of Imperialists Total power of an imperialist is affected from number of its colony [11]. First, total cost of an imperialist is calculated by (9). Then, normalized total power of an imperialist is obtained by (10). TOTCi= cost(Impi)+ £ mean{cost(colonies of ith imperialist)}
(9)
£ is positive number less than 1. NTOTPi= TOTCmax - TOTCi
(10)
TOTCmax is maximum of total costs of all imperialists. 6.5 Imperialistic Competition Imperialistic competition in our proposed method exactly same as [11]. In each step, the weakest colony in weakest imperialist is selected and other imperialists compete for possess this colony. Possession probability is obtained by (11).
ICA-WSS: A QOS-Aware Imperialist Competitive Algorithm for Web Service Selection
PPi=
NTOTC N
∑
NTOTC
143
(11)
After obtaining all Possession probabilities, array P is made as: P=[ PP1, PP2, …, PPk], k= Nimp. Then, an array R with same dominion as P is made like: R= [R1, R2,…Rk] , k= Nimp & Ri ~ U(0,1). Then, array D from subtract of P and R is made. Then the index of lager value in D is selected. For example if this index is i, the ith imperialist possess the colony. 6.6 Revolution Revolution operator doesn’t exist in [11]. Revolution operator in some case helps for escape from local minimum. Revolution operator acts as follows. Per each country, a random number r between [0 1] is produced. If r greater than revolution rate (rr), a task in the country is selected randomly. Then, the selected web service for this task is exchanged with another web service in related candidate web service set. 6.7 Eliminate Imperialist In this step, imperialist with no colonies eliminate from imperialist’s set. Figure 3 shows the steps of proposed algorithm. When only one imperialist has remained, algorithm stops and this imperialist is returned as solution. ISA-WSS Algorithm
input: ABSP & ALLCWSS & m output: COWS 1- initialize: 1-1- set the number of population (Npop), number of imperialist (Nimp), number of colonies(Ncol) revolution rate (rr) and £ 1-2- Initialize population: randomly generate Npop initial countries. //each country relevant to a composite service 1-3- evaluate the objective function for each country //cost(country)=MAX(obj.fun)-obj.fun(country) 1-4- select Nimp of the best countries as initial imperialist. The other countries will be the colonies while(num of imperialist >1) { 2- Move colonies toward their relevant imperialist 3- Exchanging position of imperialist and the best colonies 4- Compute the total power of each empire 5- Imperialistic competition 6- Revolution 7- Eliminate empires with no colonies. } 8- return last Imperialist as solution Fig. 3. The Steps of ICA-WSS
144
S. Lotfi and K. Mowlani
7 Experiments and Results This section includes experiments and evaluation of proposed algorithm. The simulation is performed using java jdk1.5 in Intellij IDEA environment and matlab 7.11.0. Simulation is performed in a personal computer with: intel. 2 GHz Dual Core cpu with 1Mb cache, 2G ram DDR2. Experiments are in two types. In first type, ICA-WSS evaluated individually and second type, includes comparison. Comparison is performed with QQDSGA that is proposed in [20]. For all experiments, value of QOS parameters are selected from range [0 1]. So, value of objective function is between 0 and 1. First experiment studies the convergence of ICA-WSS. For testing convergence, average of 10 different runs is considered. Figure 4 shows the results of this experiment. The x axis shows the iteration’s number. The y axis shows the value of optimal solution in each iteration. Value of parameters is: Npop=200, Nimp=30, Ncol =17 rr=0.02 and £= 0.05. Also: number of tasks (n)=10 and number of candidate web service for each task (m)=30.
Fig. 4. Convergence of ICA-WSS
Figure 4 shows that ICA-WSS has a good convergence and has good improvement in value of optimal solution from starting of algorithm to end. Second experiment tests the influence of number of imperialist (Nimp) on optimal solution. For perform this, first ICA-WSS is executed with 5 imperialists. In each next step, number of imperialists is increased by 1. This experiment is continued until Nimp =50. In each execution value of parameters was: Npop =200, rr=0.02, £= 0.05, n=10 and m=30. Figure 5 shows the results of this step. Best result is obtained when number of imperialists is about 30. So number of selected imperialist is important and influences the overall results. The third experiment tests the stability of proposed method. For this, value of optimal solution in 50 different cases is considered. Each case is average results of 5 different runs (overall 250 different executions). All parameters like previous experiments. Npop =200, Nimp=30, rr=0.02, £= 0.05, n=10 and m=30. Results of this test are shown in figure 6. The variance of objective function in 50 different cases is low. It endorse stability of ICA-WSS is good.
ICA-WSS: A QOS-Aware Imperialist Competitive Algorithm for Web Service Selection
145
Fig. 5. Effect of number of IMPs on optimal solution
Fig. 6. Stability of proposed method
After those three experiments, here, comparison of ICA-WSS with QQDSGA is performed. QQDSGA is a combination of two algorithms GA and SA that consider both QOS and QOE parameters. Results in [20] shows that QQDSGA is better than both SA and GA. In [20], Comparisons performed in two category. In first category, only QOS parameters are considered but in second category, both QOS and QOE parameters are considered. Our Comparisons is in first category. We don’t consider QOE parameters in Comparisons. For performing comparison, impact of deferent numbers of candidate web services (m) on objective function is analyzed. Like [20], experiment begin with m=5 and increase m by 5 in next case (m=5, 10, 15, …., 50). Last run include 50 candidate web service per each task (m=50). The value of other parameters considered like previous experiments (Npop =200, Nimp=30, rr=0.02, £= 0.05, n=10). Figure 7 shows this comparison. ICA-WSS improve quality of optimal solution. Figure 7 shows that best solution in ICA-WSS has objective function about 0.74 that better than best solution in SGA (QQSGA without QOE). In order to
146
S. Lotfi and K. Mowlani
highlight comparison, average improvement in objective function is considered. It calculated by (12). AvgImpv
N
C
O f
O
f
N
C
(12)
is last objective function (with 50 candidate WS), Obj is first fitness Objf is number of candidate web service in function (with 5 candidate WS), NumCondf last execution(=50) and NumCond is number of candidate web service in first execution (=5). Table 5 shows numerical detailed of AvgImpv. Because of raw data of QQDSGA is not accessible, we approximate them from figure 7-a.
(a)
(b) Fig. 7. Results of Comparison a) Effect of number of candidate web service on optimal solution in QQDSGA[20] b) Effect of number of candidate web service on optimal solution in ICAWSS
ICA-WSS: A QOS-Aware Imperialist Competitive Algorithm for Web Service Selection
147
Table 5. AvgImpv in ICA-WSS and QQDSGA Algorithm’s name QQDSGA
Value of AvgImpv
ICA-WSS
0.7391 0.6317 50 5
0.67 0.63 50 5
0.0008889 0.0024
It clear that AvgImpv in ICA-WSS is greater than QQDSGA. So ICA-WSS has better evolution process and can find optimal solution with better quality rather than QQDSGA.
8 Conclusions and Future Works In this study, a QOS-based Imperialist competitive algorithm for selecting web service and making composite web service is proposed. The core of our proposed method is standard Imperialist competitive algorithm that its assimilation operator is changed according to discrete environment of web service composition. Experiments show that proposed method that called ICA-WSS is an efficient algorithm. Comparison shows that ICA-WSS can find better composition in term of objective function rather than QQDSGA when QOE parameters not considered. We didn’t carry out any comparison with QOE parameters. In future our focus will be on: 1) propose an efficient way to using user constraint in selecting web service and 2) build a framework for web service composition based on ICA-WSS.
References 1. Booth, D., Haas, H., McCabe, F., Newcomer, E., Champion, M., Ferris, C., Orchard, D.: Web Services Architecture, W3C Working Group Note 11, W3C Technical Reports and Publications (February 2004), http://www.w3.org/TR/ws-arch/ 2. Dustdar, S., Schreiner, W.: A Survey on Web Services Composition. J. Web and Grid Services 1(1), 1–30 (2005) 3. Karakoc, E.: Web Service Composition under Resource Allocation Constraints. A Thesis Submitted to the Graduate School of Natural and Applied Sciences of Middle East Technical University for the Degree of Master of Science in Computer Engineering (2007) 4. Li, S.H., Huang, S.M., Yen, D.C., Chang, C.C.: Chang, Migrating Legacy Information Systems to Web Services Architecture. J. Database Management 18(4), 1–25 (2007) 5. Ma, Y., Zhang, C.: Quick Convergence of Genetic Algorithm for QOS-Driven Web Service Selection. J. Computer Networks 52, 1093–1104 (2008) 6. ITU-T Recommendation E.800, Terms and Definitions Related to Quality of Service and Network Performance Including Dependability (1994) 7. Rao, J., Su, X.: A Survey of Automated Web Service Composition Methods. In: Cardoso, J., Sheth, A.P. (eds.) SWSWPC 2004. LNCS, vol. 3387, pp. 43–54. Springer, Heidelberg (2005) 8. Karakoc, E., Senkul, P.: Composing Semantic Web Services under Constraints. J. Expert Systems with Applications 36, 11021–11029 (2009) 9. Canfora, G., Penta, M.D., Esposito, R., Villani, M.L.: A Lightweight Approach for QoS–aware Service Composition. In: Proceedings of the 2nd International Conference on Service Oriented Computing (ICSOC 2004), New York, pp. 36–47 (2004)
148
S. Lotfi and K. Mowlani
10. Yu, T., Zhang, Y., Lin, K.: Efficient Algorithms for Web services Selection with End-toEnd QoS Constraints. J. ACM Transactions on the Web 1(1), 1–26 (2007) 11. Atashpaz-Gargari, E., Lucase, C.: Imperialist Competitive Algorithm: An Algorithm for Optimization Inspired by Imperialistic Competition. In: IEEE Congress on Evolutionary Computation (CEC), Singapore, pp. 4661–4667 (2007) 12. Mozafari, H., Abdi, B., Ayob, A.: Optimization of Transmission Conditions for Thin Interphase Layer Based on Imperialist Competitive Algorithm. Int. J. Computer Science and Engineering 2(7), 2486–2490 (2010) 13. Niknama, T., Fard, E.T., Pourjafarian, N., Rousta, A.: An Efficient Hybrid Algorithm based on Modified Imperialist Competitive Algorithm and K-means for Data Clustering. J. Engineering Applications of Artificial Intelligence 24, 306–317 (2011) 14. Jain, T., Nigam, M.J.: Synergy of Evolutionary Algorithm and Socio-Political Process for Global Optimization. J. Expert Systems with Applications 37, 3706–3713 (2010) 15. Fana, X.Q., Fang, X.N., Jiangb, C.J.: Research on Web Service Selection Based on Cooperative Evolution. J. Expert Systems with Applications 38, 9736–9743 (2011) 16. Howard, R., Kerschberg, L.: A Knowledge-Based Framework for Dynamic Semantic Web Services Brokering and Management. In: Proceedings of 15th International Workshop on Database and Expert Systems Applications (DEXA), Zaragoza, Spain, pp. 174–178 (2004) 17. Rao, J., Kungas, P., Matskin, M.: Logic-Based Web Services Composition: From service Description to Process Model. In: Proceeding on IEEE International Conference on Web Service (ICWS), USA, pp. 446–453 (2004) 18. Zeng, L., Benatallah, B., Ngu, A.H., Dumas, M., Kalagnanam, J., Chang, H.: QoS-Aware Middleware for Web Services Composition. IEEE Trans. Software Engineering 30(5), 311–327 (2004) 19. Qiqing, F., Xiaoming, P., Qinghua, L., Yahui, H.: A Global QoS Optimizing Web Services Selection Algorithm based on MOACO for Dynamic Web Service Composition. In: International Forum on Information Technology and Applications (IFITA), Chengdu, China, pp. 37–42 (2009) 20. Gao, Z.P., Chen, J., Qiu, X.S., Meng, L.M.: QoE/QoS Driven Simulated Annealing-Based Genetic Algorithm for Web Services Selection. J. China Universities of Posts and Telecommunications 16, 102–107 (2009) 21. Luo, Y.S., Yong, Q., Hou, D., Shen, L.F., Zhong, X.: A Novel Heuristic Algorithm for QoSAware End-to-End Service Composition. J. Computer Communications 34, 1137–1144 (2011) 22. Liang, W.Y., Huang, C.C.: The Generic Genetic Algorithm Incorporates with Rough Set Theory– an Application of the Web Services Composition. J. Expert Systems with Applications 36, 5549–5556 (2009) 23. Ai, L., Tang, M.: A Penalty-based Genetic Algorithm for QoS-Aware Web Service Composition with Inter-Service Dependencies and Conflicts. In: International Conference on Computational Intelligence for Modeling Control & Automation (CIMCA), Vienna, Austria, pp. 738–743 (2008) 24. Yu, T., Lin, K.-J.: Service Selection Algorithms for Composing Complex Services with Multiple QoS Constraints. In: Benatallah, B., Casati, F., Traverso, P. (eds.) ICSOC 2005. LNCS, vol. 3826, pp. 130–143. Springer, Heidelberg (2005) 25. Huang, F.M.A., Lan, C.W., Yang, S.J.H.: An Optimal QoS-Based Web Service Selection Scheme. J. Information Sciences 179, 3309–3322 (2009) 26. Cardoso, J., Sheth, A., Miller, J., Arnold, J., Kochut, K.: Quality of Service for WorkFlows and Web Service Processes. J. Web Semantics 1(3), 281–308 (2004) 27. Wang, Z.J., Liu, Z.Z., Zhou, X.F., Lou, Y.S.: An Approach for Composite Web Service Selection Based on DGQoS. Int. J. Advanced Manufacturing Technology 56, 1–13 (2011)
Flexi-adaptor: A Nobel Approach for Adapting Web Content for Mobile Devices Rajibul Anam, Chin Kuan Ho, and Tek Yong Lim Faculty of Information Technology, Multimedia University, Persiaran Multimedia, 63100 Cyberjaya, Selangor, Malaysia {rajibul.anam08,ckho,tylim}@mmu.edu.my
Abstract. Mobile internet browsing usually involves a lot of horizontal and vertical scrolling, which makes web browsing time-consuming and in addition user may be interested in a section of a web page, which may not fit to the mobile screen. This requires more scrolling in both directions. In this paper, we propose to address this problem by displaying the block title and hide the block body contents from a large web page while maintaining their semantics, which displays only the condensed information. The Flexi-adaptor, reduces unnecessary information by allowing its users to see the most relevant blocks of the page and provides the block title. The Flexi-adaptor categorized each object of the web page as menu, block title and main content. Assign a weight to each object of the block by analyzing the object elements. It uses depth first traversal algorithm to select blocks, and delivers them to handheld devices. The proposed solution improves web content accessibility and delivers the target contents to the users. Keywords: Content adaptation, mobile browsing, small display, adaptive interface for small screens.
1 Inroduction Last few years, people start using the network ready mobile devices like handheld computers, PDAs and smart phones to access the internet. The term mobile device refers to a device specially designed for synchronous and asynchronous communication while the user is on the move [3]. Among the mobile devices, the mobile phone and PDA are the most popular and commonly used by the users, and one of the facilities is accesses the internet [3]. These kinds of devices provide good mobility but very limited computational capabilities and display size [7]. Still, many mobile users don’t feel easy to browse the internet via mobile devices, because of small screen, limited memory, processing speed and slow network [1][7]. Since most of the existing web contents are originally designed for display in the desktop computers, so the content delivery without layout adjustment and content adaptation makes the contents unsuitable for mobile devices. User mobile devices need to scroll the screen horizontally and vertically to find the desired content. Moreover, searching and browsing could frustrate because most of the web site is designed for the standard desktop display. This means that most of the web contents are unsuitable for mobile devices [4]. A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 149–163, 2011. © Springer-Verlag Berlin Heidelberg 2011
150
R. Anam, C.K. Ho, and T.Y. Lim
Content adaptation refers to techniques, which dynamically adjust the contents in the direction of represent the contents according to the properties of the handheld devices for better presentation. The conventional way to provide web contents to support various types of handheld devices is to create the same contents but different formats for different devices. This method is simple but the chances of making errors are very high with different handheld devices. To support new handheld devices, all the previous web contents have to be reformatted for that handheld device. Sometimes, changes in the main contents require updating of all handheld contents format. So, this is neither a practical nor a feasible solution for large volume of web contents. Since the screen size is limited, the content adaptation method needs to apply various content transformation processes including layout changes and content format reconfiguration [4][6][7]. However, simple content adaptation solution for changing multiple-column layout to a single-column layout for handheld screens also shows some disadvantages. Without the semantic analysis and relationship among the semantic objects, this kind of adaptation may cause an awkward organization of a web page and gives different meaning of information. A tool or mechanism is needed to provide the users opportunity to flexible web contents on handheld devices. Web contents are typically composed of many multimedia objects (text, images, audio and video) [12], which are semantically connected by various objects in a section. For example, an image can be illustrated by a section of a text article or a text title can abstract an article and some images. In other words, these related objects are integrated to help readers to understand what the section intend to express. Improper arrangement of these kinds of objects and their relationship may lead to the misunderstanding or loss of information to the users. Therefore, it is very important for a content adaptation mechanism to maintain the original semantic relationship among the objects during the adaptation and content delivery process. Web contents are categorized into three kinds of useful objects such as the navigation bar (main menu), navigation list (list of anchor texts links to the actual content) and content (details information) [14]. These three types of objects contain all the important contents of the web page. However, most of the web pages contain these three types of objects to represent the information. In this paper, we present a novel method which supports web page content adaptation for handheld devices. Our goal is to improve web content accessibility, deliver the target information to the users. So the users can reach to the target information from a large web page. To achieve this goal, we introduce Flexi-adaptor (Fig.1) as an automatic content adaptation system. Our algorithms automatically identify the semantic relationship among the objects from the block, identify the objects and categorize them into the navigation bar, block title and content of the block. We use toggle functionality to the block title of the each block to show and hide the contents of the block. The depth first search algorithm selects the objects and delivers them to the users. The rest of the paper is organized as follows: Section 2, summarizes the related research work. Describe the proposed solution in section 3. Section 4, discussion and result about the system. Finally, we conclude in section 5.
Flexi-adaptor: A Nobel Approach for Adapting Web Content for Mobile Devices
151
2 Related Work There are general-purpose content adaptation systems have been developed. The CMo uses the proxy-based architecture to adapt web contents for handheld devices. This system reduces the information to overload by allowing its users to see the most relevant fragment of the pages and navigate between other fragments if necessary. It captures the context of the links and applies a simple topic boundary detection technique and also uses the context to identify the relevant information in the next page with the help of vector machine [5]. Web Clipping is one of the methods that researchers are still working on it. Web Clipping is a technique where the system extracts and represents some part of HTML document for mobile browser [8]. An annotation or parser is general declares the properties that qualify a particular portion of a target document. The system annotates some part of the web page, and it provides the annotation contents to the content adaptation engine. The Web Clipping methods modify the HTML structures. This system breaks the page into small parts, makes them a new separate page and adds title and header. Sometimes it removes extra unnecessary objects from the HTML. The concept of this method is to read the web page, tag some parts of the page and regenerate the web page for the mobile browser. Xadaptor is a content adaptation system that consists of rule-based and fuzzy logic approach. The rule-based approach facilitates extensible, systematic and adaptive content adaptation and also integrates adaptation mechanisms for various content types and categories them into the rule-based approach. Rules are applied by the user define preferences. The HTML objects are transformed into content and pointer objects. Therefore, the system uses the content parser to separate the objects from the HTML tags. Moreover, the system reformats the standard tables from the HTML objects. The table reformatting algorithm uses fuzzy logic and rule-based approaches. This system is not fully automated. The users need to assign some parameters to adapt the contents [2]. Some researchers use Vision Based Page Segmentation (VIPS) algorithm [1][46][9] that manages the web structure to find the interesting objects and restructure the web pages as a block. The Web page Tailoring System follows some rules to select the interesting blocks [9]. The system removes unnecessary information, creates a new title for the block and tries to summarize the block contents information. The VIPS identifies the interesting data, change the format of the web page and also represent the information to the user with user’s interest. User preferences are used by Page Segmentation and Pattern Matching to make the information interesting for every user. Researches use Web Clipping, rule-based and VIPS to identify the objects and adapt the contents. However, users still need to scroll vertically. Sometimes users browse the contents but fail to reach the target content because of overload information [13]. Still, it is a great challenge to achieve satisfactory precision for dynamic web page segmentation, which is based on HTML element analysis.
152
R. Anam, C.K. Ho, and T.Y. Lim
3 Proposed Solution A mobile user first connects to the Flexi-adaptor with the web browser. The user types the URL of a web page in the input box (Fig.2). If the user likes to search any particular info in the page which he wants to view after adaptation, then key in the word in the search field. The users submit the request to the Flexi-adaptor, and it delivers the sub pages to the users. All the sub page contents are fitted for the mobile screen. The users do not need to scroll. Users use the next and previous buttons to navigate the sub pages. Input Original Web Page Output Adapted Web Page
Identification of Important Blocks
Identification of Type Specific Objects of a Block
Depth First Traversal Algorithm for Selecting Blocks to Display on Mobile
Assign Toggle Functionality to the block Title
Pre-process Each Block Object Elements Assign Weight to Each Object of a Block
Fig. 1. The framework of the Flexi-adaptor
Fig. 2. Homepage of the Flexi-adaptor
3.1 Identification of Important Blocks The Flexi-adaptor first parsers the HTML page to a tree. The root node is the top element of the tree. The internal node means a node with one or more children [10]. Fig.3(b), illustrates the tree presentation of the block-5 (Fig.3.a). This tree contains all the context of the block-5 HTML elements. The root of the tree is (1), internal node is
(2) and leaf node is (3). The system identifies the blocks according to the semantic relationship among the HTML elements [14]. Since web pages are always presented in HTML format, the system traverses the tree and identifies the blocks in the tree according to the HTML elements, analyze their properties and tag the blocks. The system avoids objects like JavaScript and CSS. These objects are not important for mobile view [3]. The system uses the Depth First Search to traverse the tree. It removes nodes contain elements like , to simplify the tree [3]. The semantic blocks are identified by using structure functionality [7], which arranges the layout of information objects
Flexi-adaptor: A Nobel Approach for Adapting Web Content for Mobile Devices
153
[13]. Each semantic block is an individual block content unit, and it contains a group of object. Fig. 3. (a), illustrates a typical page from “BBC”, block 1 is the menu, Block-2 is the top news, Block- 4,5,7,8,10,11 are the relevant interesting blocks. Block- 6 and 9 contain some related links and Block-3 is an advertising Block.
(b)
(a) Fig. 3. (a) Original BBC webpage categories as semantic blocks and (b) HTML Tree hierarchy presentation of Block-5 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.
Input: T is the Tree; Output: T is a Tree contains only important blocks; Start Visit(T,v) Perform visit of the node v; If v match with structure functionality Then If v is an internal node Then For all child w of v Do Visit(T,w); If w do not match with structure functionality Then Remove all the children of w; Else Remove all the children of v; End
Fig. 4. Algorithm for identifying the important blocks
Fig. 4. Illustrates the algorithm to identifying the important blocks. The input is the original HTML tree T (line 2), output tree T contains only the important blocks (line 4), v and w is the node of the tree and structure functionality [7] is the set of HTML tags, which points to the HTML blocks. The algorithm traverses the tree and searches for the structure functionality (line 8). If it matches, then it keeps the internal nodes and leaf nodes, if it does not match, it removes the internal nodes and leaf nodes from the tree. This algorithm keeps only the important block contents in the tree and removes other contents.
154
R. Anam, C.K. Ho, and T.Y. Lim
3.2 Identification of Type Specific Objects of a Block We categorized type specific object in this system as the navigation bar (menu), block heading (block title) and main content (list of anchor texts links to the actual content and details information) [11][14]. This algorithm searches for the objects (Fig.5) from the tree and sends it for categorize assignments. The categorize assignment algorithm (Fig.6) tagged the Menu Object, Block Title Object and Main Contents from the tree according to the type of the objects. 1. Input: 2. T is the Tree; 3. Output: 4. T is the Tree with tagged node; 5. Start 6. Traverse(T,v) 7. Perform visit of the node v; 8. AssignmentOfCategory(v); 9. Preprocessing(v); 10. AssignmentOfWeight(v); 11. If v is an internal node Then 12. For all child w of v do 13. Traverse(T,w); 14. End
Fig. 5. Algorithm to identifying the object from blocks
The search keyword option (Fig.2) gives the users more specific target information. If the user keys in any prefer data, the system checks the keyword user prefers to find, and it starts searching the keyword from the root to leaf node HTML elements of the tree. Fig.6. (line 7-9) algorithm shows, if it finds any keyword matches with the user preferred data, tag the object as match, remove the node from the tree, add a new left child node to the root of the tree and assign the match data node to the new left child node. All this kind HTML elements use for format the heading [7] purposes to display the title of the blocks and contains important information. The algorithm searches these HTML elements inside the tree and tags the object as heading. Fig.6.(line 1011), illustrates the algorithm searches all the block title elements and tags the individual content unit as heading. Most of the web page use hyperlinks which navigate to the main content or another portal. The algorithm tags the individual content unit according to the Hyperlinks text lengths. There are two kinds of Hyperlinks, Menu (navigation bar) and Content (navigation list) Hyperlink [14]. Navigation on a web site is achieved by a collection of links (text/images/flash/java applets) which is the website navigation menu or the website navigation bar. This navigation menu or bar is usually placed vertically on the left, right or horizontally near the top of the web page. Fig.6.(line 1216) algorithm illustrates, it searches for the navigation object then checks the text length of the object. First, tag the object as the menu, if the text length is less than 13 [14]. Second, if the text length is more than 13, then tag the object as content.
Flexi-adaptor: A Nobel Approach for Adapting Web Content for Mobile Devices
1. 2. 3. 4. 5. 6. 7. 8. 9. 10 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.
155
Input: v is the node of the tree; Output: v with tagged value; Start AssignmentOfCategory (v) If search keyword matches with v and preference enable Then Assign vp = match; Create new left child of the root and shift v to the new left child; If heading object matches with v Then Assign vp = heading; If hyperlink object matches with v Then If text length is less than 13 Then Assign vp = menu; If text length is equal or more than 13 Then Assign vp = content; If text object matches with v Then Assign vp = content; If multimedia object matches with v data and multimedia object pixels > 21 Then Assign vp = content; End
Fig. 6. Algorithm for identification of type specific objects of a block
Text and multimedia contents used to deliver the information context which is considered as main content. The system searches for the main content and tags the object. Fig.6.(line 17-18) algorithm shows, it searches for the text contents. If the node contains the text objects, then tag the object as content. Fig.6.(line 19-20) algorithm illustrates, it searches for the images, when the image width is less than 21 pixels ignore the image to tag because, less than 21 pixels either use as a symbol of navigation or fill the blank space in the web page. When the image width is more than 21 pixels, tag the object as the content because it is considered as main content. 3.3 Pre-processing of Blocks Some HTML elements are used to format the data to highlight important information as the title [7] and decoration [7] makes attractive to the readers. For mobile readers, this kind of extra formatted data presentation is not necessary. Therefore, remove and edit the HTML elements. Fig.7.(line 7), the algorithm illustrates Iscreen control the multimedia contents dimension. Iscreen contains the ratio of the multimedia contents which is suitable for the mobile screen. The algorithm (line 9) first, removes all the HTML decoration elements, change the background color of the elements property. Second, edit the multimedia contents dimension properties, if the multimedia contents dimension is more than Iscreen (line 10-20). This algorithm modifies the original dimensions of the multimedia contents and reduces the dimensions which suit in the mobile screen.
156
R. Anam, C.K. Ho, and T.Y. Lim
1. Input: 2. v is the node of the Tree; 3. Output: 4. v with modified node of the Tree; 5. Start 6. Preprocessing(v) 7. Iscreen =( DeviceWidth / DeviceHeight) × 200; 8. If v is decoration or highlight element Then 9. Remove the element from v and change the background color properties; 10. If v is multimedia element and multimedia height, width more than Iscreen Then 11. MaxHeight = Iscreen; 12. MaxWidth = Iscreen; 13. Ratio = height / width; 14. If height > MaxHeight Then 15. newheight = MaxHeight; 16. newwidth = height / Ratio; 17. Elseif width > MaxWidth Then 18. newwidth = MaxWidth; 19. newheight = width × Ratio; 20. update v with the new width and height; 21. End
Fig. 7. Algorithm to pre-processing each object elements
3.4 Assignment of Weight to Each Object in a Block The Flexi-adaptor assigns weight according to text length and multimedia objects dimension of each object from the tree. The depth first traversal algorithm uses the object weight to deliver limited contents to every subpage. The system considers Text and Multimedia objects to assign weight. The system searches all the objects from the tree, if finds any text content in the HTML element, the system count the text lengths, convert the text length into pixels and assign to the object as weight. Fig.8.(line 7-8), shows the algorithm convert the text lengths to pixels. Here vw is the total number of the pixels, 10 is the height of the character, and 30 is the threshold, vw = (30+10× number of text character); [2]. 1. Input: 2. v is the node of the Tree; 3. Output: 4. v with weight value; 5. Start 6. AssingWeight(v) 7. If v is text object Then 8. Assign weight to vw = vw + (30 + 10 × number of text character); 9. If v is image object and image height, width more than 20pixels Then 10. Assign weight to vw = vw + (height of the image × width of the image); 11. End
Fig. 8. Algorithm for assignment of weight to object
The system searches all the objects of the tree, if finds any of the multimedia contents in the HTML element, checks the properties of the multimedia contents. If the properties of the content dimensions are provided, then grab the information for
Flexi-adaptor: A Nobel Approach for Adapting Web Content for Mobile Devices
157
the weight, otherwise the system checks the multimedia content and gets the dimension by itself. Fig.8.(line 9-10), shows the algorithm where the height of the image is the height of the multimedia content and width of the image is the weight of the multimedia content. The variable vw is the area and weight of the multimedia object. vw = height of the image × width of the image; 3.5 Organizing Block Contents Using Toggle Functionality to the Block Title The block title contains the condensed information of the body (information content objects). So the user can understand the content of the body by looking at the block heading object. The Flexi-adaptor searches the tagged heading objects and content objects. It adds a navigation object to the block title and hides the block body. Fig.9 (a), illustrates the CNN homepage. This page contains the block title and block body. Fig.9 (b), shows after using the toggle functionality to the block title, extra plus symbol adds to the block title as the navigation and block body gets hidden. If the user clicks the block title plus symbol, the block body gets visible in the browser.
Block Title Block Title
(b) Block Title Block Body Block Title
Block Title Block Body Block Title
Block Body
Block Body Block Title
(a)
(c)
Fig. 9. (a) CNN homepage contains block title and block body with information, (b) adapted CNN homepage contains block title with navigation and block bodies are hidden, (c) adapted CNN homepage after clicking the block title navigation. The block body appears to the screen.
Fig.9 (c), illustrates the block heading with the minus symbol and the block body appears. Fig.10, shows the algorithm (line 7-8), if find any heading object, then assign the toggle object to the block title. Searches for the block body contents (line 9-11), find any contents then hide them from display. The hidden block body appears (line 12-13) in the screen if the user clicks the header navigation symbol.
158
R. Anam, C.K. Ho, and T.Y. Lim
1. Input: 2. v is the node of the Tree; 3. Output: 4. v with toggle object value; 5. Start 6. BlockToggle(v) 7. If heading object matches with v Then 8. Assign vt = enable toggle object; 9. If v has child node w Then 10. If all node w has information content object Then 11. Assign vt = all node wi and hide information content object; 12. If toogle the heading object Then 13. Display the hidden information content object; 14. End
Fig. 10. Algorithm for organizing block contents using toggle block title
3.6 Depth First Traversal Algorithm for Selecting Blocks to Display on Mobile Devices The depth first traversal algorithm selects the tagged objects from the tree and sub pages with the weight capacity of the objects. The total weight of objects may not be more than the mobile screen capacity for each sub page. 1. Input: 2. v is the node of the Tree; 3. t is the tagged info of the individual object unit; 4. contents is the tree individual content unit; 5. w is the weight of the individual object unit; 6. next is the next button; 7. previous is the previous button; 8. Output: 9. subpage is HTML page contains individual content units; 10. Start 11. dfs(v) 12. If next = TRUE Then 13. If weight + wi ≤ C Then 14. v = visited; 15. If ti matches with user preferences Then 16. subpage = subpage append contentsi ; 17. weight = weight + wi ; 18. For each child x of parent v 19. If x is unvisited Then 20. dfs(x); 21. Elseif previous = TRUE Then 22. If weight + wi ≤ C Then 23. v = unvisited; 24. If ti matches with user preferences Then 25. subpage = subpage append contentsi ; 26. weight = weight + wi ; 27. For each parent v of child x 28. If v is visited Then 29. dfs(v); 30. End
Fig. 11. Depth first traversal algorithm for selecting blocks
Flexi-adaptor: A Nobel Approach A for Adapting Web Content for Mobile Devices
159
Fig.11. illustrates, sub page p generator scheme from the tree. Let contentsi is the tree object, wi is the weightt, and ti is the tagged info of the object assigned by ith item, C is the capacity base on the t screen mobile dimension (Height × Width). The deepth first traversal algorithm sellects the objects and the weight smaller or equal to thhe C capacity. When the weight becomes b more than the capacity C, then process stops. T The algorithm delivers the seleccted contents in a sub page. The next and previous vallues use to direct next sub page or previous sub page. If the next is true (line 12), then the algorithm traverse the unvisited nodes of the tree, tag the node as visited and add the objects in the sub page (lin ne 13-20). If the previous is true (line 21), the algoritthm traverse the visited nodes of o the tree, tag the node as unvisited and add the objectts in the sub page (line 22-29). The T depth first traversal algorithm creates small sub paages from the tree. Fig.12, illustrate the BB BC main page adapted by Flexi-adaptor. Most of the W Web sites can be adapted by th his system. Sometimes, there are few web sites can’tt be adapted properly because th hey don’t follow the standard (W3C) format of the HTM ML.
Fig. 12. Adapted BBC B home page by Flexi-adaptor on handheld device
4 Results and Discusssion A mobile user first opens a general web browser and connects to the Flexi-adapptor. The user types the URL of o a web page in the input box Fig.1. If the user likess to search any particular info in n the web page, needs to key in the keyword to the seaarch field and submits the requesst to the Flexi-adaptor and it provides contents, which fiit to the users screen. The systeem delivers the adapted contents to the users. Fig.133.(a) illustrates the desktop versiion of CNN web page. After adapting by the Flexi-adapptor, fig.13.(b) illustrates the CN NN web page for the mobile. The background color, ffont
160
R. Anam, C.K. Ho, and T.Y. Lim
(a)
(b)
Fig. 13. (a) CNN webpage desktop version and (b) adapted CNN webpage for the mobile
size and multimedia contents are modified to suit in the mobile screen. The block title has [+] navigation symbol, and the bodies of the block contents are hidden. If the user toggle the navigation, the body of the block contents appears in the screen and the block title navigation symbol changes to [-]. If the user does not find the target contents, then the user can navigate to next or previous pages. Next part of this paper we compare our proposed method with other existing methods; discuss potential implementation issues and other considerations. 4.1 Framework Comparison There are many significant differences between our proposed system and other existing content adaptation systems. The Xadaptor [2] framework builds on five components. It is a rule-based adaptive content adaptation system; fuzzy logic to model the adaptation quality and control the adaptation decision. The Webpage Tailoring System [6] is a complete framework to adapt contents for mobile devices. This system consists of three components. It uses the mechanism that can determine, which blocks in a webpage should be retained by user preferences and arrange the blocks. The CMo [5] framework builds on three components. It captures the context of the link, applies simple topic-boundary detection technique and uses the context to identify relevant information in the next page by using vector machine. The Flexiadaptor framework consists of six components. We compare the components on table 1, which they use to develop their systems. Table.1. illustrates the comparison of the components from different frameworks. Other frameworks do not hide and display block contents dynamically. In our proposed system, arrange the block title with toggle facilities to display and hide the body contents of the block to provide the condensed contents to the users.
Flexi-adaptor: A Nobel Approach for Adapting Web Content for Mobile Devices
161
Table 1. Comparison with mobile content adapter components Components Used
(1) Xadaptor System Yes
(2) CMo System
(4) Flexiadaptor
Yes
(3) Web Page Tailoring System Yes
Representation of HTML page as Tree
Yes
Yes
Yes
Yes
Block identification Object identification
Yes
Yes
Yes
Yes
Yes
No
No
Yes
Modification object elements
Yes
No
No
Yes
Toggle the contents for display and hiding
No
No
No
Yes
Use mechanism to select the target contents for display
No
Yes
Yes
Yes
User preferences
Yes
Comments
System 1 and 3 uses predefined user preferences from database but system 2 and 4 uses dynamic preferences from the users. System 1 use structure tree, system 2 use frame tree, system 3 use DOM tree and system 4 uses simple HTML tree. System 1- 4 identify the blocks by HTML tags. System 1and 4 identify the objects by HTML tags. Some embedded and navigation objects are not important for mobile readers. System 1 modifies the objects according to the user database, but system 4 modifies the objects from the instant data. Which helps the contents to fit in the mobile screen. System 4 toggles the block title for showing and hiding the body of the block. This component hides the body of all the blocks, which makes the page very small and shows only the condensed information. System 2 delivers the related information. System 3 delivers information according to the tag pattern matching, and system 4 delivers exact information.
Table.2. illustrates the comparison of the algorithms. Our proposed blocks identification algorithm complexity is O(n) because Fig.4. Line 10, uses an iterative loop and recursive function. In the FindBlocks [5] algorithm line 3 and 14 uses iterative loops and the complexity is O(n). Second, proposed objects identification algorithm's complexity is O(|n|+|e|) because Fig.5. Line 12, uses the iterative loop
162
R. Anam, C.K. Ho, and T.Y. Lim
and recursive function. However, the FindContext [5] algorithm line 6 and 10 use nested iterative loop. So the complexity becomes O(n2). Third, Fig. 7, Our object elements modification algorithm complexity is O(n) and sub page navigation algorithm [15] line 4 complexity is O(n). Fourth, our contents height and width calculation algorithm complexity is O(n) and sub page navigation algorithm [15] line 5 complexity is O(n). We can see that over-all complexity of our proposed algorithm is same with other algorithms. However, proposed object identification algorithms complexity is less than the FindContext [5] algorithm. Table 2. Comparison the complexity of the algorithm Algorithm Blocks Identification Algorithm Objects Identification Algorithm Object Elements Modification Algorithm Contents Height and Width Calculation Algorithm
Time Complexity of Proposed System
Time Complexity of Other Systems
O(n) O(|n|+|e|) O(n)
O(n) [5] O(n2) [5] O(n) [15]
O(n)
O(n) [15]
4.2 Visual Comparison The Web Page Tailoring System, Xadaptor and our proposed system delivers the contents according to the user preferences. However, there are differences in the visual outlook. Our system shows only the block title and hides the body of the block. So, all the condensed data are displays in the screen. The Web Page Tailoring System, users can zoom in which part of the page he wants to read. The multimedia content sometime gets oversized to the mobile screen because The Web Page Tailoring System doesn’t adapt all kinds of contents. The Xadaptor and proposed system adapt the multimedia and text contents with rule base approach. The Xadaptor and Web Page Tailoring System display all the contents in a single column page but our proposed system delivers all the condensed contents in sub pages.
5 Conclusion In this paper, we proposed a new method for facilitating the browsing of a large webpage on a handheld device. The Flexi-adaptor convert's HTML page into a tree and identify the semantic blocks. Each object of the blocks gets tagged and contains weight value. The Pre-processor modify the contents. The toggle block title algorithm uses to hide and show the body contents. The Depth First Traversal algorithm selects the contents. The system generates sub pages for the handheld devices. Our approach enables a new browsing experience to the users. The block title will appear in the screen which carries the most significant information. A new browsing method overcomes the limitation of a mobile device with a small screen and makes them truly useful for information access.
Flexi-adaptor: A Nobel Approach for Adapting Web Content for Mobile Devices
163
References 1. Xiao, X., luo, Q., Hong, D., Fu, H., xie, X., Ma, W.-Y.: Browsing on Small Displays by Transforming Web Pages into Hierarchically Structured Subpages. ACM Transactions on the Web 3(1) Article 4, (January 2009) 2. He, J., Gao, T., Hao, W., Yen, I.-L., Bastani, F.: A Flexible Content Adaptation System Using a Rule-Based Approach. IEEE Transactions on Knowledge and Data Engineering 19(1), 127–140 (2007) 3. Blekas, A., Garofalakis, J., Stefanis, V.: Use of RSS feeds for Content Adaptation in Mobile Web Browsing. In: International cross-disciplinary workshop on Web accessibility (W4A), pp. 79–85 (May 2006) 4. Lee, E., Kang, J., Choi, J., Yang, J.: Topic-SpecificWeb Content Adaptation to Mobile Devices. In: International Conference on Web Intelligence, pp. 845–848 (December 2006) 5. Borodin, Y., Mahmud, J., Ramakrishnan, I.V.: Context Browsing with Mobiles - When Less is More. In: International conference on Mobile systems, applications and services, pp. 3–15 (June 2007) 6. Kao, Y.-W., Kao, T.-H., Tsai, C.-Y., Yuan, S.-M.: A personal Web page tailoring toolkit for mobile devices. Computer Standards & Interfaces 31(2), 437–453 (2009) 7. Ahmadi, H., Kong, J.: Efficient Web Browsing on Small Screens. In: International Conference on Advanced Visual Interfaces, pp. 23–30 (May 2008) 8. Horia, M., Ono, K., Abe, M., Koyanagi, T.: Generating transformational annotation for web document adaptation: tool support and empirical evaluation. Journal of Web Semantics 2(1), 1–18 (2004) 9. Lee, E., Kang, J., Park, J., Choi, J., Yang, J.: ScalableWeb News Adaptation To Mobile Devices Using Visual Block Segmentation for Ubiquitous Media Services. In: International Conference on Multimedia and Ubiquitous Engineering (MUE 2007), pp. 620–625 (April 2007) 10. Pan, R., Wei, H., Wang, S., Luo, C.: Auto-adaptation of Web Content: Model and Algorithm. In: IET 2nd International Conference on Wireless, Mobile and Multimedia Networks (ICWMMN 2008), pp. 507–511 (October 2008) 11. Chen, J., Zhou, B., Shi, J.: Function-Based Object Model Towards Website Adaptation. In: International World Wide Web Conference, pp. 587–596 (May 2001) 12. Yang, S.J.H., Zhang, J., Chen, R.C.S., Shao, N.W.Y.: A Unit of Information–Based Content Adaptation Method for Improving Web Content Accessibility in the Mobile Internet. ETRI Journal 29(6), 794–807 (2007) 13. Xu, K., Zhang, D., Zhu, M., Gu, T.: Context-Aware Content Filtering & Presentation for Pervasive & Mobile Information Systems. In: International Conference on Ambient MEdia and Systems and Workshops, Article 20 (February 2008) 14. Lee, E., Kang, J., Choi, J., Yang, J.: Topic-Specific Web Content Adaptation to Mobile Devices. In: International Conference on Web Intelligence, pp. 845–848 (December 2006) 15. Gupta, A., Kumar, A., Mayank, Tripathi, V.N., Tapaswi, S.: Mobile Web: Web Manipulation for Small Displays using Multi-level Hierarchy Page Segmentation. In: International Conference on Mobile Technology, Application & Systems, pp. 599–606 (September 2007)
Performance Evaluation of the Developed Automatic Musical Note Translating System Chooi Ling Si Toh, Chee Kyun Ng, and Nor Kamariah Noordin Department of Computer and Communication Systems Engineering, Faculty of Engineering, University Putra Malaysia, UPM Serdang, 43400 Selangor, Malaysia [email protected], {mpnck,nknordin}@eng.upm.edu.my
Abstract. The automatic musical note translating system refers to the ability of a computer to translate the captured or recorded melody into musical note automatically and instantly. This paper presents the development of such system, whereby a musician can compose music directly without any extra process or procedure, in real-time environments. In this automatic translating system, the information of the captured or recorded melody from microphone or musical instrument is first analyzed through digital signal processing (DSP) to extract its fundamental frequency. The extracted fundamental frequency is then compared with the predefined frequency of musical notes in the computer. The matched musical note frequency will be distinguished at the developed musical stave interface instantly. The performance of the developed automatic musical note translating system will be evaluated with different sampling rates. Results show that the optimized sampling rate to provide most accuracy in the developed system is 44100 Hz. Keywords: Melody, musical note, automatic translating system, fundamental frequency, sampling rate.
1 Introduction A great piece of composition is the conjunction of inspiration and creativity of a composer. Inspiration is mostly sourced by composer’s mood, character, memories and their observation to the surroundings. The inspired melody line is then enriched by the creativity of the composer, where harmony and dynamic will be exerted to thicken the performance of the inspired melody line. Music is always the best medium to express the feeling and virtual idea across the boundary of geography as it is perceived as a global language that everyone speaks. It is the poet writing on our soul where it touches the soul without any literary reading. Music composition is always ascribed to the art work of people who owned musical background. Generally, musical note is mainly composed based on its chord and melody [1]. In order to compose a piece of music chord, ones must have the knowledge of musical notation elements which are pitch and rhythm. Pitch governs the melody and harmonic whereas rhythm elaborates the timing of musical sounds and silences [2]. With this knowledge, composers are able to elucidate their composition in a written form – the Music Score. A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 164–178, 2011. © Springer-Verlag Berlin Heidelberg 2011
Developed Automatic Musical Note Translating System
165
In recent years, music industry has occupied the knowledge of computer science and engineering to develop the music score where the composer can elaborate the musical notation through mouse click while composing music [3], [4]. Also, the pitch and rhythm can be displayed neatly at the monitor compared to traditional handwritten form. Furthermore, hardware devices also involve in music production in order to produce a high quality of music product [5]. The electronic musical instruments have been invented tremendously as the application of electronic industry is expanding. These musical instruments are mainly used in contemporary musical composition. The electronic piano are currently developed to integrate with personal computer to provide a platform for composer to compose their melody art work. A special designed software tool is used to enable the pitch of melody to be displayed according to the triggered keystroke on the piano keyboard [6]. However, the composing music activities should not restrict to merely individual who have the musical background. Everyone should be a music composer as long as they have the inspiration with a melody in their thought. Melody is subjective and it differs among individuals. A great melody might be inspired by ones who do not have the music background. But the great melody line will be abandoned when it does not being recorded or written down. This is ascribed to the waste of art as the precious idea which could not be expressed due to the lack of literacy in musical notation. Although the advance of integration of computer science and engineering has developed various software and hardware equipments to enhance the musical composition workflow, there is no system tool that is suitable for such individuals to express their ideas in musical written form. This paper presents the design and development of an automatic translating system that translates the captured or recorded melody into musical note by using digital signal processing (DSP). A composer can compose music directly in real-time to generate musical score without extra processes or procedures. In this developed system, the captured or recorded melody is quantized as a digitized waveform which representing an acoustical music signal. This is an attempt to derive the musical note from the signal such that a musical score can be produced. By using the pitch detection algorithm from the cepstral analysis, the fundamental frequency of acoustic signal is extracted. The extracted fundamental frequency is then compared with the predefined frequency of musical notes in the computer. The matched musical note frequency will be distinguished at the developed musical stave interface instantly. Finally, the performance of the automatic musical note translating system is evaluated with different sampling rates and also with different sizes of random access memory (RAM) in the computer. Results show that the optimized sampling rate to provide the highest accuracy in the automatic musical note translating system is 44100 Hz regardless of the sizes of computer RAM. The rest of this paper is organized as follows: Section 2 presents the characteristic of musical acoustic signal. Section 3 explains on how to detect the pitch of recorded acoustic signal. The development of the automatic musical note translating system is described in Section 4. The performance evaluations of the developed system are discussed in Section 5. Finally, this paper is concluded in Section 6.
166
C.L. Si Toh, C.K. Ng, and N.K. Noordin
2 Characteristic of Melody Signal The musical acoustic signal is a periodic signal with frequencies that are multiple integers of a fundamental frequency which is called harmonics. The proportion is according to the associated harmonics in this sum to determine the sound timbre, while the sound pitch is determined by the fundamental frequency. Henceforth, the frequency of a harmonic musical acoustic is called the fundamental frequency. A harmonic of a wave is a component frequency of the signal that is an integer multiple of the fundamental frequency. If the fundamental frequency is f, the harmonics have frequencies of 2f, 3f, 4f. etc. The harmonics have the property that they are all periodic at the fundamental frequency; therefore the sum of harmonics is also periodic at that frequency [6]. Harmonic frequencies are equally spaced by the width of the fundamental frequency. These frequencies can be found by repeatedly adding of that frequency. For example, if the fundamental frequency is 25 Hz, the frequencies of harmonics are 50 Hz, 75 Hz, 100 Hz, etc. A plucked guitar string or a struck drum head or struck bell, which naturally oscillated with not only one, but several frequencies which are known as partials. Many of these partials are integer multiples of the fundamental frequency; these are called harmonics. The harmonic wave motion is depicted as in Fig. 1.
Fig. 1. The illustration of harmonics for a vibrating string
When a note is played on a musical instrument, such as Middle “C” played on a piano, it does not comprise of just one frequency but a complex and made up of many frequencies that are combined to yield the audible sound signal. Table 1 shows an example of middle “C” frequency value and its initial harmonics. When these periodic
Developed Automatic Musical Note Translating System
167
waveform sums up, the frequency waveforms for the Middle “C” (C4) note is formed as illustrated in Fig. 2. Note that the musical signal is not a pure sine wave although human tends to hear a musical acoustic that may sound at a particular frequency. The first or predominant frequency that human heard is fundamental frequency whereas the additional frequencies that make up the note are referred to harmonics which forms the timbre of musical sound signal. Nevertheless, the musical sound signal is not restricted to only pitch musical instrument, but also human vocal [6]. Table 1. Fundamental frequency and harmonics of Middle “C” C4 (Middle C)
262 Hz
Fundamental
C5
523 Hz
First Harmonic
G5
785 Hz
Second Harmonic
C6
1046 Hz
Third Harmonic
E6
1318 Hz
Forth Harmonic
Fig. 2. The waveform of Middle “C” (C4)
3 Melody Pitch Detection Algorithm A primary underlying task in this section is the extraction of features from acoustic signal, which can be used for developing the proposed automatic musical note translating system. One of the important features from acoustic signal is fundamental frequency or commonly referred to as pitch. The acoustic signal such as human voice and musical melody are ascribed to harmonic signal. Fundamental frequency corresponds to the rate at which the acoustic signal vibrates. To identify the pitch of harmonic signal, cepstral analysis is applied instead of spectral analysis. The cepstral analysis method is known as a harmonic signal analysis tool commonly used to measure the fundamental frequency of speech. It separates a strong pitched component from the rest of spectrum [7]. Cepstral analysis method applies the principle where a frequency spectrum of a time harmonic signal that consists of a series of impulse responses at the fundamental
168
C.L. Si Toh, C.K. Ng, and N.K. Noordin
frequency and its harmonics, occurs at integer multiples of the fundamental frequency. This is a reasonable model of many vocal and instrumental sounds whose spectrums can be considered as the sum of an excitation sequence, e(n) – the original vibration impulses typically at the pitch of the sound, and the resonances, h(n) – the discrete impulse response of the vocal tract. Hence, the sequence of an acoustic signal, x(n) can be represented as
x ( n ) = e( n ) * h ( n )
(1)
In the autocorrelation function, the effects of the vocal source and vocal tract are convolved with each other. This results in broad peaks and in some cases multiple peaks in the autocorrelation function. In frequency domain, the convolution relationship between vocal source and vocal tract effects becomes a multiplicative relationship of
X (ω ) = E (ω ) * H (ω )
(2)
where X(ω) = F{s(n)}, E(ω) = F{e(n)}, H(ω) = F{h(n)} and F{} is Discrete Fourier Transform (DFT) function. The multiplicative relationship between vocal source and vocal tract resulted in cepstrum is transformed into an additive relationship such that both vocal source and vocal tract are nearly independent or easily identifiable and separable as shown in Fig. 3. The vocal source can be viewed as a sequence of quasiperiodic pulses. The DFT of these pulses is a line spectrum where the lines are spaced at harmonics of the original frequency, which is shown as the thin wiggly lines in Fig. 3. In contrast, the spectrum of the response of the vocal tract is a slowly varying function of frequency, which is shown as the thick undulating line in Fig. 3.
Fig. 3. The separation of vocal source and vocal tract
It is possible to separate the part of the cepstrum, which represents the vocal source signal. For the melody pitch determination, the real part of cepstrum is sufficient. The real part of cepstrum of the discrete signal x(n), Cepx(n) is defined as
Developed Automatic Musical Note Translating System
Cepx ( n) =
2π j kn 1 N −1 N ⋅ X ( k ) e N k =0
169
(3)
where X(k) is logarithmic magnitude spectrum of x(n) as 2π N −1 − j kn X (k ) = log x(n) ⋅ e N n=0
(4)
Hence, the cepstrum of a discrete-time signal s(n), Cepx(k) can be defined as an inverse N-point DFT for the logarithmic magnitude of N-point DFT for s(n) [9], and it is given by
Cepx(k ) =
2π 2π j kn − j kn 1 N −1 N −1 N N log x ( n ) e ⋅ e ⋅ N k =0 n=0
(5)
The cepstrum of an acoustic signal segment is having a peak at the period of fundamental frequency of the segment. The discrete function Cepx(k) is used to search for the largest peak in the pitch period region of interest. If this peak value is above the detection threshold, the pitch period is defined as the location of this peak. If this peak is below this threshold, a pitch will not be detected [8]. For discrete-time signals, the cepstrum of a signal is defined as the inverse DFT (IDFT) for the logarithmic magnitude of the DFT for the signal. That is, the cepstrum of a discrete-time signal x[n], c[n] is defined as
c[n] =
1 2π
π
π log X (e )e iw
iwn
dw
(6)
−
where the DFT of the signal is defined as
X (e iw ) =
∞
x[n]e
−iwn
(7)
n = −∞
Note that c[n], being an IDFT, is nominally a function of a discrete index n. If the input sequence is obtained by sampling an analog signal, i.e., x[n] = ax(n/fs), then it would be natural to associate the time with the index n in the cepstrum. Authors in [3] introduced the term quefrency for the name of the independent variable of the cepstrum in (6). This new term is useful to describe the fundamental properties of the cepstrum. For example, low quefrencies correspond to slowly varying components in the logarithmic magnitude spectrum, while high quefrencies correspond to rapidly vary components of the logarithmic magnitude spectrum. To apply the derived mathematical function of cepstrum in the melody pitch determination, the DFT, which is computed with a Fast Fourier Transform (FFT) algorithm, is used with a sample in frequency version of the DFT of a finite-length sequence [1], [3]. The windowed signal is then transformed into frequency domain to
170
C.L. Si Toh, C.K. Ng, and N.K. Noordin
obtain the spectrum that is contained in the windowed signal by using DFT. The function of signal, X[k] after the DFT is defined as N −1
X [ k ] = x[ n]e −i ( 2πk / N ) n
(8)
n =0
The logarithmic magnitude of X[k] is computed for power spectrum. The function of logarithm X[k], Xˆ [k ] is fixed to
Xˆ [k ] = log X [k ] + i arg{X [k ]}
(9)
The function of cepstrum for x[n] discrete signal, Xˆ [n] where the DFT in (6) has been replaced by the finite DFT computation which can be derived as
1 N −1 Xˆ [ n] = Xˆ [ k ]e i ( 2πk / N ) n N n =0
(10)
After the substitution of (8) into (9) and (9) into (10), the predefined function of cepstrum such as (5) can be obtained. Figure 4 depicts the operations of cepstrum [3].
Fig. 4. The separation of vocal source and vocal tract
The cepstrum consists of cepstral peak occurring at a high quefrency which equal to the pitch period in seconds and low quefrency information corresponding to the formant structure in the logarithm spectrum [8]. To obtain an estimation of the fundamental frequency from the cepstrum, a peak in the quefrency region corresponding to acoustic signal fundamental frequencies is observed. In Fig. 5, an example of fundamental frequency estimation from the cepstrum of acoustic signal frame is presented. The cepstral peak detector uses the full-band of acoustic signal for processing. Each block of 240 samples is weighted by a 240-point Hamming window and the cepstrum of that block is computed. The cepstral peak value and its location is determined. If the value of this peak exceeds a fixed threshold, the section is called voiced and the pitch period is the location of the peak. If the peak does not exceed the threshold, a zero-crossing count is made on the block. If the zero-crossing count exceeds a given threshold, the block is marked as unvoiced. Otherwise, it is called voiced and the period is the location of the maximum value of the cepstrum [3].
Developed Automatic Musical Note Translating System
171
Fig. 5. The waveform of (a) an acoustic signal segment and (b) its real cepstrum
A 512-point FFT was found sufficient to produce an accurate computation of the cepstrum. The cepstral peaks corresponding to the voiced segments are clearly resolved and quite sharp. Hence, the peak picking scheme is exploited to determine the cepstral peak in the interval of 2.5 - 15 ms, which corresponding to pitch frequencies between 60 - 400 Hz [4]. The result of cepstrum computation is a time sequence, just as the input signal itself. If the input signal has a strong fundamental pitch period in the region interested, the pitch period will show up as a peak in the cepstrum. The fundamental period of this pitch can be determined by calculating the time distance from time 0 to the time of peak. Figure 6 depicts the pitch determination from cepstrum plot. From the cepstrum plot for a recorded acoustic signal, the cepstral peak that marked by an asterisk is found at 2.52 ms. This indicated the signal will be cycled at a period of 2.52 ms and thus the fundamental frequency of this recorded note is 396 Hz.
Fig. 6. An example of pitch extraction
172
C.L. Si Toh, C.K. Ng, and N.K. Noordin
4 System Development The developed automatic musical note translating system is initiated with the analog to digital signal conversion of the input melody. Once the analog signal is sampled and converted to digital signal, the sampled data of digital signal is stored as information that carries the characteristic of the fundamental frequency of the captured melody. These sampled data is then retrieved for the pitch determination by using the cepstral analysis. The cepstral analysis extracts the pitch of a concerned acoustic signal region by determining the peak of the fundamental frequency in the cepstrum. After the pitch value is determined, the corresponding musical note will be synthesized according to its standard reference and displayed on a developed musical stave as its interface template. In order to make the system performance more precise, the evaluation of sound level and voice activity detection including the background noise filtering will be taken into consideration to assure only the effective voice region are processed in the cepstral analysis. Therefore, the musical translating system consists of five main stages such as sound data collection, voiced region determination, fundamental frequency extraction, pitch matching and normalization, and autoshape coordinate allocation as shown in Fig. 7. Voiced Region Determination
Sound Data Collection
Fundamental Frequency Extraction
Autoshape Coordinate Allocation
Pitch Matching & Normalization
Fig. 7. System overview of the developed musical translating system
4.1 Sound Data Collection The preferable data quantization will be in 16-bit for better digitization and reducing lost of data. It is also preferable to record the melody in an environment with less noise signal to ensure the characteristic of the sampled sound data is not being attenuated such that it retains as much as possible of the original sound data characteristic. The sources of collected melody could be from various inputs either human vocal singing or various types of musical instruments such as keyboard, string, brass and woodwind. The collected lengthy digitized sound data is then segmented into smaller frame with the length of power of two of FFT points before preceded to the pitch analysis. 4.2 Fundamental Frequency Extraction Melody can be a harmonic signal which contains more than one spectral peak. Cepstral analysis is implemented to determine the fundamental frequency of the
Developed Automatic Musical Note Translating System
173
segmented melody sound data. The time-domain sound data is transformed into frequency-domain using FFT algorithm with the purpose of analyzing the sound spectrums. Since the sound data is a harmonic signal, the resulted spectrums are further analyzed to restrict its power values for determining the cepstral peak. By determining the restricted frequency with the highest power, the fundamental frequency of the windowed region can be found. 4.3 Pitch Matching and Normalization The estimated fundamental frequency value of the segmented sound data rarely hits the exact predefined musical frequency for the precise pitching of musical note due to the digitization lost and noises. Thus, a frequency range is defined to allocate the estimated fundamental frequency to match with the proper musical note. The upper bound and lower bound of the frequency ranges for all standard musical notes are assigned by Upper bound = standard pitch frequency + diff
(11)
Lower bound = standard pitch frequency – diff
(12)
diff = Difference between two consecutives notes/2
(13)
where
The identical computed musical notes tend to repeat at every short distance of time as the sound data is segmented into a small frame for cepstral analysis. Thus, these computed musical notes will be normalized before being recognized and displayed on musical score template to avoid the redundant notation displays for same sound data. 4.4 Autoshape Coordinate Allocation The computed musical note from the previous stage is projected to be displayed on musical score template as its standard notation. The autoshapes such as circle, horizontal line are used to construct the musical notation elements. These autoshapes are drawn on axes accordingly to display the computed or recognized musical notes. 4.5 Voiced Region Determination In order to yield a precise system performance where only voiced signal will be involved in fundamental frequency analysis, the captured melody will be analyzed for voiced region. Voice activity detection (VAD) and sound intensity level evaluation will be implemented to extract only the effective voiced region are processed for the cepstral analysis. The VAD applies the concept of zero crossing rates to detect the active voice region. The rate of the framed sound data for crossing the zero level is used to determine whether the framed sound data is a voiced region or an unvoiced region. In order to determine the zero crossing rate of a framed sound data, the amplitude of both consecutive sound data within the frame is evaluated for its sign changes as z(n) =
1 2
N
sgn [x (m+ 1)] − sgn [x (m )] m =1
(14)
174
C.L. Si Toh, C.K. Ng, and N.K. Noordin
where x(m) ≥ 0 + 1 sgn[x(m)] = (15) − 1 x(m) < 0 A positive sound data will gives a +1 sign meanwhile a negative sound data gives a – 1 sign. For an unvoiced region, the sound data tends to cross the zero level at higher rate as the amplitude of unvoiced signal is changes frequently from negative to positive and vice versa. Thus, it tends to accumulate the summation of sign in (14) and yield to a high zero crossing rate. While for a voiced region, the sound data are either remain at positive amplitude or negative amplitude. This will yield a low zero crossing rate. Thus, by determining the zero crossing rate, the voiced and unvoiced region of melody can be determined. The sound intensity level for the framed sound data is set with a threshold level of 50 dBA. To determine the sound level of a captured melody, the frequency spectrum of the segmented melody is obtained by using FFT algorithm such that the sound data is processed in frequency domain.
5 Performance Evaluation of the Developed System To evaluate the performance of the developed system, a piano melody C4 is played and captured by the system. Figure 8 shows plots of the amplitude, spectrum, and cepstrum for the captured C4 note. From the cepstrum plot, it shows that the peak is located at 0.003902s. Thus the fundamental frequency for the captured C4 signal is 1/0.003902 = 256 Hz, which is in the predefined domain of C4 signal. When the melodies of C4 D4 E4 F4 G4 A4 B4 and C5 are played and captured by the system. The sound data are analysed and its results in term of amplitude and frequency over timeline are shown in Fig. 9. The computed fundamental frequency value of these melodies are then matched with the predefined frequency domain and its corresponding musical notes are displayed as shown in Fig. 10. The evaluated accuracy performance of the system is environment dependent. The silent surrounding tends to yield a higher accuracy rate because the noise effect on acoustic signal very low. Figure 11 shows the accuracy performance for each note of melodies with ten trials.
Fig. 8. The amplitude, spectrum, and cepstrum plots for the captured C4 note
Developed Automatic Musical Note Translating System
175
Fig. 9. The amplitude and frequency over timeline of the captured C4 D4 E4 F4 G4 A4 B4 and C5 melodies
Fig. 10. The displayed musical notes of the captured C4 D4 E4 F4 G4 A4 B4 and C5 melodies
Fig. 11. The accuracy performance for each musical note of the captured melodies
176
C.L. Si Toh, C.K. Ng, and N.K. Noordin
An experiment has been conducted for the different sampling rates and the length of FFT points. The sampling rates of 11025, 22050, 44100 and 48000 Hz are exploited, while the length of FFT points that has been considered are 4096, 8192, 16384, 32768 and 65536. There is a hypothesis that the high performance computer with higher speed computing rates may produce high accuracy performance. Hence, an experiment has been done from a computer of 2.20 GHz microprocessor speed with different RAMs of 1, 2, 3 and 4 GHz. However, findings show that the accuracy and delay performances of musical notes matching are almost similar regardless of the different size of RAMs as shown in Figs. 12 and 13. It shows that the existing computing speed is sufficient to the developed automatic musical note translating system. Figure 12 shows the percentage of accuracy over different sampling rates. It shows that the accuracy is increase with higher sampling rates. At higher sampling rates especially 44100 Hz, the captured melody is being sampled at higher rate and thus it allows higher degree of preservation for the melody details. A more precise pitch value would be yield for segmented sound data at higher sampling rates. The results show that the optimum sampling rate that produces highest accuracy is 44100 Hz rather than 48000 Hz. Figure 13 shows the computational time over different length of FFT points. When the FFT length is longer, the system will load more sound data before start to perform FFT algorithm after all points are valid. Although the FFT algorithm computes DFT in an easier way and avoids redundant calculation, yet it still consumes some time and it becomes more significant for larger number of FFT points. Thus, lower FFT point length will yield shorter computational time. However, the results in Fig. 13 show that the 44100 sampling rate sound data needs higher computational time than 48000 sampling rate sound data. This is because the 44100 sampling rate sound data consumes more time such that higher accuracy of musical notes being translated can be achieved.
Fig. 12. The accuracy rate across different sampling rate for RAM of 2 GHz
Developed Automatic Musical Note Translating System
177
Fig. 13. The computational time over different length of FFT points for RAM of 2 GHz
6 Conclusions The automatic musical note translating system has been developed and its performance has also been evaluated. Basically, this system is developed to display the musical note automatically for a captured melody by determining the fundamental frequency of the signal. The performance evaluations of the developed system show that the optimized performance can be achieved at 44100 Hz sampling rate. It indicates the 44100 Hz is the optimum sampling rate for all audio frequencies that human can hear. Nevertheless, the developed musical translating system is best being implemented in noiseless environment as the acoustic signal is being less attenuated by noise signal and thus yields a better result.
References 1. Jo, J.Y., Kim, Y.H., Kang, H.J., Lee, J.S.: Chord-Based Musical Composition and Incorporating it into Auto-Accompaniment Instrument. In: Future Generation Communication and Networking (FGCN 2007), vol. 2, pp. 429–432 (2007) 2. Coyle, E.J., Shmulevich, I.: A System for Machine Recognition of Music Patterns. In: Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, vol. 6, pp. 3597–3600 (1998) 3. Bogert, B.P., Healy, M.J.R., Tukey, J.W.: The Quefrency Alanysis of Time Series for Echoes: Cepstrum, Pseudo Autocovariance, Cross-Cepstrum and Saphe Cracking. In: Rosenblatt, M. (ed.) Proceedings of the Symposium on Time Series Analysis. ch.15, pp. 209–243. Wiley, New York (1963)
178
C.L. Si Toh, C.K. Ng, and N.K. Noordin
4. Ahmadi, S., Spanias, A.S.: Cepstrum-Based Pitch Detection using a New Statistical V/UV Classification Algorithm. IEEE Transactions on Speech and Audio Processing 7(3), 333–338 (1999) 5. Ference, M., Lemon, H.B., Stephenson, R.J.: Analytical Experimental Physics, 3rd edn. The University of Chigago Press (1956) 6. Mastorakis, N.E., Gioldasis, K.D., Koutsouvelis, D., Theodorou, N.J.: Study and Design of An Electronic Musical Instrument which Accurately Produces The Spaces of The Byzantine Music. IEEE Transactions on Consumer Electronics 41(1), 118–124 (1995) 7. Tokuda, K., Kobayashi, T., Imai, S.: Adaptive Cepstral Analysis of Speech. IEEE Transactions on Speech and Audio Processing 3(6), 481–489 (1995) 8. Noll, A.M.: Cepstrum Pitch Determination. Journal of the Acoustical Society of America 41(2), 293–309 (1967)
Motion Tweening for Skeletal Animation by Cardinal Spline Murtaza Ali Khan1 and Muhammad Sarfraz2 1
2
Department of Computing Science Royal University for Women Bahrain [email protected] Department of Information Science Kuwait University Kuwait [email protected]
Abstract. Motion tweening (Inbetweening) is the process of generating intermediate frames between keyframes to create an illusion of motion. Motion tweening is a key process to generate computer based animations. This paper presents a simple and efficient method of tweening of motion data for skeletal animations. The proposed method generates smooth animation using cubic Cardinal spline. Keyframes are taken as control points and spline interpolation is performed to generate the in between frames smoothly. In order to facilitate the input, the keyframes can be manually specified or automatically detected from the motion data. Keywords: keyframes, animation, motion, tweening, cardinal spline, interpolation, control points.
1 Introduction Motion data is extensively used in video games, computer animation, and animated videos. Interestingly out of the three nominees for the 2006 Academy Award for Best Animated Feature, two of the nominees (Monster House and Happy Feet) used motion capture data [1]. The importance of using motion data in several fields is growing rapidly. This requires to investigate better methods to manipulate motion data. The aim of this research is to find a simple and efficient method of motion tweening using Cardinal spline. In our work, we focused on tweening of motion data that belongs to skeletal animations. Even though the method can be applied to other types of motion data as well. Motion tweening uses keyframes with an interpolation scheme to construct new frames. This essentially requires a set of meaningful, high-level control knobs (control points) to the animator or system. Motion tweening must be efficient for use in a run-time environment. This paper describes a technique that is very efficient and works well in real-time for motion tweening using Cardinal spline. A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 179–188, 2011. © Springer-Verlag Berlin Heidelberg 2011
180
M.A. Khan and M. Sarfraz
Spline and curve are widely used in computer-aided design and computer graphics because of the simplicity of their construction, accuracy of evaluation, and their capability to interpolate and approximate complex shapes [16], [3]. Spline interpolation is used to generate additional points given few sample points (control points/keypoints). Due to continuity of Cardinal spline, it is well suited for generating smooth animation. In addition to that implementation of Cardinal spline is simple and efficient and can be employed for real-time systems. Organization of the rest of the paper is as follows. Related work is discussed in Section 2. Section 3 describes basic concepts and terminology about animation and motion data. Construction and interpolation of Cardinal spline is explained in Section 4. Section 5 is the most important section and it describes the details of motion tweening strategy of the proposed method. Simulation and results are presented in Section 6. Final concluding remarks are in Section 7.
2 Related Work A good reference on modeling, capture, and animation of human motion is [15]. This book contains articles contributed by several researchers. General Cardinal spline interpolation schemes are investigated by many authors [13], [9], [4], [10]. A scheme to construct new motions from existing motions is proposed in [14]. This scheme used low order polynomials and B-splines functions to create the interpolation space between example motions. Our method does not generate new motion from existing motions but generates motions from keyframes. The method proposed in [12] is about geostatistics based motion interpolation. This method treats motion interpolations as statistical predictions of missing data in an arbitrarily definable parametric space. However, the method of [12] has limitations, since the number of necessary samples usually increases exponentially in the dimension of a control space. The work in [17] used a local spline keyframe interpolation method for generating animations, but the method is applicable to only 2D animations such as cartoons in comics. Our method is not restricted to 2D animations and can be used for generating both 2D and 3D animations. A method to generate a smooth morphing between the source and the target images by drawing input curves on the features of the objects is presented by [8]. This method uses B´ezier curves and shape interpolation techniques . Our method uses Cardinal spline interpolation rather than B´ezier curves and does not require to draw any input curve.
3 Basics of Animation and Motion Data Following is the basic terminology used to describe different aspects of a animation, motion data and motion data file format. – Skeleton: It is a whole character on which motion data is applied to create animation. Figure 1 shows a skeleton of human figure [5]. – Bone: A skeleton is comprised of a number of bones. Each bone represents the smallest segment within the motion that is subject to individual translation and/or rotation changes during the animation.
Motion Tweening of Skeletal Animation
181
– Channel or Degree of Freedom (DOF): Each bone within a skeleton can be subject to translation, rotation and scale changes over the course of the animation, where each parameter is referred to as a channel or degree of freedom (DOF). The changes in the channel data over a period of time produce animation. Translation and scale changes are measured in unit of length, while rotational changes are measured in degrees or radians. – Frame/Frame-rate: Every animation is comprised of a number of frames where for each frame the channel data for each bone is defined. Frame-rate refers to number of frames generated in one second. Generally frame rate varies between 24 to 60 frames per second (fps).
Fig. 1. A skeleton of human figure
We can plot the motion data of a joint for each channel separately such that frame numbers are along horizontal axis and translation or rotation values are along vertical axis. This kind of plot is called motion curve of motion signal. Figure 2 shows motion curve of right-shoulder joint (rotation along z-axis) in 153 frames. Several file formats exist to store the motion data [11]. We used BVH (Bio Vision Hierarchy) file format. This format is simple and supported by many motion capture devices and animation programs. A BVH file has two sections, a header section which describes the initial hierarchy of the skeleton; and a data section which contains the motion data. BVH can store motion for a hierarchical skeleton. This means that the motion
182
M.A. Khan and M. Sarfraz
of the child bone is directly dependent on the motion of the parent bone. Figure 3 shows hierarchy of joints for a segment of a skeleton. For example in Figure 3, Hips joint has children LeftHip, RightHip, and LowBack. LeftHip has a child joint LeftKnee. LeftKnee has a child joint LeftHeel, etc. Figure 4 shows how BVH file format stores hierarchical skeleton. The motion section of a BVH file contains number of frames, frame rate and the actual motion data. Each line is one sample (one frame) of motion data. −15 −20
Motion Data (Degree)
−25 −30 −35 −40 −45 −50 −55 −60 −65
1
16
32
48
64 80 96 Frame Numbers
112
128
144 153
Fig. 2. Motion curve of right shoulder joint (rotation along z-axis) in 153 frames
4 Cardinal Spline This section describes the construction of parametric cubic Cardinal spline. Cardinal spline is a C1 continuous curve. Cardinal spline interpolates piecewise cubics with specified endpoint tangents for each segment. A Cardinal spline segment, as shown in Figure 5, is defined by four control points, i.e., Pj−1 , Pj , Pj+1 and Pj+2 . The jth segment of Cardinal spline interpolates between two middle control points, i.e., Pj and Pj+1. The end control points, i.e., Pj−1 and Pj+2 are used to calculate the tangents at Pj and Pj+1 . Equations for boundary conditions of jth segment are written as: 1 (1 − T ) Pj+1 − Pj−1 , 2 1 = (1 − T ) Pj+2 − Pj , 2
Pj = Pj+1
where parameter T is Tension and it controls looseness/tightness of spline.
(1) (2)
Motion Tweening of Skeletal Animation
Fig. 3. Hierarchical skeleton
Fig. 4. Hierarchical skeleton in BVH file format
183
184
M.A. Khan and M. Sarfraz 22
Pj+2
Cardinal Spline Control Points
Pj+1
20
18
16
14
P
j
12
10 P 8 15
j−1
20
25
30
35
40
Fig. 5. The jth Cardinal spline segment, T = 0
For l joined segments, there are 2l conditions for continuity of functions and 2l conditions for continuity of slopes. Finally the equation of Cardinal spline for jth segment is written as follows: Q(ti ) =(−sti3 + 2sti2 − sti )Pj−1 + [(2 − s)ti3 + (s − 3)ti2 + 1]Pj + [(s − 2)ti3 + (3 − 2s)ti2 + sti ]Pj+1
(3)
+ (sti3 − sti2)Pj+2 , where ti is parameter of interpolation , 0 ≤ ti ≤ 1, and s is related to Tension by s = (1−T ) 2 . In order to generate n points between Pj and Pj+1 inclusive, the parameter ti is divided into (n − 1) intervals between 0 and 1 inclusive, and Q(ti ) is evaluated (interpolated) at n values of ti .
5 Tweening Strategy This section elaborates the strategy of motion tweening. Our objective is to generate smooth animation using few keyframes. We will describe the strategy of motion tweening of a single channel of a joint. Same strategy is applied for all the channels of all the joints. Initially the user has to specify the keyframes. These keyframes are taken as control points of Cardinal spline. There are three possibilities to specify keyframes which are as follows:
Motion Tweening of Skeletal Animation
185
1. Keyframes can be specified from the existing motion data at regular interval, no human intervention is required. 2. User manually input/select the keyframes at non-regular interval, human intervention is required. 3. Feature point extraction algorithm is use to specify keyframes at non-regular interval, no human intervention is required. Several feature point extraction algorithms based on curvature analysis are proposed in the literature, such as in [2], [6], [7], etc. We modified the algorithm of [6] to extract the keyframes from motion data. Curvature based algorithms are sensitive to some threshold value, specified as an input. We took default values of all the input parameters specified in [6]. Once the keyframes are specified/extracted then Cardinal spline interpolation is used to generate in between frames between keyframes. A single Cardinal spline segment is constructed between each adjacent pair of keyframes by taking adjacent keyframes as middle control points, i.e., Pj and Pj+1 of Cardinal spline. Therefore, for n keyframes the number of Cardinal spline segments are n − 1. Since, in addition to control points of the current segment, Cardinal spline needs control points of previous and next segments (see Eq. 3). Therefore, we used keyframes of previous, current and next segments as control points of a current Cardinal spline segment and obtained the interpolated data of current segment using Eq. (3). An interesting question is how to obtain the keyframes of first and last segments? Because they do not have previous and next segments respectively. We opted to take P−1 = P0 , for the first segment and Pl+2 = Pl+1 , for the last segment. All the keyframes and their in-between frames constitute a complete motion curve of the channel. Figure 6 shows the keyframes and motion curve of right-shoulder (rotation along z-axis) obtained by Cardinal spline interpolation between keyframes. Note that this motion curve is similar but smoother than Figure 2, where no motion tweening was used. It is also worth to mention that in the proposed method, only keyframes are needed to be saved, while in-between frames are generated on the fly using keyframes. Where as in the conventional systems, all the frames need to be saved. Algorithm 1 summarizes the motion tween strategy.
Algorithm 1. Motion Tweening 1: Extract or specify the keyframes. First and last frames are always taken as keyframe. 2: Specify number of in between frames between every pair of keyframes. 3: Take each pair of adjacent keyframes as middle control points, i.e., Pj and Pj+1 of Cardinal spline. Take previous and next keyframes as first and last control points (Pj−1 and Pj+2 ) respectively. 4: Obtain the motion tweening data by Cardinal spline interpolation between every pair of keyframes Pj and Pj+1 . 5: If the result is not satisfactory then go to step 1, add more keyframes and repeat the process. 6: Save the keyframes.
186
M.A. Khan and M. Sarfraz −10
Interpolated fMotion Data (Degree)
Interpolated Motion Data Keyframes −20
−30
−40
−50
−60
−70
1
16
32
48
64 80 96 Frame Numbers
112
128
144 153
Fig. 6. Interpolated data (Motion tweening) with keyframes for right shoulder joint
6 Simulation and Results Simulation is done on motion capture data of various animation sequences. Table 5 provides details about environment of simulation. Even though the proposed method needs only keyframes. But in order to see how effectively the proposed method performs tweening that resembles original motion, we took complete sequences and extracted the keyframes from these sequences then performed the motion tweening on the keyframes. Table 2 gives the details of two animations namely Ladder-Climb and Walk-Turn used Table 1. Environment of simulation Hardware Inter Core Duo 2.4 GHz Operating System Linux (Ubuntu 10.04 LTS) Programming Language MATLAB R2010
Table 2. Animations used in the simulation Animation Name Frame Rate (fps) Number of Frames Ladder-Climb 30 153 Walk-Turn 24 119
Motion Tweening of Skeletal Animation
187
Fig. 7. Frame 133 obtained by motion tweening of keyframes of Ladder-Climb animation
Fig. 8. Frame 7 obtained by motion tweening of keyframes of Walk-Turn animation
in the simulations. In the Ladder-Climb animation, a character climbs on a ladder. In the Walk-Turn animation, a character first walks straight then turn back 180 degree and walks again. Figures 7 and 8 show the frames obtained by motion tweening of LadderClimb and Walk-Turn keyframes respectively.
188
M.A. Khan and M. Sarfraz
7 Conclusion Motion data is widely used in many applications such as video games, computer animation, and animated videos. Motion tweening is an essential process to generate computer based animations using keyframes. We presented a simple and efficient method of tweening of motion data to generate smooth animation using cubic Cardinal spline. For each channel of a joint, keyframes are specified manually or extracted automatically as an input to the method. The keyframes serve as control points of Cardinal spline to generate in between frames by Cardinal spline interpolation. We presented the algorithm of motion tweening using Cardinal spline. The proposed method is well suited to generate skeletal animations. Simulation results show that the proposed method yields very good results for variety of animations.
References 1. http://www.videocapturehardware.org/video-capture-artifacts/ 2. Awrangjeb, M., Lu, G.: Robust image corner detection based on the chord-to-point distance accumulation technique. IEEE Transactions on Multimedia 10(6), 1059–1072 (2008) 3. Bartels, R.H., Beatty, J.C., Barsky, B.A.: An Introduction to Splines for Use in Computer Graphics and Geometric Modeling. Morgan Kaufmann (1995) 4. Bejancu, A.: A new approach to semi-cardinal spline interpolation. East Journal on Approximations 6(4), 447–463 (2000) 5. http://www.cs.wisc.edu 6. He, X.C., Yung, N.H.C.: Curvature scale space corner detector with adaptive threshold and dynamic region of support. In: Proceedings of the 17th International Conference on Pattern Recognition (ICPR 2004), vol. 2, pp. 791–794. IEEE Computer Society, Washington, DC, USA (2004) 7. He, X.C., Yung, N.H.C.: Corner detector based on global and local curvature properties. Optical Engineering 47 (2008) 8. Johan, H., Koiso, Y., Nishita, T.: Morphing using curves and shape interpolation techniques. In: Proc. Of The Pacific Graphics 2000 (PG 2000), 1 Rao, And Shah / View Interpolation, pp. 348–358 (2000) 9. Kochanek, D.H.U.: Interpolating splines with local tension, continuity, and bias control. Computer Graphics 18(3), 33–41 (1984) 10. Silbermann, M.J., Tagare, H.D.: Local cardinal spline interpolation and its application to image processing. In: Proceedings of SPIE, pp. 272–283 (February 1992) 11. Menache, A.: Understanding Motion Capture for Computer Animation. Morgan Kaufmann (2010) 12. Mukai, T., Kuriyama, S.: Geostatistical motion interpolation. In: ACM SIGGRAPH 2005 Papers, pp. 1062–1070. ACM, New York (2005) 13. Renka, R.J.: Algorithm 716: TSPACK: Tension spline curve-fitting package. ACM Transactions on Graphics (TOG) 19(1), 81–94 (1993) 14. Rose, C., Bodenheimer, B., Cohen, M.F.: Verbs and adverbs: Multidimensional motion interpolation using radial basis functions. IEEE Computer Graphics and Applications 18, 32–40 (1998) 15. Rosenhahn, B., Klette, R., Metaxas, D.: Human Motion: Understanding, Modelling, Capture, and Animation. Springer, Heidelberg (2007) 16. Sarfraz, M.: Interactive curve modeling with applications to computer graphics, vision and image processing. Springer, Heidelberg (2008) 17. Sun, N., Ayabe, T., Okumura, K.: An animation engine with the cubic spline interpolation. In: International Conference on Intelligent Information Hiding and Multimedia Signal Processing, pp. 109–112 (2008)
Preliminary Investigation on the Use of Augmented Reality in Collaborative Learning Wannisa Matcha and Dayang Rohaya Awang Rambli Computer and Information Department, Universiti Teknologi PETRONAS, Bandar Seri Iskandar, Tronoh, Perak, 31750, Malaysia [email protected], [email protected]
Abstract. This paper presents the result of the study for collaborative Augmented Reality (AR) based system. AR technology has been investigated for use as a medium for enhancing collaboration by adding virtual information to the real world and supporting the interaction in real time. However, this technology is still in its infant state especially in the education field. More studies are needed so as to provide the value information to this field. The main focus of this study is to investigate the potential of AR spaces such as markers in providing communication cues and supporting collaboration in learning environment. The techniques used in the study are through questionnaire, observation and think aloud protocol. Results from the study shows that AR supports the natural means of communication and interaction. This preliminary study strengthens the conjecture that AR has the potential to be used as a shared medium in collaborative learning. Keywords: Augmented Reality in Education, Computer-Supported Collaborative Learning, Collaborative Augmented Reality.
1 Introduction Collaborative Learning is one of the learning approaches which aim to help students to learn collaboratively in group in order to accomplish a given task [1]. There are several empirical studies which have shown positive outcome of the collaborative learning [2]. Moreover, the advancement of technologies had made it possible to enhance collaborative learning by employing information technology to support communication and collaboration. This approach is known as Computer-Supported Collaborative Learning (CSCL) [3]. Many researchers have long studied CSCL by applying different technologies such as the use of the internet technology to make it available at any time any places, the use of Virtual Reality (VR) to make lessons more interesting by immersed the user to the virtual world and simulate the scenarios. However, the rapid changes of technology make the current learning methods obsolete or less attractive to students [4]. Therefore, it is essential to continually explore new technologies or enhance the current ones to make the learning environment more attractive for the students [4]. Augmented Reality (AR) is an emerging technology which overlays virtual images into real world. It allows the user to manipulate the virtual object by interacting with the physical object and the result is displayed in real time [5]. AR offers new experience in A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 189–198, 2011. © Springer-Verlag Berlin Heidelberg 2011
190
W. Matcha and D.R.A. Rambli
learning which has never been possible before with other technologies. Collaborative AR is a term used to refer to AR systems which aim to support collaboration either in the workplace or learning environment, co-locate or remote distance. However, this research focuses on face-to-face collaboration in learning environment. Since this field is still in its infancy. There is a need for more usability studies to provide the value information to this field and to mature the technology. Its potential for single user has long been studied. But researches concerning how AR can support collaborative learning are few [6, 7]. Hence this research aims to fulfill that gap. The primary question in this study is do AR features support collaboration and provide communication cues for co-located collaborative learning? To provide collaboration space, AR needs to support communication cues both verbal and nonverbal such as gesture, gaze speech etc. Several evaluation techniques such as observation, questionnaire and open-ended comment have been applied in this study. The rest of the paper is organized as follows. The related work is presented in the next section. The evaluation context is briefly described in section 3 followed by the result of the study in section 4. The paper is concluded in section 5.
2 Related Work Even though AR has been in the research field for more than forty years, but its formal usability study has only recently begun [6]. While, its potential in education is just being explored [6]. There is still huge gap in the research, especially in collaboration, particularly in education field [6]. Based on the survey research by Dünser et al. in 2008, out of 161 papers, only 10 papers were focused on collaboration, and only 2 out of the 10 were in educational context [7]. However, recently the number of research in collaborative AR has increased. Some of the researches work that have evaluated collaborative AR interface in learning environment are discussed below. Construct3D was developed with the aim to facilitate face-to-face collaboration between teachers and students in geometry learning. Various user evaluations had been conducted [8]. The results of usability studies were used to improve the system. The study also reported that students felt that the system was more useful, more satisfying as compared to desktop application; users felt that the system can be used in real class. However, not much work had been done to evaluate how the system supports the collaboration and how it can be used as shared space in collaborative learning environment. Protein Magic Book (PMB) was used to investigate peer learning by comparing the effectiveness of learning between individual and peer group as well as between learning through AR and without AR technology condition. Using questionnaire and think aloud protocol, the study also aimed to explore the cognitive load between individual and peer learners. The preliminary result showed that students performed better in individual learning using AR. Since peer learning is a more complex situation, students required more time to discuss and learn the material together with their partners [9]. Asai et al. developed lunar surface navigation system for exhibition purpose which aimed to assist collaboration between parents and children by enhancing museum experience through the use of tabletop AR and virtual environment. The system was developed based on the quasi-role play that assumes children will act as astronauts
Preliminary Investigation on the Use of Augmented Reality in Collaborative Learning
191
while parents will act as mission commanders. A user study by the use of questionnaires and observation was conducted to investigate the properties of the lunar surface navigation system. The usability study was mainly focused on motivation, suitability, usability, comprehension and media capability of the system. The result of the study shows that the system was “playful” and had facilitated collaboration between parents and their children [10]. Most of the researches in this field have been concerned more on whether AR could be used as a tool in collaborative learning. Thus far, most of the studies have shown significant positive feedbacks. However, not much works has been conducted to address the concerns on how AR could support collaboration in detail, which features of AR that provide support for collaboration and which communication cues are provided by the AR technology.
3 Experimental Design 3.1 Method As mentioned by many researchers, there is no particular usability method that has been designed for Augmented Reality system [11, 12]. However, the necessity to investigate and evaluate the collaborative AR is important in order to mature the technology. Hence many researchers have tried to come out with a method for evaluation of AR system. According to Dünser et. al (2008), AR user evaluation can be classified into five categories which are objective measurements, subjective measurements, quantitative measurements, qualitative analysis, usability evaluation techniques and informal evaluations [7]. Similarly, Monk et al. also agreed that multidimensional approaches are needed in evaluating the usability of the system [13]. Hence, in this preliminary study subjective measurement and qualitative analysis adapted from the work of Kiyokawa et al. [14] which were used to investigate the communication behaviors of co-located users in collaborative AR for working environment have been applied. In this preliminary study, subjective user’s feedback and perception on using the system will be gathered through the use of questionnaires [7]. In quantitative analysis, a video recording of the experiment will be made to observe the user’s interaction with the system, and the collaboration occurs during the experimentation [7]. 3.2 The Prototype Augmented Reality Experiment (AReX) is an Augmented Reality prototype which aims to help students in learning the concept of light dispersion and combination through experiment. AReX allows students to interact with the system through the use of marker patterns to manipulate the virtual object and the outcome is displayed on a monitor screen. The AReX system includes a worksheet, marker patterns, a camera and a computer. The lab worksheet includes the instruction for conducting AR based experiment. The marker patterns are printed black and white square shape. Each marker represents the elements needed in light dispersion experiment. Whilst the
192
W. Matcha and D.R.A. Rambli
camera captures the markers and real environment; computer calculates the position, overlays the virtual image over the markers and displays the real environment and virtual object on the monitor screen as shown in figure 1.
Fig. 1. AReX prototype and experiment set up
3.3 Participants and Tasks The participants who volunteered in this experiment are students who have different backgrounds. Six of the participants are from engineering field while another six are from Computer Science Department. Neither one of the students knows the technology nor familiar with Augmented Reality technology. The students were grouped into pairs to test the system. The test was conducted with one pair at a time. Firstly, when the students arrived they were asked to fill the questionnaire regarding their science preferences. The questions are as shown in table 1. Table 1. Questions for science preference test No.
Questions
Q1
Do you like science subject?
Q2
Is it difficult to understand some lesson in class?
Q3
Do you have problem in learning physic in class?
Q4
Do you like doing experiment?
Q5
How often do you practice the lab experiment?
Q6
Do you prefer to do experiment with group members?
Q7
Do you want to repeat the experiment in any time you want?
Preliminary Investigation on the Use of Augmented Reality in Collaborative Learning
193
After they had completed the questionnaire, the students were briefed on AR technology and the AReX system. Next, they were given five to ten minutes to familiarize themselves with AR and the AReX system. Then the students will start doing the experiment according to the lab worksheet provided for them. They were given three tasks to complete. During the experiment, the students’ action will be observed and video recorded. After completing the experiment, the students were asked to fill in a second the questionnaire. The questions are as shown in table 2. Table 2. AReX experiment usability questions No.
Question Questions about visibility
Q1
The view of the real world was very natural Questions for partners
Q2
it was very easy to see my partner
Q3
I could very easily tell where my partner was looking
Q4
I could very easily tell where my partner was pointing Questions about the overall preferences
Q5
It was very easy to perform the task
Q6
I like the condition very much Questions about communication richness
Q7
I looked at my partner very often
Q8
I used pointing gestures very often
Q9
It was very easy to see my partner
Q10
I could very easily understand what my partner was doing
Q11
I could easily tell when my partner was looking at me
Q12
I could very easily tell where my partner was looking
4 Result and Analysis 4.1 Science Preference To evaluate user’s feedback regarding interaction with the AR system, the users’ preference in science is must be observed since it might have some effect on the system. The questions used to gather information about user’s preference in science learning are as shown in table 1. Figure 2 presents the result based on science preference answered by students before the experiment.
194
W. Matcha and D.R.A. Rambli
100% 90% 80% 70% strongly disagree
60%
disagree
50%
neutral agree
40%
strongly agree 30% 20% 10% 0% Q1
Q2
Q3
Q4
Q5
Q6
Q7
Fig. 2. Result of user preference test on science learning
The results show that students like science subjects. However, more than sixty percent had difficulty in learning science subjects. For physics, the result indicate that forty percent face learning difficulties such as understanding the concept etc. while more than seventy percent of the students like doing the experiment, but their rarely practice it. One of the comments from the students is that the only chance they have to practice lab experiments is during lab session, other than that it is difficult for them to practice. The results also indicated that more than ninety percent of the students prefer doing lab experiment in group since they can share information. While more than sixty percent of the students would like to repeat the experiment if it is available. Some students also commented that classroom environment can be quiet boring sometimes, especially when the lesson is about scientific concept. They also face problem in understanding the lesson since they cannot visualize the scientific concept of the topic.The result of the study indicates that science subject could be used as a case study for application of AR as a collaborative learning tool. Since the real practice in classroom, the lab is group based learning activities while the students also prefer to do experiment in group. Based on the feedbacks from the students, AR can be used to alleviate the problem in understanding scientific concepts and to provide a tool for practicing lab experiments prior to conducting actual experiments. 4.2 Collaborative AR The questionnaires used in this study are adapted from the work of Kiyokawa et al. which was aimed to evaluate the users’ behaviors for Collaborative AR in face-toface working environment [14]. The questions are as shown in table 2.
Preliminary Investigation on the Use of Augmented Reality in Collaborative Learning
195
6
5
Scores
4
3
2
1
0 Q1
Q2
Q3
Q4
Q5
Q6
Q7
Q8
Q9
Q10
Q11
Q12
Questions
Fig. 3. Average score of AR system as a collaborative shared space
Students were asked to fill the questionnaire after the experiment (1-strongly disagree, 5 strongly agree). Figure 3 presents the average scores of AR space in supporting the collaboration for light dispersion experiment. The first question (Q1) concerns the visibility of the real world. Most of the users feel that the view of the real world in this experiment is natural since they can easily see their partner. Second part of the questions (Q2-Q4) focused on the users’ feedback on collaboration. The results show that it was not difficult for the students to tell where their partner was currently looking at and where he/she was pointing at. The next questions (Q5&Q6) are focused on user’s opinion in using the AReX application. Most of the users feel that it was easy to perform the task and they liked doing the experiment using AR technology.
Fig. 4. Users discussing the experiment
196
W. Matcha and D.R.A. Rambli
The last part of the questionnaire (Q7-Q12) was aimed to gather information about the communication richness. The result show that most of the users often look at their partner even though there were some users who did not look at their partner at all since they paid most of the attention to the task space. This is in line with the work of Kiyokawa et al. that in wall case which similar to desktop application, the communication separation seam might draw away the users’ attention to the application more than their partner. However, users use point gestures quiet often. It was also observed that the users used point gestures quiet often even though some argued that they did not use point gestures that often. However, most of the users agree that they can easily and naturally see their partners. And they can easily understand what they partner was doing The result also show that the use of monitor based display creates the separation seam because they can become too focused on the task in the screen; some of them ignored their partner and could not feel when their partners were looking at them. Based on the observation of the video recording, most of the users focused on the monitor display and they seldom look at their partner. But pointing gestures were used to support communication; they used words like “this”, “that” and point to communicate. Students tend to use their hands to explain their understanding to their partners such as when describing the shape of prism that affects the order of colors. Some of them also used their hands to refer to the different layers of the colors. Based on the result, it is obvious that AR supports various communication cues such as speech, pointing gestures, and gaze. However, the use of monitor seems to draw the user’s attention more on the screen and the application, and less attention is paid to their partner. Based on the think aloud protocol and open-ended discussion after the experiment, the students said that the AR technology has managed to grab their attention and it looks interesting. The users felt excited with the technology as reflected in their comments: “wow! How can the images appear on the paper? “That’s fantastic” Some of the users felt like they were really holding the object but when they tried to touch it, they cannot touch that object. This is reflected in their comment as such “I feel real, like I am holding it”. While some of the users still prefer the real experiment as compared to AR experiment. However, they agreed that the AR experiment will be very useful to explain or demonstrate the concept of the lessons. Some of their comments are: “It will be good if the lecturers use it during class because it is always difficult to figure out the concept at the first time”. “If the experiment cannot be conducted in the lab, this AR application can be a really good alternative of doing experiment”. While some other users commented that they would love to have this online so that they can always do the experiment and explore the concept of the lesson. “I love to have it, so that I can always practice and re-do the experiment”. “It will be good if I can do the experiment in my own room and whenever I want”.
Preliminary Investigation on the Use of Augmented Reality in Collaborative Learning
197
5 Discussion Most of the average scores of the usability study for AReX project were rated more than 3.5 which indicates that the system supports collaboration by supporting the visibility of real world (Q1), collaboration between partners (Q2-Q4), user satisfactions (Q5-Q6) and communication cues (Q7-Q12). However, among the entire question, Q7 received 3.25 which is the lowest score. Q7 concerns on how often the user looks at their partner. Even though most of the users agree on that but there are some disagreement. This result is inline with the work of Billinghurst et al. (2002). He claimed that the desktop application might create communication separation from the task space [15]. Since the use of monitor seemed to draw the user’s attention more to the screen and the application, thus less attention is paid to their partner. This separation also affected the score of Q11 and Q12 to be at 3.50. Q2, Q5 and Q6 received that highest average score. Q2 addresses if the users feel easy to see their partner. Q5 and Q6 are concerned with the ease of performing and users’ satisfactions. This result is inline with the work of Kiyokawa et al. He claimed that sharing the same viewpoint ease the communication [14]. Based on the open-ended comment after the experiment, the students said that the AR technology grabs their attention. The students like the AR technology and AReX project itself. This might be because the students nowadays are technologist. They are familiar with the use of technology in real life. Hence they can easily adapt to the new technology. Even though AR is not a new technology for researchers, but for educators and students, AR is still considered as new. The students who participated in the experiment got excited with the new technology. They paid attention when working on the AReX system. The students remarked that it will be good if the teacher uses this technology particularly in teaching the scientific conceptual. This is because AR can simulate virtual objects in real time. With this features, students can easily visualize and learn the concept. However, as AR is still a new technology in the educational field, the AR prototype needs more improvement in information providing, interaction and interfaces.
6 Conclusion Results preliminary study on collaborative AR in science learning environment has shown that AR supports various communication cues. Students like the condition of AR system. However, the prototype needs to be improved more on instruction and information provision. The current prototype does not support much of information, less explanation on the state occurred in the experiment and the consequences of the experiment. However, the AR system shows a significant support for collaborative learning environment by supporting various types of communication cues.
References 1. Smith, B.L., MacGregor, J.T.: What is Collaborative Learning? In: Goodsell, A., Maher, M., Tinto, V., Smith, B.L., MacGregor, J. (eds.) Collaborative Learning: A Sourcebook for Higher Education, the National Center on Postsecondary Teaching, Learning, and Assessment, Pennsylvania State University (1992)
198
W. Matcha and D.R.A. Rambli
2. Lu. W.: GroupSpace as Support for Collaborative Learning (1994), http://ldt.stanford.edu/~luwuping/study/ portfolio/GroupSpace.pdf 3. Ludigsen, S.R., Mørch, A.I.: Computer-Supported Collaborative Learning: Basic Concepts, Multiple Perspectives, and Emerging Trends. In: McGaw, B., Peterson, P., Baker, E. (eds.) The International Encyclopedia of Education, 3rd edn., Elsevier (in press) 4. Martín-Gutiérrez, J., Contero, M., Alcañiz, M.: Evaluating the usability of an augmented reality based educational application. In: Aleven, V., Kay, J., Mostow, J. (eds.) ITS 2010, Part I. LNCS, vol. 6094, pp. 296–306. Springer, Heidelberg (2010) 5. Azuma, R.: A Survey of Augmented Reality. Presence: Teleoperators and Virtual Environments 6(4), 355–388 (1997) 6. Zhou, F., Duh, H., Billinghurst, M.: Trends in Augmented Reality Tracking, Interaction and Display: A review of Ten years of ISMAR. In: IEEE international symposium on Mixed and Augmented Reality, Cambridge, UK, September 15-18, pp. 193–202 (2008) 7. Dünser, A., Grasset, R., Billinghurst, M.: A survey of evaluation techniques used in augmented reality studies. In: ACM SIGGRAPH ASIA 2008 courses on - SIGGRAPH Asia 2008, pp. 1–27 (2008) 8. Kaufmann, H., Schmalstieg, D.: Mathematics and Geometry in Education with Collaborative Augmented Reality. In: ACM SIGGRAPH 2002 Conference, New York, pp. 37–41 (2002) 9. Chen, Y.C.: Peer learning in an AR- based learning environment. In: 16th International Conference on Computers in Education, Taipei, Taiwan, pp. 291–295 (2008) 10. Asai, K., Sugimoto, Y., Billinghurst, M.: Exhibition of lunar surface navigation system facilitating collaboration between children and parents in science musuem. In: VRCAI 2010, Seoul, South Korea, December 12-13, pp. 119–124 (2010) 11. Balog, A., Pribeanu C., Iordache, D.: Augmented Reality in Schools : Preliminary Evaluation Results from a Summer School. In: Engineering and Technology, pp. 114-117. (2007) 12. Bach, C., Scapin, D.L.: Obstacles and perspectives for evaluating Mixed Reality system usability. Presented at Workshop MIXER “Exploring the design and engineering of MR system”, IUI-CADUI 2004, Funchal, Island of Madeira, Portugal (2004) 13. Monk, A.F., McCarthy, J., Watts, L., Daly-Jones, O.: Measures of process. In: Thomas, P.J. (ed.) CSCW requirements and evaluation, pp. 125–139. Springer, Heidelberg (1996) 14. Kiyokawa, K., Billinghurst, M., Hayes, S.E., Gupta, A., Sannohe, Y., Kato, H.: Communication behaviors of co-located users in collaborative AR interfaces. In: International Symposium on Mixed and Augmented Reality, pp. 139–148 (2002), http://ieeexplore.ieee.org/lpdocs/epic03/ wrapper.htm?arnumber=111508 15. Billinghurst, M.: Shared Space: Explorations in Collaborative Augmented Reality. Ph.D. Dissertation, University of Washington (2002)
Development of 6LoWPAN Adaptation Layer with Fragmentation and Reassembly Mechanisms by Using Qualnet Simulator Chiaw Wei Chan, Gee Keng Ee, Chee Kyun Ng, Fazirulhisyam Hashim, and Nor Kamariah Noordin Department of Computer and Communication Systems Engineering, Faculty of Engineering, University Putra Malaysia, UPM Serdang, 43400 Selangor, Malaysia [email protected], [email protected], {mpnck,fazirul,nknordin}@eng.upm.edu.my
Abstract. This paper presents the development of 6LoWPAN adaptation layer with fragmentation and reassembly mechanisms by using Qualnet simulator. The developed adaptation layer is based on 6LoWPAN working group specifications and is added into the current existing Qualnet libraries. The IPv6 protocol stacks, however, are not currently supported and provided by Qualnet simulator. Thus, a modified version of IPv4 based addressing scheme to emulate the IPv6 protocol stacks for developing the 6LoWPAN adaptation layer has been successfully implemented and validated in Qualnet simulator. This developed adaptation layer has also been examined over the IEEE 802.15.4 standard WSN environment. The performance of the developed 6LoWPAN is evaluated and compared to WSN in terms of packet delivery ratio, throughput, average end-to-end delay and total average energy consumption. Keywords: 6LoWPAN, WSN, adaptation layer, IEEE 802.15.4, fragmentation, reassembly, Qualnet.
1 Introduction Wireless sensor network (WSN) IEEE 802.15.4 standard consists of many sensor nodes which are able to perceive changes in environment and perform certain actions. These sensor nodes are the tiny devices equipped with radio, microcontroller, sensors and battery. WSN is envisioned to be the future pervasive computing in different types of fields including environmental monitoring, military surveillance, and inventory tracking. Normally, the sensed information in WSN is passed to Internet and the control message returns to sensor nodes through a connection with an external network. Thus, the integrated sensor nodes with Internet will allow WSN to be widely deployed in many other fields later [1]. WSN becomes increasingly important especially in embedded applications. These applications require low cost, low power and low data nodes that communicating over A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 199–212, 2011. © Springer-Verlag Berlin Heidelberg 2011
200
C.W. Chan et al.
multiple hops to cover a large geographical area. In conjunction with the emergence of large scale sensor network, it requires survivability, stability, and mobility in sensor networks. Challenges will be present especially when large scale of sensor network is deployed in future. The small and low power sensor device in WSN is not suitable to be loaded with high resource Internet protocol (IP) capabilities. Since the IP version 4 (IPv4) 32-bit address space is unable to address problems associated with the rapid consumption of network address, the introduction of the IP version 6 (IPv6) 64-bit address space gives solution to rapid growth of Internet. Besides providing huge address spaces, IPv6 is also able to support the address’s auto-configuration and mobility. Hence, the IPv6 over low power wireless personal area network (6LoWPAN) is proposed to standardize the end-to-end communication between sensor nodes and external IP networks [2], [3]. One of the challenges in 6LoWPAN is how to adapt IPv6 packet over IEEE 802.15.4 standard WSN. The minimum size of IPv6 maximum transmission unit (MTU) is 1280 bytes. Since the IEEE 802.15.4 MAC layer can only support the maximum 127 bytes of data units, it fails to accommodate the 1280 bytes of IPv6 MTU. Hence, in order to provide a smooth evolution from WSN to 6LoWPAN, an adaptation layer is introduced between MAC and Network layers to provide a seamless connection between IPv6 layer and IEEE 802.15.4 MAC layer [4], [5]. In this paper, the 6LoWPAN adaptation layer is designed and developed with fragmentation and reassembly functionalities using Qualnet in order to support the packet transition across the layer protocol stacks. The fragmentation mechanism is performed whenever the received packet at the end of adaptation layer is larger than IEEE 802.15.4 MAC frame size. The packet is fragmented into smaller size with new header appended such that the payload of bottom layer is able to carry the received data. The reassembly mechanism always emerges as a pair with fragmentation mechanism. The adaptation layer will reassemble the fragmented packets whenever the received packets at the other end are identified as 6LoWPAN fragments. The Qualnet simulator, however, does not currently provide the IPv6 protocol stacks. In order to overcome this issue, the IPv6 protocol stacks for developing the 6LoWPAN adaptation layer has been successfully emulated by modifying IPv4 based addressing scheme. The performance of the developed 6LoWPAN has been evaluated and compared to WSN in terms of packet delivery ratio, throughput, average end-to-end delay and total average energy consumption. Both 6LoWPAN and WSN developed in Qualnet simulator are based on the IEEE 802.15.4 standard. It is shown that the developed adaptation layer with its fragmentation and reassembly mechanisms in 6LoWPAN has greatly improved the throughput and packet delivery ratio as expected by 201.34% and 14% respectively compared to WSN. This enhancement is caused by the adapted datagram size that indicates the original packet size for reassembly purpose helps in the process of reassemble packet. Hence, this reduces the chances of packet lost at intermediate node when the synthesized reassembly buffer is full compared to IP fragmentation. As a tradeoff from these results, the total average energy consumption in 6LoWPAN is higher than WSN by 274.4%, since big amount of energy is consumed for sending higher amount of fragmented packet. Also, the average end-toend delay in 6LoWPAN is higher by 33.66% which caused by experiencing two time processes of fragmentation and reassembly compared to only one time process in WSN.
Development of 6LoWPAN Adaptation Layer with Fragmentation
201
The rest of this paper is organized as follows. Section 2 presents the overview of 6LoWPAN. Section 3 presents the important elements in 6LoWPAN adaptation layer. The fragmentation and reassembly mechanisms in the adaptation layer are described in Section 4. The performance evaluations for 6LoWPAN and WSN are discussed in Section 5. Finally, this paper is concluded in Section 6.
2 6LoWPAN Overview The 6LoWPAN is defined based on the low power personal area network (PAN) with the integration of IPv6 over IEEE 802.15.4 standard. It enables IPv6 to be applied to WSN sensor nodes to provide end to end communication. There are some basic specifications distinguished in 6LoWPAN such as it supports large address space of IPv6, both short 16-bit or IEEE 64-bit extended MAC address, and operates as ad-hoc with star and mesh network topologies, besides operated in low power, low bandwidth, low data rates, low cost, and small packet size [6], [7]. Since 6LoWPAN utilized IPv6 as upper network layer for low rate wireless PAN (WPAN) its technology has attracted extensive attention with offering several technical advantages associated to IPv6. Due to the integration of IPv6, the next generation IP, with low rate WPAN, the deployment of 6LoWPAN will be easy and widely accepted in the future which is suitable to WSN. 6LoWPAN is able to support the increased number of sensor devices with the new addressing scheme. This exactly meets the need of disposing large-scale and high density of low rate WPAN. It also supports the stateless address auto-configuration since the MAC address is read automatically once sensor nodes are operated for reducing overhead configuration on the hosts comparing to stateful addressing scheme. Hence, the implementation of low rate WPAN using IPv6 technology will make the network access becomes easier compared to others [5]. 2.1 Network Architecture The sensor nodes in 6LoWPAN are defined into three types: reduced functional device (RFD), full functional device (FFD) and PAN coordinator. RFD is considered as an end device with limited function, which can only communicate to FFD. FFD can communicate with both RFD and FFD by sending its MAC layer beacon frames. PAN coordinator is considered as a coordinator to coordinate the bilateral communication by synchronizing the services through the MAC layer beacons transmission [6]. The communication between the sensor nodes in 6LoWPAN and other end of IP network nodes are shown in Fig. 1. Note that in each 6LoWPAN topology there is only one PAN coordinator. The end device of RFD first sends packet to FFD in order to route the packet to 6LoWPAN gateway which is defined by PAN coordinator. The data of sensor nodes is then forwarded to IP network nodes the 6LoWPAN gateway [5], [8]. Normally, the 6LoWPAN network topology is designed as either star or mesh topology. In the star network topology, the role of PAN coordinator acts as a single central controller to establish communication link between sensor devices. Thus, each device has one hop left from the PAN coordinator. However, in the mesh network topology, the sensor
202
C.W. Chan et al.
devices can communicate with each other when their provided radio range allows them to do so. The multi hops to another hop is possible to route the message from one sensor node to the others [4].
Fig. 1. 6LoWPAN network architecture
2.2 Network Protocol Stack Internet Task Working Force (IETF) has defined 6LoWPAN adaptation layer to provide interoperability of IPv6 to sensor networks [2]. 6LoWPAN employs IEEE 802.15.4 protocol as its bottom layer standard, where layer one and layer two of 6LoWPAN protocol are adopting the IEEE 802.15.4 standard physical (PHY) and MAC layers respectively as shown in Fig. 2 [4].
Fig. 2. The reference model of 6LoWPAN protocol stack
Development of 6LoWPAN Adaptation Layer with Fragmentation
203
The intention of 6LoWPAN adaptation layer is to enable efficient communication between network layer and IEEE 802.15.4 standard layer. The maximum MAC frame size of 127 bytes is unable to accommodate 1280 bytes MTU of IPv6. The other 25 bytes of MAC frame overhead and 21 bytes of link-layer security overhead are imposed, and it leaves only 81 bytes MAC frame payload size. In upper network layer the 40 bytes of overhead is allocated for IPv6 header which reduces the 81 bytes MAC frame payload size to 41 bytes. It will be further reduced to only 33 or 21 bytes when user datagram protocol (UDP) or transmission control protocol (TCP) is taken into consideration. The general packet format of subsequent layers in 6LoWPAN protocol stack is shown in Fig. 3 [8].
Fig. 3. The general packet format of subsequent layers in 6LoWPAN protocol stack
3 6LoWPAN Adaptation Layer In order to implement the seamless communication between IEEE 802.15.4 standard MAC layer and IPv6 network layer, an adaptation layer between them has been proposed by 6LoWPAN working group to perform header compression, fragmentation and layer-two forwarding [2]. In the header compression mechanism, the 6LoWPAN defines HC1 encoding as an optimized compression scheme for the link-local IPv6 communication. Some IPv6 header fields such as IPv6 length fields and IPv6 addresses are eliminated from its packet as long as the adaptation layer can derive them from the headers in the link-layer frame. Furthermore, the header fields that come from adaptation, network and transport layers usually carry the common value. Hence, in order to reduce transmission overhead, header compression mechanism is used to compress those header fields into a few bits while reserving an escape value for the less appeared bits.
204
C.W. Chan et al.
In the fragmentation mechanism, when the IPv6 packets cannot fit into the IEEE 802.15.4 MAC frame payload size (81 bytes), these packets are fragmented into multiple link-layer frames such that the IPv6 minimum MTU can be accommodated and as a requirement for reassembling them at the other end. The first fragmentation header (4 bytes) and the subsequent fragmentation header (5 bytes) which includes an extra byte for offset field in the 6LoWPAN fragmentation mechanism are shown in Fig. 4. The datagram size header field is used to specify the entire IP packet size before the adaptation layer fragmentation. This field value shall be the same for all link-layer fragments of an IP packet. The datagram tag header field is used to identify all fragments of a single original packet. Basically, all fragments of a single packet have the same value in this field. Another header field, the datagram offset field presents only in the second and subsequent fragments, and shall specify the offset of a fragment (increments of 8 octets) from the beginning of payload datagram. Usually the implicit value of the datagram offset field in the first fragment is zero [2].
Fig. 4. The (a) first and (b) subsequent fragmentation headers
In the layer-two forwarding of IPv6 datagram mechanism, the adaptation layer can forward the link-layer level addresses at the end of each IP hop. Alternatively, the IPv6 network layer can accomplish intra-PAN routing via layer-three (adaptation layer) forwarding, where each 802.15.4 radio hop is an IP hop [10]. To accomplish the multi hops packet forwarding, the 6LoWPAN has defined the mesh header (4 or 5 bytes) as shown in Fig. 5. Basically, the mesh header is used to standardize the way to encode the hop limit and the link layer’s source and destination of the packets.
1
1
0
0
O
O
F
F
Hops left (4 bits)
0xF
Originator address (16-64 bits) Hops left (8 bits)
Final address (16-64 bits)
Originator address (16-64 bits)
Final address (16-64 bits)
Fig. 5. Mesh header
The different mechanisms in adaptation layer require different types of header, the first two bits of the header is used to identify this header type. As shown in Figs. 4 and 5, the bit pattern 11 is used for the fragmentation header while bit pattern 10 is
Development of 6LoWPAN Adaptation Layer with Fragmentation
0 0
1 1
205
Dispatch (6 bits)
0x3F
Dispatch (8 bits)
Fig. 6. Dispatch header
used for the mesh header. The bit pattern 00 is reserved to identify the coexistence of non-6LoWPAN frames. The bit pattern 01 is used for the dispatch header (1 or 2 bytes) as shown in Fig. 6.
4 Fragmentation and Reassembly Mechanisms The developed adaptation layer between IPv6 network layer and IEEE 802.15.4 MAC layer will perform the fragmentation and reassembly mechanisms. The basic fragmentation and reassembly mechanisms are shown in Fig. 7. Since the Qualnet simulator does not currently provide the IPv6 protocol stacks, an IPv6 packet is emulated by modifying the IPv4 based addressing scheme. This emulated IPv6 packet can be developed by enlarging the 576 bytes of IPv4 MTU to the minimum 1280 bytes of IPv6 MTU. The emulated IPv6 packet needs to be fragmented before sent to IEEE 802.15.4 MAC layer. The fragmented packets will be reassembled to the original emulated IPv6 packet before sent to network layer at the other network end.
Fig. 7. The overview of fragmentation and reassembly mechanisms
4.1 Fragmentation Mechanism Fragmentation is a process of breaking down IPv6 packet into multiple link-layer fragments. The maximum available MAC frame size is reduced to 81 bytes where 25 bytes is reserved for MAC frame overhead and 21 bytes is reserved for security overhead. Thus, a packet will only be fragmented when the adaptation layer receives an IPv6 packet with the minimum size of 81 bytes. 6LoWPAN fragments are formed
206
C.W. Chan et al. 6LoWPAN Fragmentation IPv6 Header Header Header length
Payload Packet length
Fig. 8. 6LoWPAN packet
by appending IPv6 header and 6LoWPAN fragmentation header on the fragmented payload as shown in Fig. 8. Algorithm of fragmentation is summarized in Fig. 9. Recall that the process of fragmenting packet will only take place when the IPv6 size is at least 81 bytes. Once this condition is fulfilled, the measurement of IPv6 header length of the received IPv6 packet is recorded as unfragmented length. This is because the IPv6 header will be removed first at the beginning of fragmenting process. Then, the measurement of a fragmented packet length will be computed based on the received packet payload size. During the fragmenting process, each fragment will be appended with the previously eliminated IPv6 header, followed by the 6LoWPAN fragmentation header. All of these successful fragments from a same IPv6 packet will be queued to ensure a complete fragmentation process is done for a particular IPv6 packet before sending to IEEE 802.15.4 MAC layer.
Fig. 9. The flow of fragmentation mechanism
Development of 6LoWPAN Adaptation Layer with Fragmentation
207
4.2 Reassembly Mechanism Reassembly is a process to combine fragmented packets into original IPv6 packet upon the reception at the other end of adaptation layer. Before performing reassembly mechanism, each received packet at the end of adaptation layer is examined. The algorithm of reassembly is summarized in Fig. 10.
Fig. 10. The flow of reassembly mechanism
5 Performance Evaluation between 6LoWPAN and WSN In this section, the performance of the developed 6LoWPAN is evaluated and compared with WSN in terms of packet delivery ratio, throughput, average end-to-end delay and total average energy consumption using Qualnet simulator. Table 1 shows the list of simulation parameters for the simulation setup. These simulation parameters are including radio type, MAC and Network protocols, IP fragmentation unit, routing protocol, number of nodes, area size, simulation time, and packet size. The configuration parameter of IP fragmentation unit for 6LoWPAN is given as 1280 bytes instead of 70 bytes which is for WSN as shown in Table 1. The packet size for 6LoWPAN is fixed to 1280 bytes instead of 512 bytes which is for WSN.
208
C.W. Chan et al. Table 1. List of simulation parameters
Parameter Radio Type
Measurements 802.15.4 Radio
MAC Protocol
802.15.4
Network Protocol
IPv4
IP fragmentation unit
1280, 70 bytes
Routing protocol
AODV
Number of nodes
5, 10, 15, 20, 25, 30 nodes
Area size
500 meters x 500 meters
Simulation time
30 minutes
Packet size
1280, 512 bytes
In each simulation, the number of sensor nodes in the simulation scenario is increased by five nodes. Due to the processed computing limitation, the maximum number of sensor nodes that can be simulated is up to 30 which can be shown in Fig. 11. In each scenario only one PAN coordinator is used which is responsible to route and forward data from sending node to receiving node. Two nodes which have large enough distance between each other will use constant bit rate (CBR) application. The node placement in each scenario is done by setting node placement model as random in Qualnet. This is mainly because each sensor node in a network may have unpredicted movement or condition (turn on and off) which is caused by the environment, functionality, and so on. Therefore, the random node placement in Qualnet will eventually cause some fluctuation in the performance graphs of this section.
Fig. 11. The simulation scenario of 30 sensor nodes
Development of 6LoWPAN Adaptation Layer with Fragmentation
209
5.1 Packet Delivery Ratio Packet delivery ratio is calculated as the total packet received with respect to total packet sent. As depicted in Fig. 12, the packet delivery ratio for 6LoWPAN is 14% higher than WSN. This is because the datagram size in 6LoWPAN fragment header is used to indicate the original size of a packet before fragmentation. The receiving end will create a buffer for reassembling process upon the reception of first fragment. Also, the arrival of fragments does not necessary in order. Since WSN employs more fragment flag header field without indicating the original size of a packet as what the 6LoWPAN fragment header done, the created buffer might not be sufficient to accommodate all fragments. Therefore, 6LoWPAN experiences less packet lost in the intermediate node compared to WSN. 5.2 Throughput The throughput is calculated as ratio of the total bits received with respect to the difference of time between the last received packets and the first received packet. Fig. 13 illustrates that 6LoWPAN has 201.34% higher throughput than WSN. Since WSN has lower packet delivery ratio and experienced more delay which is induced by the packet lost, therefore WSN may experience a lower throughput compared to 6LoWPAN. 5.3 Average End-to-End Delay The average end-to-end delay is calculated as the sum of delays from each packet received with respect to the number of received packet. Fig. 14 describes that the average end-to-end delay of 6LoWPAN is 33.66% higher than WSN. This is because 6LoWPAN uses route-over mechanism to route packet. The route-over mechanism requires each fragment that received at intermediate node to be reassembled first before the packet is re-fragmented and forwarding out to the next hop toward the destination node. Thus, each sensor node in 6LoWPAN is suffered for two times of delay which is induced from the network and adaptation layers for the repeating fragmentation and reassembling mechanism in the intermediate node. The number of required hops is increased with the number of nodes, since the data is transmitted between two nodes with a large distance. 5.4 Total Average Energy Consumption The maximum number of bits sent by a node is defined by total battery energy divided by the required energy per bit. The used radio energy is Mica Motes. This model reads the power consumption in three different modes: transmit, receive and idle modes. The energy consumption for each sensor node is the sum of all energy used in these three modes. Thus, the total average energy consumption is calculated by summing energy consumption for each node with respect to total number of nodes. The total number of bits sent by 6LoWPAN is higher than WSN due to introduction of 6LoWPAN fragmentation header and 40 bytes of IPv6 header appended in the packet. From the results shown in Fig. 15, the total average energy consumption in 6LoWPAN is higher than WSN by 274.4%. This is due to the huge amount of energy
210
C.W. Chan et al.
is consumed for sending higher amount of fragmented packet. Nevertheless, the consumed energy is almost same independent of number of nodes. This is due to all present nodes in the network may not be participated in the communication.
Fig. 12. Performances of packet delivery ratio for 6LoWPAN and WSN
Fig. 13. Performances of Throughput for 6LoWPAN and WSN
Development of 6LoWPAN Adaptation Layer with Fragmentation
Fig. 14. Performances of average end-to-end delay for 6LoWPAN and WSN
Fig. 15. Performances of total average energy consumption for 6LoWPAN and WSN
211
212
C.W. Chan et al.
6 Conclusions The 6LoWPAN adaptation layer with fragmentation and reassembly mechanisms has been developed and simulated by using Qualnet simulator. Since the IPv6 protocol stacks are not currently supported and provided by Qualnet libraries, an emulated IPv6 packet with a minimum 1280 bytes MTU has been successfully implemented and validated by enlarging the 576 bytes of IPv4 MTU in the Qualnet simulator. The performance of 6LoWPAN with this developed adaptation layer is evaluated and compared to WSN in terms of packet delivery ratio, throughput, average end-to-end delay and total average energy consumption. The obtained results show that 6LoWPAN has higher packet delivery ratio and throughput by 14% and 201.34% respectively compared to WSN. However, as a tradeoff, 6LoWPAN experiences an extra 33.66% average end-to-end delay and 274.4% higher total average energy consumption than WSN. These are the drawbacks of the introduction of fragmentation and reassembly mechanisms into 6LoWPAN adaptation layer.
References 1. Damaso, A.V.L., Domingues, J.P.O., Rosa, N.S.: SAGe: Sensor Advanced Gateway for Integrating Wireless Sensor Networks and Internet. In: IEEE 24th International Conference on Advanced Information Networking and Applications Workshops (WAINA), pp. 698–703 (2010) 2. Zimmermann, A., Silva, J.S., Sobral, J.B.M., Boavida, F.: 6GLAD: IPv6 Global to Link Layer Address Translation for 6lowpan Overhead Reducing. Next Generation Internet Networks, 209–214 (2008) 3. Zeeb, E., Behnke, R., Hess, C., Timmermann, D., Golatowski, F., Thurow, K.: Generic Sensor Network Gateway Architecture for Plug And Play Data Management in Smart Laboratory Environments. In: IEEE Conference on Emerging Technologies & Factory Automation (ETFA), pp. 1–8 (2009) 4. IEEE Standard 802.15.4-2003: Wireless Medium Access Control and Physical Layer Specifications for Low-Rate Wireless Personal Area Networks (2003) 5. Ma, X., Luo, W.: The Analysis of 6LoWPAN Technology. In: IEEE Pacific-Asia Workshop on Computational Intelligence and Industrial Application (2008) 6. Shin, M.K., Kim, H.J.: L3 Mobility Support in Large-Scale IP-Based Sensor Networks (6LoWPAN). In: 11th International Conference on Advanced Communication Technology, pp. 941–945 (2009) 7. Kushalnagar, N., Montenegro, G., Schumacher, C.: IPv6 over Low-Power Wireless Personal Area Networks (6LoWPANs): Overview, Assumptions, Problem Statement, and Goals. RFC4919 (2007) 8. Ee, G.K., Ng, C.K., Noordin, N.K., Ali, B.M.: A Review of 6LoWPAN Routing Protocols. In: Proceedings of the Asia Pacific Advanced Network, APAN (2010) 9. Montenegro, G., Kushalnagar, N., Hui, J., Culler, D.: Transmission of IPv6 Packets over IEEE 802.15.4 Networks. RFC4944 (2007) 10. Ee, G.K., Ng, C.K., Noordin, N.K., Ali, B.M.: Path Recovery Mechanism in 6LoWPAN Routing. In: International Conference on Computer and Communication Engineering, ICCCE (2010)
Performance Evaluation of Switching between Routing Protocols in Mobile Ad Hoc Networks Annapurna P. Patil, K. Rajani kanth, Desai Neel, N. Ganesh, and P. Amala Department of Computer Science & Engineering, M S Ramaiah Institute of Technology, Bangalore-54, Karnataka, India {annapurnap2,rajanikanth}@msrit.edu
Abstract. Ad-hoc networks employ routing protocols so that out-of-range nodes can communicate with each other via intermediate nodes. The innate dynamicity and complexity of mobile ad-hoc networks has resulted in numerous routing protocols for ad-hoc networks being proposed. Furthermore, numerous variants and hybrids of these existing protocols continue to be reported in literature. This diversity appears to be inherent to the field and it seems unlikely that there will ever be a ‘one-size-fits-all’ solution to the ad-hoc routing problem. However, typical deployment environments for routing protocol for ad-hoc networks still force the choice of a single fixed protocol and the resultant compromise can easily lead to sub-optimal performance, depending on current operating conditions. In our work we try to address this problem by switching between protocols that best suits the current working environment. Details about the current network environment are obtained by sensing the network. The protocols chosen for conducting the switching operation are AODV (reactive) and DSDV (proactive). We measure the performance of such an implementation and compare the results with their monolithic counterparts to see if such a switching is feasible. Keywords: MANETs, Routing Protocols, Switching.
1 Introduction A Mobile Ad Hoc Network (MANET) [1], sometimes called a mobile mesh network, is a self-configuring network of mobile devices connected by wireless links. Some of the properties MANETs should possess are rapid deployment, self configuration, existing infrastructure independence, etc. MANETs find many applications due to their quick and economically less demanding deployment. They have increasingly been applied to civilian applications. Since the advent of wireless communication the requirements for providing better service in MANET is growing exponentially. In areas where there is very little or no communication infrastructure or the existing infrastructure is expensive or inconvenient to use, wireless mobile users will still be able to communicate through the formation of a MANET. In the next generation of wireless communication systems, there will be a need for the rapid deployment of independent mobile users. Significant examples include establishing survivable, A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 213–222, 2011. © Springer-Verlag Berlin Heidelberg 2011
214
A.P. Patil et al.
efficient, dynamic communication for emergency/rescue operations, disaster relief efforts, and military networks [1]. A MANET is an autonomous collection of mobile users that communicate over relatively bandwidth constrained wireless links. Since the nodes are mobile, the network topology may change rapidly and unpredictably over time. The network is decentralized, where all network activity including discovering the topology and delivering messages must be executed by the nodes themselves, i.e.., routing functionality will be incorporated into mobile nodes. Their main asset resides in the fact that they are not tributary to fixed installations [2]. MANETs use routing protocols to ensure out of range nodes can commune with each other via the intermediate nodes [3]. Unfortunately, it is hard to design generically-applicable routing protocols in the MANET environment. This is basically for two main reasons: First, MANETs are inherently characterized by dynamic variations in network conditions—for example in terms of network size, topology, density or mobility. Second, MANETs are subject to a diverse and dynamic set of application requirements in terms of quality of service (QoS) demands and traffic Patterns (i.e. in terms of messaging, request-reply, multicast, publish-subscribe, streaming, etc.). In response to these two pressures, MANET researchers have been proposing an ever-proliferating range of routing protocols, e.g. AODV [4], DYMO [5], OLSR [6], ZRP [7], TORA [8] and GPSR [9] to name a few. However none of these proposals come close to providing optimal routing under the full range of operating conditions encountered in MANET environments; and it is becoming clearer that the ‘one-size-fit-all’ ad-hoc routing protocol is an impossibility. We therefore believe according to the survey done that future MANET systems will need to employ multiple routing protocols for ad-hoc networks and to support switching between these as runtime conditions dictate. The paper is organized as follows: Section 2 gives details about the Literature survey. Section 3 provides a brief description of the related works. In section 4 we discuss about the proposed work and section 5 gives the conclusions about our work.
2 Literature Survey 2.1 Routing Protocols for Ad-Hoc Networks The design space of routing protocols for ad-hoc networks can be divided into three broad categories [10]. Proactive Routing Protocols: Proactive protocols continuously learn the topology of the network by exchanging topological information among the nodes in the network. Thus, when there is a need for a route to a destination, such route information is available immediately. If the network topology changes too frequently, the cost of maintaining the network information might be very high. If the network activity is low, the information about actual topology might even not be used. Ex DSDV Reactive Routing Protocols: The reactive routing protocols are based on some sort of query-reply dialog. Reactive protocols proceed for establishing route(s) to the
Performance Evaluation of Switching between Routing Protocols in MANETs
215
destination only when the need arises. They do not need periodic transmission of topological information of the network. Ex: DSR, AODV [4] Hybrid Routing Protocols: Often reactive or proactive feature of a particular routing protocol might not be satisfactory; instead a mixture might yield better solution. Hence, in the recent studies, several hybrid protocols are also proposed. Ex: ZRP [7], MPOLSR [11]. The biggest determining factor for the protocol choice is the size of the network: generally, proactive protocols are better suited to smaller networks, reactive protocols better to larger networks, and hybrid protocols to networks that can be structured hierarchically. But where the network varies in size (e.g. grows), an initial choice of protocol (e.g. proactive) can become sub-optimal. As another example, a reactive protocol will do well where pairs of interacting source-destination nodes tend to be stable, while proactive protocols are typically better where interaction patterns are more dynamic (although only where the network is not too big). In addition, peer-to-peer services running over MANETs tend to prefer proactive protocols [12] and applications requiring QoS differentiation can benefit from intelligent path selection as enabled by multipath routing algorithms like TORA [8] or Multipath DYMO [13], although these carry overhead that is unnecessary for other applications. Apart from proposing many new protocols in each of the above categories, researchers have investigated numerous variations on already-existing protocols. For example, path accumulation [5], pre-emptive routing [14], multi-path routing [13], power-efficient routing [15], fish-eye routing [16], and numerous styles of flooding [6, 17, 18, 19] are examples of techniques that can be ‘switched on’ to improve a particular property of an underlying base protocol under certain operating conditions, but which may be counter-productive under other conditions. 2.2 Protocols Selected for Evaluation • Ad-Hoc on-Demand Distance Vector(AODV) This is a reactive protocol, even though it uses some features of a proactive protocol. AODV takes the interesting parts of DSR and DSDV, in the sense that it uses the concept of route discovery and route maintenance of DSR and the concept of sequence numbers and sending of periodic hello messages from DSDV. Routes in AODV are established and maintained only when and as long as needed.[23]. AODV takes advantage of route tables. In these it stores routing information as destination and next hop addresses as well as the sequence number of a destination. A node also keeps a list of the precursor nodes, which route through it, to makes route maintenance easier after link breakage. To prevent storing information and maintenance of routes that are not used anymore each route table entry has a lifetime. If during this time the route has not been used, the entry is discarded. AODV does not need any central administrative system to control the routing process. Reactive protocols like AODV tend to reduce the control traffic messages overhead at the cost of increased latency in finding new routes.
216
A.P. Patil et al.
• Dynamic Destination-Sequenced Distance- Vector Routing Protocol (DSDV) DSDV [1], [12] is developed on the basis of Bellman–Ford routing [12] algorithm with some modifications. In this routing protocol, each mobile node in the network keeps a routing table. The routing table contains the list of all available destinations and the number of hops to each destination. Each table entry is tagged with a sequence number, which is originated by the destination node. Periodic transmissions of updates of the routing tables help maintaining the topology information of the network. If there is any new significant change for the routing information, the updates are transmitted immediately. So, the routing information updates might either be periodic or event driven. DSDV protocol requires each mobile node in the network to advertise its own routing table to its current neighbors. The advertisement is done either by broadcasting or by multicasting. By the advertisements, the neighboring nodes can know about any change that has occurred in the network due to the movements of nodes. The routing updates could be sent in two ways namely the first one is called a ‘‘full dump’’ and the second is ‘‘incremental.’’ In case of full dump, the entire routing table is sent to the neighbors, where as in case of incremental update, only the entries that require changes are sent. • AODV v/s DSDV [24]: Convergence time, packet delivery ratio and throughput were taken as performance metrics. In order to cover most if not all the types of scenarios the algorithms might face, both the node density (number of nodes) and the node mobility (pause time) were varied. The node density (number of nodes) was varied in the range [10,100] in steps of 10 (10 different node densities).The upper limit of this range was chosen to be 180 because the simulation time is 180s in all the cases. Thus a pause time of 180 implies that the nodes pause in their initial positions for 180 seconds – the entire duration of the simulation. Hence this represents the case where nodes are completely static. Similarly, pause time 0 represents very high mobility where the nodes are in constant motion. Thus each algorithm was tested over 10 node densities x 10 pause times = 100 scenarios. The following results were observed [23]. It was found AODV converges faster for low node density and high mobility network environments while DSDV converges faster for high node density and low mobility network environments. The key conclusion is that no single protocol or class of protocols is well suited to more than a subset of the operating conditions to be found in any given MANET environment at any given time .
3 Related Work We are not alone in recognizing the importance of dynamic switching between protocols in MANETs. Many frame works have been proposed in literature to support this cause.
Performance Evaluation of Switching between Routing Protocols in MANETs
217
Protocol Frameworks: MANET researchers have recently developed a number of frameworks; prominent among them are ASL [20], PICA [21] and MANETKit [3]. ASL, for example, enhances underlying system services and provides MANET-specific APIs such that routing protocols can be developed in user-space. PICA alternatively provides multiplatform functionality for threading, packet queue management, socket-event notifications to waiting threads, and network device listing, as well as minimizing platform-related differences in socket APIs, and kernel route table manipulation. MANETKit provides the developer with an extensible set of common ad-hoc routing protocol functionality (encapsulated in components), and tools to configure and reconfigure protocol graphs implemented as nested CFs. In addition, the popular Unik-olsrd [22] implementation of OLSR supports a plug-in framework which has been well used by researchers [16, 17]. Intelligent MANET Routing System [25]: Mobile Ad Hoc Network consists of many routing protocols, which are capable of routing data packets from source to destination and they all aim to find the most suitable path. Research has been going on in Artificial Neural Network (ANN) to enhance the quality of the selected path. Example: The use of a modified Hopfield ANN to solve the shortest path problem, to guarantee fast computation appropriate to multi-hop radio networks and also to prevent the packet loss. A hybrid system has also been designed, combing Hopfield networks and genetic algorithm to solve the problem of optimal routing in computer networks. MANETKit [3]: MANETKit is an OpenCom ComponentFramework (CF) that supports the development, deployment and dynamic reconfiguration of routing protocols for ad-hoc networks. It provides the developer with an extensible set of common ad-hoc routing protocol functionality (encapsulated in components), and tools to configure and reconfigure protocol graphs implemented as nested CFs.
4 Proposed System Design and Implementation The overall system function flow diagram is shown in Figure 1. From Figure 1 we see that the DSDV protocol is initially run on the given network scenario to obtain the number of nodes. This is then passed as input to the switching code which has the protocol switching table embedded in it. The code selects the best suiting protocol for the network environment and runs the simulation using NS2 [26]. The NAM file is for visual simulation and the required parameters are extracted from the trace file. The parameters thus got (convergence time) are compared with the monolithic implementation of the protocols. This process repeats for the specified number of trials. In our work we simulate the required network environment using NS2 as the network simulation tool. The cost Metrics being Node Density, and Pause Time. Values for the above metrics are randomly generated which defines the current network environment for simulation. The environment is then sensed and accordingly one of AODV or DSDV is chosen as the routing protocol. The selection is driven by the information in table 1 which is pre-fed to the switching code.
218
A.P. Patil et al.
Fig. 1. System function flow diagram Table 1. Input to the Switching Tool Nodes = 80
Pause Time = 100
AODV/DSDV
AODV/DSDV
DSDV
Performance Evaluation of Switching between Routing Protocols in MANETs
219
5 Experimental Results We have tested the tool for convergence time, throughput and packet delivery ratio under four conditions: Low node density and low mobility, Low node density and high mobility, High node density and low mobility, High node density and high mobility. Network scenario: The simulations are conducted using the network simulator ns2 [26]. Random Waypoint mobility model is used. The physical layer simulates the behavior of IEEE 802.11 (as included with ns2). Each node has a radio range of 250 meter, and uses TwoRayGround as the radio propagation model. All the scenarios are based on the following basic parameters: cbr (constant bit rate) traffic, topology of size 500 m x 500 m, maximum speed of each node 20 m/s, simulation time 180s,transmission rate (packet rate) 10 m/s. Simulations were carried out for number of nodes varying from 30 to 120 with interval of 30 and also with pause time varying from 30 to 120 with intervals of 30.For the varying pause time and number of nodes the convergence time was recorded for simulation running AODV, DSDV and Switching tool. The switching tool out performed in all the four tested conditions. Even in conditions where AODV or DSDV fails the switching tool has not failed. The results are shown in the graphical form as in Figure: 2-5.
Fig. 2. Comparisons for 30 nodes
Fig. 3. Comparisons for 60 nodes
Fig. 4. Comparisons for 90 nodes
Fig. 5. Comparisons for 120 nodes
220
A.P. Patil et al. Table 2. Results for Throughput Number of Nodes
30
30 87.9528
60 76.011
90 68.7194
120 43.1611
60 90 120 30 60 90 120 30 60 90 120
85.166 74.708 80.4389 87.0194 91.7611 92.388 83.5638 88.9735 95.7878 80.3539 78.990
80.244 84.3999 78.8416 94.5638 98.1667 99.425 82.8333 77.2555 72.9527 74.1916 82.975
64.5305 53.5972 58.8805 99.3889 87.7388 86.9139 85.1833 61.575 61.4583 83.9873 88.9753
46.9944 46.1805 38.8361 82.4165 81.4660 79.2276 74.9674 45.7888 45.3331 79.8881 72.9876
AODV
Pause Time (Sec)
DSDV
Switching Tool
Table 3. Results for Packet delivery ratio
AODV
Pause Time (Sec)
DSDV
Switching Tool
30 60 90 120 30 60 90 120 30 60 90 120
30 65.1520 63.106 55.3546 59.5413 64.6228 67.9746 68.5329 61.8811 64.3121 68.8943 56.3333 59.3212
Number of Nodes 60 90 30.9785 20.8435 32.7249 19.5679 34.3827 16.2826 32.0699 17.8537 38.5052 30.1867 40.719 26.6402 40.477 26.3655 33.7181 25.8214 31.4273 18.6775 29.6978 18.6295 30.2247 26.5201 33.7462 26.7789
120 9.0502 14.8552 9.6838 7.8809 16.2879 16.6475 16.7143 16.0295 8.8509 11.7731 18.6723 16.0137
6 Conclusions In this work a switching tool to switch between the routing protocols in MANETs is designed and its performance evaluated. AODV and DSDV are chosen as the base algorithms. This switching obtains the number of nodes in the environment and accordingly switches between AODV and DSDV protocols. The net task done was to compare the performance of the switching tool with individual implementations of AODV and DSDV protocols via realistic scenario simulations using Network Simulator 2 (NS2) v2.34, under varying mobility of the nodes and node densities. These protocols are evaluated using the network performance metrics convergence time, throughput and packet delivery ratio.
Performance Evaluation of Switching between Routing Protocols in MANETs
221
Extensive simulations of the switching tool and the algorithms AODV and DSDV were conducted over a wide range of network scenarios. When the results of simulation were analyzed it is found that switching tool out performs AODV and DSDV in most of the cases considered for the convergence metric. Switching tool with throughput and packet delivery ratio parameters proved to be better than AODV and DSDV only at certain points as documented in table 2 and 3.
References 1. Siva Ram Murthy, C., Manoj, B.S.: Ad Hoc Wireless Networks Architectures and Protocols, 2nd edn. Prentice hall of India (2007) 2. Quintero, A., Pierre, S., Macabeo, B.: A routing protocol based on node density for ad hoc networks. Ad-Hoc Networks 2, 335–349 (2004) 3. Ramdhany, R., Grace, P., Coulson, G., Hutchison, D.: MANETKit: Supporting the Dynamic Deployment and Reconfiguration of Ad-Hoc Routing Protocols. In: Bacon, J.M., Cooper, B.F. (eds.) Middleware 2009. LNCS, vol. 5896, pp. 1–20. Springer, Heidelberg (2009) 4. Perkins, C., Royer, E.: Ad-hoc On demand distance vector routing. Internet Draft RFC 3561 (July 2003) 5. Chakeres, I., Perkins, C.: Dynamic MANET on-demand (DYMO) routing., draft-ietfmanetdymo-11, IETF’s MANET WG (November 2007) 6. Clausen, T., Dearlove, C.: Optimized link state routing protocol, v2, draft-ietf-manetolsrv2-03.txt (2007) 7. Haas, Z.J., Pearlman, M.R., Samar, P.: The zone routing protocol (ZRP) for ad-hoc networks, Internet Draft, draft-ietf-manet-zone-zrp-04.txt (July 2002) 8. Park, V.D., Corson, M.S.: A highly adaptive distributed routing algorithm for mobile wireless network. In: Proceedings of 16th IEEE Conference on Computer and Communications Societies (INFOCOM 1997), Kobe, Japan, vol. 3, pp. 1405–1413 (April 1997) 9. Karp, B., Kung, H.T.: Greedy perimeter stateless routing for wireless networks. In: Proc. 6th Annual ACM/IEEE International Conference on Mobile Computing and Networking, MobiCom (2000) 10. Vijaya Kumar, G., Vasudeva Reddyr, Y., Nagendra, D.M.: Current Research Work on Routing Protocols for MANET: A Literature Survey. (IJCSE) International Journal on Computer Science and Engineering 2(3), 706–713 (2010) 11. Yi, J., Cizeron, E., Hamma, S., Parrein, B., Lesage, P.: Université de Nantes, Nantes Atlantique Universités, Implementation of Multipath and Multiple Description Coding in OLSR 12. Borgia, E., Conti, M., Delmastro, F., Gregori, E.: Experimental Comparison of Routing and Middleware solutions for Mobile Ad Hoc Networks: Legacy vs Cross-Layer approach. In: Proc. of E-WIND: Workshop on Experimental Approaches to Wireless Network Design and Analysis, in conjuction with ACM SIGCOMM 2005, Philadelphia, pp. 82–87 (August 2005) 13. Galvez, J.J., Ruiz, P.M.: Design and performance evaluation of multipath extensions for the DYMO protocol. In: 32nd IEEE Conference on Local Computer Networks, October 15 (2007)
222
A.P. Patil et al.
14. Goff, T., Abu-Ghazaleh, N.B., Phatak, D.S., Kahvecioglu, R.: Preemptive Maintenance Routing in Ad Hoc Networks. Journal of Parallel and Distributed Computing, Special Issue on Wireless Mobile Communication and Computing 63(2), 123–140 (2003) 15. Mahfoudh, S., Minet, P.: An energy efficient routing based on OLSR in wireless ad hoc and sensor networks. In: Proc. 22nd International Conference on Advanced Information Networking and Applications – Workshops (2008) 16. Gerla, M., Hong, X., Pei, G.: Fisheye State Routing Protocol (FSR) for Ad Hoc Networks. In: IETF MANET Working Group Internet Draft, Rockwell Scientific Company, June 17 (2002) 17. Santiváñez, C.A., Ramanathan, R., Stavrakakis, I.: Making link-state routing scale for adhoc networks. In: Proc. 2nd ACM International Symposium on Mobile Ad-Hoc Networking (October 2001) 18. Bani-Yassein, M., Ould-Khaoua, M.: Applications of probabilistic flooding in MANETs. International Journal of Ubiquitous Computing and Communication (January 2007) 19. Haas, Z.J., Halpern, J.Y., Li, L.: Gossip-based ad-hoc routing, INFOCOM 2002. IEEE/ACM TRANSACTIONS ON NETWORKING 14(3) (June 2006) 20. Kawadia, V., Zhang, Y., Gupta, B.: System Services for Ad-Hoc Routing: Architecture, Implementation and Experiences. In: MobiSys (2003) 21. Calafate, C.M.T., Manzoni, P.: A multi-platform programming interface for protocol development. In: 11th Euromicro Conference on Parallel, Distributed and Network-Based Processing (2003) 22. Implementation of the OLSR routing protocol, Unik-olsrd, http://www.olsr.org/ 23. Patil, A.P., Sambaturu, N., Chunhaviriyakul, K.: Convergence time evaluation of algorithms in MANETs. International Journal of Computer Science and Information Security 5(1) (2009) 24. Patil, A.P., Harish, R.: Comparing the Performance of reactive and Hybrid protocols in MANETs using Convergence Time. Preliminary WOCN 2010 Conference Program, Colombo, Sri Lanka (2010) 25. Saeed, N.H., Abbod, M.F., Al-Raweshidy, H.S.: Intelligent MANET Routing System. In: 22nd International Conference on Advanced Information Networking and Applications – Workshops 26. The ISI Network Simulator NS2, http://www.isi.edu/nsnam/ns
Processing of Multidimensional Range Query Using SIMD Instructions Peter Chovanec and Michal Kr´ atk´ y ˇ Department of Computer Science, VSB–Technical University of Ostrava 17. listopadu 15, Ostrava, 708 33, Czech Republic {peter.chovanec,michal.kratky}@vsb.cz
Abstract. Current main stream CPUs provide SIMD (Single Instruction Multiple Data) computational capabilities. Although producers of current hardware provide other computational capabilities like multicores CPU, GPU or APU, an important feature of SIMD is that it provides parallel operations for one CPU’s core. In previous works, authors introduced an utilization of the SIMD instructions in some indexing data structures like B-tree. Since multidimensional data structures manage n-dimensional tuples or rectangles, the utilization of these instructions seems to be straightforward in operations manipulating these n-dimensional objects. In this article, we show the utilization of SIMD in the R-tree data structure. Since the range query is one of the most important operation of multidimensional data structures, we suppose the utilization of SIMD in range query processing. Moreover, we show properties and scalability of this solution. We show that the SIMD range query algorithm is up-to 2× faster then the conventional algorithm.
1
Introduction
Spatial data are often utilized in various application areas including medicine, cartography, computer vision, molecular biology, and many others. Query processing in high-dimensional spaces has therefore been a very prominent research topic over the last few years. Many data structures indexing vector spaces have been developed in past years [23], for example: n-dimensional B-tree [9], Rtree [12], R*-tree [3], Signature R-tree [17], X-tree [29] or UB-tree [2]. One of the most important queries supported by these data structures is the range query [32]. This query can be defined by various spatial objects, in this article we consider the query defined by an n-dimensional rectangle. The R-tree is one of the most popular multidimensional data structure which is utilized in commercial database systems. This data structure bounds spatially near points by multidimensional rectangles. It supports various types of queries, e.g. point and range queries. In the case of the R-tree, the range query returns tuples in a query multidimensional rectangle (QR). The range query can be written as the following pseudo SQL statement:
Work is partially supported by Grant of GACR No. GAP202/10/0573.
A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 223–237, 2011. c Springer-Verlag Berlin Heidelberg 2011
224
P. Chovanec and M. Kr´ atk´ y
SELECT * FROM T WHERE ql1 ≤ T.attr1 ≤ qh1 AND . . . AND qln ≤ T.attrn ≤ qhn
SIMD (Single Instruction, Multiple Data) instructions can increase the performance when exactly the same operations are performed on multiple data objects. Typical applications are communication, digital signal, and graphics processing [10,26,27,14]. Various systems based on the SIMD paradigm have been introduced in last decades. Systems like MMX, SSE1 , 3DNow!2 , AltiVec3 may be considered as the major ones. A complete list of SIMD architectures and their comparison is presented in [28]. 3DNow! is an extension of the x86 instruction set developed by AMD in 1998. Streaming SIMD Extensions (SSE) is a SIMD extension of the x86 instruction set, designed by Intel and introduced in 1999 in their Pentium III series of CPUs as a reply to AMD’s 3DNow!. 3DNow! and SSE include scalar and packed floating point and integer instructions. Whereas 3DNow! includes 64bit-length vector registers, SSE includes 128bit-length vector registers. AltiVec was developed between 1996 and 1998 by a collaborative project of Apple Computers, IBM, and Motorola. Since SSE provides 128bit-length registers enabling us to perform four parallel operations over 32bit integers and floats, we suppose this SIMD extension in our article. We must note that AltiVec provides the same functionality. Although producers of current hardware provide other computational capabilities like multi-cores CPU, GPU or APU4 , the important feature of SIMD is that it provides parallel operations for one CPU’s core. There have been much recent works on CPU behavior (especially CPU’s caches) in database systems [1,5,11,19,21,22]. In [6], authors introduced an efficient implementation of sorting on multi-core SIMD CPU architecture. Full table scan using SIMD technology has been shown in [31]. Authors have presented the full table scan over multiple compressed columns utilizing embedded Vector Processing Units (VPUs). An architecture sensitive layout of the index tree called FAST has been presented in [15]. In [33], authors introduced an utilization of the SIMD instructions in some database operations. Since multidimensional data structures manage n-dimensional tuples or rectangles, the utilization of these instructions seems to be straightforward in operations manipulating these n-dimensional objects. In this article, we show that it is not quite true. In this article, we depict the utilization and properties of the SIMD range query algorithm in a multidimensional data structure. Since the R-tree is the most popular multidimensional data structure and the range query is one of the most important operations of multidimensional data structures, we show the utilization of SIMD in range query processing of the R-tree. The outline of the paper is as follows: In Section 2, we briefly review the Rtree and its variants. In Section 3, up-to-date SIMD extensions are described. 1
2 3 4
http://software.intel.com/en-us/articles/using-intelstreaming-simd-extensions-and-intel-integratedperformance-primitives-to-accelerate-algorithms/ http://www.amd.com/us/products/technologies/3dnow/ Pages/3dnow.aspx http://www.freescale.com/webapp/sps/site/ overview.jsp?code=DRPPCALTVC http://fusion.amd.com/
Processing of Multidimensional Range Query Using SIMD Instructions
225
Section 4 presents the utilization of SIMD for range query processing in the Rtree. Moreover, we show properties and scalability of this solution. In Section 5, we put forward experimental results. Finally, we outline possible areas of our future work and conclude the paper.
2
R-tree and Its Variants
Since 1984 when Guttman proposed his method, R-trees [12] have become the most cited and most used as reference data structure in this area. The R-tree is a height-balanced tree based on the B+ -tree with at least 50% utilization guaranteed. This data structure supports point and range queries and some forms of spatial joins as well. A general structure of the R-tree is shown in Figure 1. R2
p9
p5
R5
p8 p2
R3 R4
R3 R4
p6 p10
R1
R1 R2
p1
p4
R6
p7
p2 p4 p8
p10 p6
R5 R6
p9 p1 p5
p7 p3 p11
p3 p11
Fig. 1. Planar representation and general structure of the R-tree
It is a hierarchical data structure representing spatial data by the set of nested n-dimensional minimum bounding rectangles (MBR). Each MBR is defined by two tuples QL and QH, where QLi ≤ QH i , 1 ≤ i ≤ n. If N is an inner node, it contains pairs (Ri , Pi ), where Pi is a pointer to a child of the node N . If R is the inner node MBR, then the rectangles Ri corresponding to the children Ni of N are contained in R. Rectangles at the same tree level may overlap. If N is a leaf node, it contains pairs (Ri , Oi ), so called index records, where Ri contains a spatial object Oi . Each node of the R-tree contains between m and M entries unless it is the root and corresponds to a disk page. Other properties of the R-tree include the following: – Whenever the number of node’s children drops below m, the node is deleted and its descendants are distributed among the sibling nodes. The upper bound M depends on the size of the disk page. – The root node has at least two entries, unless it is a leaf. – The R-tree is height-balanced; that is, all leaf nodes are at the same level. The height of an R-tree is at most logm N −1 for N index records (N > 1). Many variants of the R-tree have been proposed during the last decades. Although the original R-tree algorithms tried only to minimize the area covered by
226
P. Chovanec and M. Kr´ atk´ y
MBRs, R∗ -tree [3] takes other objectives into account, e.g. the overlap among MBRs. R+ -tree [25] was introduced as a variant that avoids overlapping MBRs in intermediate nodes of the tree and an object can be stored in more than one leaf node. In general, we distinguish R-trees indexing integer or float tuples and point data or rectangle (spatial) objects. When point data are considered, the R-tree includes tuples in leaf nodes. In the case of rectangle objects, leaf nodes include MBRs of spatial objects together with pointers to a representation of spatial objects. In this article, we consider all these variants.
3
Streaming SIMD Extensions
SIMD processes multiple data in parallel with a single instruction, resulting in significant performance improvement; 4 computations at once (see Figure 2). SSE introduced by Intel in 1999 is based on the SIMD paradigm and comprise a set of extensions to the Intel x86 architecture. SSE was subsequently expanded by Intel to the current version SSE5. SSE2 adds new math instructions for doubleprecision (64-bit) floating point and also extends MMX integer instructions to operate on 128-bit XMM registers. SSE2 enables us to perform SIMD operations on any data type (from 8-bit integer to 64-bit float). Other versions are incremental upgrades of older versions. In this article, the label SSE represents all versions of SSE. Four 32 bit values packed in a 128 bit register 127
xmm0
A3
xmm1
B3
95
A2 B2
63
*
A1 B1
31
A0
0
B0
xmm2 = _mm_mul_ps ( xmm0, xmm1)
= xmm2
A3 * B3
A2 * B2
A1 * B1
A0 * B0
Fig. 2. Parallelism based on a SIMD paradigm
SSE originally defines 8 new 128-bit registers (xmm0 – xmm7) for singleprecision floating-point computations. It is possible to work with these data types: – – – –
m64 represents a 64-bit MMX register (8, 16 or 32 bit packed integers) m128 represents 4 packed single precision floating-point values m128d represents 2 packed double precision floating-point values m128i represents packed integer values (8, 16, 32 or 64-bit)
SSE contains over 200 instructions and defines two types of operations: scalar and packed. Scalar operation only operates on the least-significant data element
Processing of Multidimensional Range Query Using SIMD Instructions
227
(bit 0–31) and packed operation computes all four elements in parallel. SSE includes many types of instructions sets, e.g. arithmetic, comparisons and so on. 3.1
Packed Compare Instructions
In the next section, we will show that all operations of range query processing utilize compare operations, therefore we describe the packed compare instructions in this section. Each packed comparison instruction performs a comparison of two registers. The size of each register is fixed set to 128 bits, therefore one register can contain two 64-bit pieces, four 32-bit pieces, and eight 16-bit pieces, respectively. For the packed form, these all values of registers are compared by one instruction, and a 128-bit mask is returned. The mask is set to 0xffffffff for each element where the comparison is true and 0x0 where the comparison is false. A return value is created by the instruction movemask that creates a 4-bit mask from the most significant bits of the four values in register (see Figure 3). In our experiments, we did not use the mm test all zeros instruction due to its high time consumption. xmm2 = _mm_cmplt_ps ( xmm0, xmm1) 127
95
63
31
2.0
0
1.0
xmm0
4.0
<
<
<
<
xmm1
3.0
3.0
3.0
3.0
xmm2
0x0
0x0
0xFFFFFFFF 0xFFFFFFFF
result = _mm_movemask_ps( xmm2 )
16
result
3.0
0
0x0 2, the SIMD algorithms are faster. Therefore, it is necessary to know if the number of cycles for the conventional algorithms > 2; we show these results in the following section. Although it seems that these SIMD algorithms are useful only when n%3 = 0 or n%4 = 0 (if PackCount = 4), it is not true, we can test two neighbourly items by one SIMD operation for n = 2. Algorithm 4. SIMD IsIntersected Algorithm Input : SIMDRegister QRQL and QRQH, Minimal Bounding Rectangle MBR, Space dimension n Output: true if QR intersects MBR, otherwise else 1 for i ← 0 to (n / PackCount) - 1 do 2 mm prefetch(MBR.QL [i * PackCount ]), MM HINT NTA); 3 MBRQL = mm load ps(MBR.QL [i * PackCount ]); 4 resultReg = mm cmplt ps(QRQH [i], MBRQL); if mm movemask ps( resultReg) != 0 then 5 return false; 6 7 mm prefetch(MBR.QH [i * PackCount ]), MM HINT NTA); 8 MBRQH = mm load ps(MBR.QH [i * PackCount ]); 9 resultReg = mm cmpgt ps(QRQL [i], MBRQH); 10 if mm movemask ps( resultReg) != 0 then 11 return false; 12 13 end 14 return true; 5
The experiments were executed on an Intel Xeon X5670 2.93Ghz, 12.0 MB L2 cache; 96GB of DDR333; Windows Server 2008 R2.
Processing of Multidimensional Range Query Using SIMD Instructions
231
Algorithm 5. SIMD IsInRectangle Algorithm Input : SIMDRegister QRQL and QRQH, Tuple T, Space dimension n Output: true if T is in QR, otherwise else 1 for i ← 0 to (n / PackCount) - 1 do 2 mm prefetch(T[i * PackCount ]), MM HINT NTA); 3 reg = mm load ps(T[i * PackCount ]); 4 resultReg = mm cmplt ps(reg, QRQL [i]); if mm movemask ps( resultReg) != 0 then 5 return false; 6 7 resultReg = mm cmpgt ps(reg, QRQH [i]); if mm movemask ps( resultReg) != 0 then 8 9 return false; 10 11 end 12 return true;
5
Experimental Results
In our experiments, we compare the performance of range query processing using conventional and SIMD algorithms. Since we want to show the scalability of the SIMD implementation, we show results for the R∗ -tree and sequential scan in a data set. In our implementation, 128-bit SIMD registers are used. All experiments are executed with software prefetching and aligned memory access. All data structures have been implemented in C++ and compiled for x86 a x64 by Microsoft C++6 . The improvement of x64 was 2–7% for integer values and 8–17% for float values compared to x86. Since this improvement is the same for conventional and SIMD algorithms, we depict results only for x86. In our test, we utilized 8 real collections. The first collection, titled XML, represents a set of paths in an XML document [16]. We selected a subset generated from the XMark collection [24] including 8-dimensional tuples. The second collection, titled CARS7 , includes spatial records related to California, USA. The third collection, titled METEO8 , includes 5-dimensional tuples related to a meteorological measurement in the Czech Republic, Europe. The fourth collection is TIGER [30], that is a standard Tiger/Line data set used for testing spatial databases. We chosen a Wyoming data set from 2006 and index type 2. It includes 2D points and we do not include any topological information. We extracted over 5 million of points from this set. The fiveth collection WORDS9 contains five text collections in the form of bags-of-words. The sixth collection STOCKS10 represents historical stock data from 1970–2010, including daily open, close, low, high and trading volume figures. The seventh collection, titled WEATHER11 , contains daily maximum and minimum temperatures. A purpose 6 7 8 9 10 11
http://msdn.microsoft.com/visualc http://www.census.gov/geo/www/tiger/ http://portal.chmi.cz/ http://archive.ics.uci.edu/ml/datasets/Bag+of+Words http://www.infochimps.com/datasets/nasdaq-exchange-daily-1970-2010open-close-high-low-and-volume ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/daily/grid/
232
P. Chovanec and M. Kr´ atk´ y Table 1. Test Data Collections
Collections XML CARS METEO POKER TIGER WORDS STOCK WEATHER Dimension 8 4 5 11 2 3 11 7 #Tuples 15,884,160 3,318,583 9,989,993 1,000,000 5,889,786 483,450,157 19,610,499 98,947,174 Domain Integer Float Integer Integer Integer Integer Float Float Tuple Size [B] 32 16 20 44 8 12 44 28 Size [MB] 349 106 199 23.4 112 7,720 997 2,970
of the last collection POKER12 is to predict poker hands. In Table 1, we see some characteristics of these data collections. In Table 2, the basic characteristics of created R-trees are depicted. The node size is 2,048 B, the average utilization is 55-65%. In the case of the sequential scan, we consider an array with the 100% utilization of pages. Table 2. Basic Characteristics of Created R-trees Data Collections Tree Height #Inner nodes #Leaf nodes Index size [MB]
XML CARS METEO POKER TIGER WORDS STOCK WEATHER 5 3 4 5 3 5 6 5 35,235 1,400 7,409 3,176 844 179,877 75,495 176,574 538,625 49,127 199,009 35,345 53,187 6,858,253 835,725 2,943,405 1,175 103 423 78.9 105 13,400 1,730 5,955
When the disk access cost is considered [18], the sequential scan in a data collection means sequential reads from the secondary storage, on the other hand, the R-tree reads pages randomly. Although the random read is generally a more expensive operation than the sequential read, even when we turn SSD into account, there are some works showing that it is possible to process a range query with a set of sequential reads [7]. The disk access is the most expensive operation during the range query processing, but we want not to compare this cost for the sequential scan and the R-tree in this article, therefore we quite ignore it. We only compare the conventional and SIMD range query algorithms. In our experiments, we test the performance for 40 range queries for each data collection divided to 4 query sets according to the selectivity. The first query set contains range queries with the highest selectivity; the result size is 1. The second and third query sets include range queries with the result size in 2, 999 and 1,000, 99,999, respectively. And finally, the fourth query set includes range queries with the lowest selectivity, the result size is in 100,000, 3,000,000. All range queries are repeatedly executed 10× and results are averaged. In Table 3, results of the sequential scan are depicted. Evidently, the number of IsInRectangle invocations is equal to the number of tuples in the data collections. Since these data collections include only point data, IsIntersected is not invoked in this case. We see that the number of cycles (of the conventional algorithm) is close to 1 for queries with higher selectivity. When a tuple is matched by a query rectangle, all coordinates must be compared and the number of cycles is equal to the space dimension. Consequently, the number of cycles per 12
http://archive.ics.uci.edu/ml/datasets/Poker+Hand
Processing of Multidimensional Range Query Using SIMD Instructions
233
Table 3. Query Statistics for Sequential Scan Collection (Dimension) XML (8)
CARS (4)
METEO (5)
POKER (11)
TIGER (2)
WORDS (3)
STOCKS (11)
WEATHER (7)
Query Result Time [s] IsInRectangle IsInRectangle Group Size Conventional SSE Avg. Cycle Calls 1 1.0 0.143 0.198 1.071 2 529.4 0.197 0.223 1.602 15,884,160 3 38,526.8 0.291 0.252 2.537 4 320,312.2 0.514 0.277 4.668 1 1.0 0.0344 0.0404 1.0001 2 619.4 0.0352 0.0408 1.0056 3,360,277 3 58,655.6 0.0423 0.0418 1.3679 4 1,443,912.0 0.0701 0.0491 2.4786 1 1.0 0.0825 0.1184 1.0020 2 334.4 0.0856 0.1225 1.071 9,989,993 3 24,585.0 0.0944 0.1265 2.1758 4 1,416,650.0 0.1665 0.1376 2.1929 1 2 3 4 1 2 3 4
1.0 283.2 22,194.8 425,567.3 1.0 510.6 40,992.1 856,908.4
0.015 0.022 0.041 0.063 0.049 0.048 0.052 0.067
0.014 0.019 0.027 0.033 0.063 0.064 0.064 0.064
1.275 2.125 4.827 8.021 1.000 1.002 1.079 1.445
1. 2. 3. 4. 1. 2. 3. 4. 1. 2. 3. 4.
1.0 528.1 23,922.5 475,487.9 1.0 414.2 27,072.1 906,472.7 1.0 630.3 48,201.8 554,468.2
2.190 2.349 2.970 3.555 0.290 0.304 0.488 0.644 1.154 1.172 1.344 1.718
3.201 3,209 3.232 3.523 0.292 0.295 0.321 0.397 1.381 1.422 1.441 1.478
1.000 1.115 1.429 1.709 1.414 1.5 2.1 3.793 1.017 1.032 1.261 1.830
1,000,000
5,889,786
483,450,157
19,610,499
98,947,174
operation increases when the result size increases. Evidently, the SIMD range query algorithm is useful for queries with low selectivity in the case of the sequential scan, therefore we need to know it before the range query is processed; we must utilize a method for result size estimation [20,4]. In the previous section, we described that the SIMD implementation is more efficient if the number of cycles > the number of packed data processed / 2. In the case of SSE using 128bit registers, the number of cycles must be greater than 2. In the case of a higher selectivity, we see the number of cycles is nearly 1 and the SIMD implementation can not by useful. When the result size increases, the number of cycles increases and SIMD algorithms outperforms the conventional algorithms. In Table 3, we see that for a low selectivity the SIMD range query algorithm is up-to 2× faster then the conventional algorithm. Although it seems that it is sometimes appropriate to combine SIMD and the conventional algorithms, it is not truth. Let us consider the query group 4 of the XML collection. The average cycle value is 4.668. A straightforward solution is to process the first 4 comparisons by SSE and next comparisons (mostly one) by the conventional comparison. However, this technique is less efficient than 2 invocations of the SIMD IsInRectangle (0.34s compared to 0.277s).
234
P. Chovanec and M. Kr´ atk´ y Table 4. Query Statistics for the R-tree
Collection
Query
(Dimension)
Group
XML (8)
1 2 3 4
1.0 529.4 38,526.8 320,312.2
CARS (4)
1 2 3 4
METEO (5)
POKER (11)
TIGER (2)
WORDS (3)
STOCKS (11)
WEATHER (7)
Result
Time [s]
Size Conventional
IsInRectangle
IsIn IsIntersected Is Rectangle Intersected Calls Avg. Cycle Calls
SSE
Avg. Cycle
0.00013 0.00426 0.0096 0.1174
0.00005 0.00214 0.0043 0.0502
5.0357 6.3991 7.4407 7.8425
35.8 2,420.0 48,461.6 344,707.4
3.7329 5.1375 6.3151 7.1563
109.4 1,016.8 3,942.2 18,111.2
1.0 619.4 58,655.6 1,443,912.0
0.00063 0.0029 0.01172 0.3166
0.00054 0.0018 0.0047 0.1721
1.0582 2.1437 3.4319 3.8879
278.4 2,429.4 90,305.6 1,491,049.0
1.2878 1.5277 2.6805 3.6159
343.0 429.0 2,857.0 26,592.8
1. 2. 3. 4.
1.0 334.4 24,585.0 1,416,650.0
0.00029 0.00201 0.0739 0.3222
0.00018 0.0007 0.0419 0.1654
3.084 3.6148 3.8881 4.5519
61.4 1,667.8 71,440.8 1,875,707.0
1.3977 2.3339 3.4057 4.3859
284.2 744.8 6,849.0 57,635.6
1. 2. 3. 4.
1.0 283.2 22,194.8 425,567.3
0.0015 0.0113 0.0708 0.4502
0.0009 0.0052 0.036 0.2227
1.838 3.046 5.038 8.458
104.4 2,005.0 10,641.8 677,512.2
4.717 6.685 9.118 10.425
941.4 3,149.4 6,631.2 37,135.2
1. 2. 3. 4. 1. 2. 3. 4. 1. 2. 3. 4. 1. 2. 3. 4.
1.0 510.6 40,992.1 856,908.4 1.0 528.1 23,922.5 475,487.9 1.0 414.2 27,072.1 906,472.7 1.0 630.3 48,201.8 554,468.2
0.0011 0.00256 0.01036 0.2094 0.0052 0.08222 0.1941 0.5299 0.00081 0.00926 0.0609 0.2359 0.0017 0.0147 0.0897 0.2745
0.0010 0.00219 0.00577 0.1518 0.0048 0.06084 0.1419 0.3921 0.00059 0.00575 0.0312 0.1221 0.0011 0.0087 0.0498 0.1418
1.008 1.185 1.945 1.993 1.070 1.503 1.945 2.306 3.410 4.542 6.370 9.088 3.882 5.423 5.821 5.843
346.0 28,962.0 113,922.4 1,818,593.4 379.4 1,503,499.6 2,930,820.8 6,943,686.4 80.2 17,396.1 244,813.9 2,187,849.0 146.1 11,963.7 581,346.3 2,741,288.5
1.136 1.213 1.702 1.960 1.249 1.541 2.115 2.424 4.103 5.332 6.587 9.304 3.068 4.177 4.951 5.101
929.8 2,493.4 3,479.8 21,833.8 772.8 188,842.4 229,459.2 263,650.0 1,116.8 22,687,6 65,717.6 205,096.5 513.1 3,710.8 78,567.4 375,949.4
Table 4 show query processing results for the R-tree. Although the number of cycles increases for the increasing result size like in the case of sequential scan, the number of cycles is quite higher. It is therefore only relevant MBRs and tuples are accessed (they include closer values), it means the higher count of their values must be compared. Consequently, the SIMD implementation is useful event in the case of higher selectivities. We see that the performance of range queries is up-to 2× more efficient in the case of the SIMD implementation. In the case of a 2-dimensional data collection, it is necessary to process two neighboring tuples by one packed introduction. Table 5 shows that the SIMD comparison of two tuples by one instruction is almost 2× faster than two sequential SIMD comparisons. Evidently, selectivity of queries does not have so high impact on the efficiency of query processing in this case. It is not possible to use this technique for 3-dimensional spaces, since we need wider SIMD registers than 128bits. However, the 256-bit wide SIMD instructions (see [13]) will enable us to process 8 parallel compare operations and compare two 3-dimensional tuples by one packed instruction. In the previous section, we depicted the SIMD algorithms where the query rectangle is loaded into SIMD registers before the range query starts. Although it speeds up the processing time, it is not possible to use it for spaces with dimension > 12 (when we consider 8 SSE registers). If the space dimension is higher, we must load the query rectangle in each IsInRectangle call. In Table 6, we see the result for the three variants: conventional, SIMD where QR is
Processing of Multidimensional Range Query Using SIMD Instructions
235
Table 5. Collection TIGER (2D): Query statistics for Sequential Scan – 2D Tuples Processing Query Result Conventional SIMD - Load and SIMD - Load and Group Size Algorithm Comparison of Comparison of 1 tuple 2 tuples 1 1.0 0.049 0.063 0.034 2 510.6 0.048 0.064 0.034 3 40,992.1 0.052 0.064 0.034 4 856,908.4 0.067 0.064 0.035
loaded before the range query processing starts, and SIMD where QR is loaded in each IsInRectangle call. We see that the last SIMD algorithm outperforms the conventional algorithm for higher selectivities. Table 6. Collection XML (8D): Query statistics for Sequential Scan – QR Loading Query Result Conventional SIMD - QR Loaded SIMD – QR Loaded Group Size Algorithm Before Range Query in IsInRectangle 1 1.0 0.143 0.198 0.225 2 529.4 0.197 0.223 0.259 3 38,526.8 0.291 0.252 0.294 4 320,312.2 0.514 0.277 0.340
6
Conclusions
In this article, we compared the conventional and SIMD algorithms for range query processing. Although we shown that SIMD algorithms are useful only for queries with the low selectivity in the case of the sequential scan, for the R-tree we shown different properties. The number of compare operations processed by a range query algorithm is higher than in the case of the sequential scan and the SIMD algorithm is useful regardless to the selectivity of a query. We shown that the SIMD range query algorithm is up-to 2× faster then the conventional algorithm. In our future work, we would like to compare existing approaches to parallelization of data structures: GPU, APU, and multi-cores CPU using SIMD.
References 1. Ailamaki, A., DeWitt, D.J., Hill, M.D., Skounakis, M.: Weaving Relations for Cache Performance. In: Proceedings of the International Conference on Very Large Data Bases, VLDB 2001 (2001) 2. Bayer, R.: The Universal B-Tree for Multidimensional Indexing: General Concepts. In: Masuda, T., Tsukamoto, M., Masunaga, Y. (eds.) WWCA 1997. LNCS, vol. 1274, pp. 198–209. Springer, Heidelberg (1997) 3. Beckmann, N., Kriegel, H.-P., Schneider, R., Seeger, B.: The R∗ -Tree: An Efficient and Robust Access Method for Points and Rectangles. In: Proceedings of the ACM International Conference on Management of Data, SIGMOD 1990 (1990)
236
P. Chovanec and M. Kr´ atk´ y
4. Bruno, N., Chaudhuri, S., Gravano, L.: STHoles: A Multidimensional WorkloadAware Histogram. In: Proceedings of the ACM International Conference on Management of Data, SIGMOD 2001 (2001) 5. Chen, S., Gibbons, P.B., Mowry, T.C.: Improving Index Performance Through Prefetching. In: Proceedings of the ACM International Conference on Management of Data, SIGMOD 2001 (2001) 6. Chhugani, J., Nguyen, A.D., Lee, V.W., Macy, W., Hagog, M., Chen, Y.-K., Baransi, A., Kumar, S., Dubey, P.: Efficient Implementation of Sorting on MultiCore SIMD CPU Architecture. Proceedings of the VLDB Endowment 1(2) (2008) 7. Chovanec, P., Kr´ atk´ y, M., Baˇca, R.: Optimization of Disk Accesses for Multidimensional Range Queries. In: Bringas, P.G., Hameurlain, A., Quirchmayr, G. (eds.) DEXA 2010. LNCS, vol. 6261, pp. 358–367. Springer, Heidelberg (2010) 8. Drepper, U.: What Every Programmer Should Know About Memory. Technical report, Red Hat, Inc. (2007), http://magazine.redhat.com/ 2007/10/26/ulrich-drepper-memory-article-series/ 9. Freeston, M.: A General Solution of the n-dimensional B-tree Problem. In: Proceedings of the ACM International Conference on Management of Data (SIGMOD 1995). ACM Press (1995) 10. Fung, Y., Ho, T., Cheung, W., Ercan, M.: Parallel Solution for Railway Power Network Simulation. In: Proceedings of IEEE Pacific Rim Conference on Communications, Computers and Signal Processing (2001) 11. Graefe, G., Larson, P.: B-tree Indexes and CPU Caches. In: Proceedings of the ICDE Conference. IEEE (2001) 12. Guttman, A.: R-Trees: A Dynamic Index Structure for Spatial Searching. In: Proceedings of the ACM International Conference on Management of Data (SIGMOD 1984), pp. 47–57. ACM Press (June 1984) 13. Intel. Intel advanced vector extensions programming reference (2008), http://softwarecommunity.intel.com/isn/downloads/intelavx/ Intel-AVX-Programming-Reference-31943302.pdf 14. Kanthak, S., Schutz, K., Ney, H.: Using SIMD Instructions for Fast Likelihood Calculation in LVCSR. In: Proceedings of Acoustics, Speech, and Signal Processing, ICASSP 2000 (2000) 15. Kim, C., Chhugani, J., Satish, N., Sedlar, E., Nguyen, A.D., Kaldewey, T., Lee, V.W., Brandt, S.A., Dubey, P.: FAST: Fast Architecture Sensitive Tree Search on Modern CPUs and GPUs. In: Proceedings of the ACM SIGMOD Conference (2010) 16. Kr´ atk´ y, M., Pokorn´ y, J., Sn´ aˇsel, V.: Implementation of XPath Axes in the Multidimensional Approach to Indexing XML Data. In: Lindner, W., Fischer, F., T¨ urker, C., Tzitzikas, Y., Vakali, A.I. (eds.) EDBT 2004. LNCS, vol. 3268, pp. 219–229. Springer, Heidelberg (2004) 17. Kr´ atk´ y, M., Sn´ aˇsel, V., Zezula, P., Pokorn´ y, J.: Efficient Processing of Narrow Range Queries in the R-Tree. In: Proceedings of IDEAS 2006. IEEE CS Press (2006) 18. Lightstone, S.S., Teorey, T.J., Nadeau, T.: Physical Database Design: the Database Professional’s Guide. Morgan Kaufmann (2007) 19. Manegold, S., Boncz, P., Kersten, M.: What Happens During a Join? Dissecting CPU and Memory Optimization Effects. In: Proceedings of the International Conference on Very Large Data Bases, VLDB 2000 (2000)
Processing of Multidimensional Range Query Using SIMD Instructions
237
20. Muralikrishna, M., Dewitt, D.: Equi-Depth Histograms for Estimating Selectivity Factors for Multi-Dimensional Queries. In: Proceedings of the ACM SIGMOD Conference on Management of Data (SIGMOD 1988), pp. 28–36 (1988) 21. Rao, J., Ross, K.A.: Cache Conscious Indexing for Decision-Support in Main Memory. In: Proceedings of the International Conference on Very Large Data Bases, VLDB 1999 (1999) 22. Rao, J., Ross, K.A.: Making B+ Trees Cache Conscious in Main Memory. In: Proceedings of the ACM International Conference on Management of Data, SIGMOD 2000 (2000) 23. Samet, H.: Foundations of Multidimensional and Metric Data Structures. Morgan Kaufmann (2006) 24. Schmidt, A.R., et al.: The XML Benchmark. Technical Report INS-R0103, CWI, The Netherlands (April 2001), http://monetdb.cwi.nl/xml/ 25. Sellis, T.K., Roussopoulos, N., Faloutsos, C.: The R+ -Tree: A Dynamic Index for Multi-Dimensional Objects. In: Proceedings of the International Conference on Very Large Data Bases (VLDB 1997), pp. 507–518. Morgan Kaufmann (1997) 26. Servetti, A., Rinotti, A., De Martin, J.: Fast Implementation of the MPEG-4 AAC Main and Low Complexity Decoder. In: Proceedings of Acoustics, Speech, and Signal Processing, ICASSP 2004 (2004) 27. Shahbahrami, A., Juurlink, B., Vassiliadis, S.: Performance Comparison of SIMD Implementations of the Discrete Wavelet Transform. In: Proceedings of Application-Specific Systems, Architecture Processors, ASAP 2005 (2005) 28. Slingerland, N., Smith, A.J.: Multimedia Extensions for General Purpose Microprocessors: A Survey. In: Technical report CSD-00-1124. University of California at Berkeley (2000) 29. Berchtold, S., Keim, D.A., Kriegel, H.-P.: The X-Tree: An Index Structure for High-Dimensional Data. In: Proceedings of the International Conference on Very Large Data Bases, VLDB 1996 (1996) 30. U.S. Department of Commerce, U.S. Census Bureau, Geography Division. TIGER/Line Files, 2 edn. Alabama, Autauga County (2006), http://www.census.gov/geo/www/tiger/ 31. Willhalm, T., Popovici, N., Boshmaf, Y., Plattner, H., Zeier, A., Schaffner, J.: SIMD-Scan: Ultra Fast In-Memory Table Scan Using On-Chip Vector Processing Units. Proceedings of the VLDB Endowment 2(1) (2009) 32. Yu, C.: High-Dimensional Indexing. LNCS, vol. 2341. Springer, Heidelberg (2002) 33. Zhou, J., Ross, K.A.: Implementing Database Operations Using SIMD Instructions. In: Proceedings of the ACM International Conference on Management of Data, SIGMOD 2002 (2002)
A Comparative Evaluation of Web Service Discovery Approaches for Mobile Computing Nor Azizah Saadon and Radziah Mohamad Software Engineering Department, Faculty of Computer Science and Information Systems, Universiti Teknologi Malaysia, 81310 UTM, Skudai, Johor, Malaysia {azizahsaadon,radziahm}@utm.my
Abstract. Nowadays, with the advancement in mobile technologies, the use of web services has seen an explosion in interest for mobile computing environment. However, the heavyweight nature of conventional web services to be deployed on mobile devices brings new challenges in mobile computing in the coming future. Web services technology and mobile computing domains are converging at their intersection which leads to Mobile Web Services (MWS) which enables the service access in users’ mobile device. As the numbers of web services are increasing dramatically, the web service discovery becomes more important to discover the usable web services in effective and efficient manner. This paper discusses an overview of mobile challenges in MWS recent research, and a summary of the issues. Moreover, current approaches of Mobile Web Services Discovery are classified and these approaches are compared to some criteria. The results of this study will help researchers to deliver more applicable solutions with the most appropriate approaches for Mobile Web Services Discovery. Keywords: Web Services, Mobile Web Service, web service discovery, mobile computing, mobile web service discovery.
1
Introduction
Web services technology and mobile computing are converging at their intersection, known as Mobile Web Services (MWS). This concept has seen an explosion in interest and becoming more important in the development of web services in mobile computing. The worldwide revenue for MWS generated nearly USD 1.2 trillion in 2010 and is expected to break USD 1.7 trillion by the end of 2015 [1]. Nowadays, as the information is available anytime and anywhere from diverse mobile devices, the demands for MWS are growing dramatically as a new paradigm of web services in the mobile computing environment [11][13]. However, several challenges concerning the limitation of mobile devices such as the CPU processing power and smaller memory pose new and unique characteristics for web services in the mobile environment. A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 238–252, 2011. c Springer-Verlag Berlin Heidelberg 2011
A Comparative Evaluation of WSD Approaches for Mobile Computing
239
MWS are web services in a mobile environment whereby a user has to search access and invoke services through a mobile device [5]. In MWS architecture, the mobile service providers publish WSDL (Web Service Description Language) descriptions of services to the mobile broker. Then, the mobile clients discover services from the mobile broker based on the WSDL descriptions. The characteristics of MWS architecture differ from conventional web services in terms of mobile devices which can be a client, provider, and even a broker. There are a lot of issues in MWS that need to be considered which are still in the early stage of research. Thus, the research questions that motivated this study are: – What are the issues of mobile web services that lead to the distinction of the unique characteristics compared to conventional one? – What are the mobile challenges which have driven to the various approaches in mobile web services? – What are the approaches that have been proposed for performance enhancements in mobile web services? Current conventional web services do not take into account about the specific characteristics for mobile computing environment. Therefore, it is necessary to consider the mobile challenges to fit into the web service technologies. This paper discusses an overview of different approaches in a mobile web services recent research and a summary of mobile requirements for web service in mobile computing. The aims of this study are at studying the concrete elements of MWS. The goal of this study differs from existing works, as this study aims to distinguish the unique characteristics of MWS compared to the conventional web services. The MWS issues are discussed and focused on the discovery of web service as it is an important part to discover the most relevant web services that match the request between clients’ request and service offered. Moreover, current approaches in MWS discovery are classified and these approaches are compared with some criteria. The results of this study will help researchers to deliver more applicable solutions with the most appropriate existing approaches to discover MWS. The layout of this paper is as follows: Section 2 discusses the mobile challenges in web service technologies and summarizes its mobile requirements. Section 3 discusses the MWS discovery issues as the focus of this work. Section 4 describes the classification of current approaches in MWS discovery. Section 5 describes the comparative evaluation with some criteria. Then, the discussion of comparative evaluation is given in Section 6. Finally, the conclusion of this work is given in Section 7.
2
Mobile Challenges in Web Services
Basically, MWS perform in a resource constraint environment compared to conventional web service architecture which is mainly designed for the environment of desktops and wired networks. To apply existing web services to mobile environment is not straightforward, since conventional web services do not meet
240
N.A. Saadon and R. Mohamad
the key aspects of mobile computing environment. Based on the exploratory literature study in MWS [2][4][5][13][14][16][17][20][21][23]the issues related to limitations of web services in mobile computing environment were identified. There are device capabilities, network connectivity, mobility, message exchange and security, in which each category has their own characteristics. The following are the details for the identified category. Device Capabilities: Pocket-sized mobile devices differ significantly from laptops or desktops in their shape, size, target user, functionality, features and service. Therefore, the physical constraints of mobile devices can be characterized to the ability of CPU processing power, smaller memory, smaller storage capacities, limited battery power, smaller screen size display, keypad or touch-screen capabilities and smaller bandwidth usage [11][13]. The limitation of processing power is tightly affected by the limitation of battery life because the battery life will decrease with a lot of computation. Besides that, the limited screen size display in mobile devices requires only important aspects of user interface to be displayed. The type of output display is another issue that regarding if the mobile devices can support the format. In addition, the limitation of keypad or touch-screen in mobile devices should consider the symbols and characters to be input and also the inputs amount should be kept in minimum. Work in [4], [5] and [21] consider the device capabilities in their approaches. The specific features of device specifications are exploited from the Composite Capabilities/ Preference Profile (CC/PP) standard. Network Connectivity: Limited bandwidth and high latency are the characteristics in MWS which normally will operate in wireless network connection[23]. Smaller bandwidth capability in mobile network leads to the limited data rates because the cost charged by mobile operator is expensive. Moreover, in wireless network environment, intermittent connectivity and network disconnection are the issues because of the stability of the wireless network itself. When the mobile user moves out of network range, this prompts to the disconnection. In such scenario, the request message or response message may fail to be delivered to the correspond requester. Work in [13] overcomes this issue by migrating the web services to adjacent mobile device. This is to provide continuous functionality with unstable network connectivity. Mobility: Mobility is one of the distinguished characteristics in MWS compared to the conventional web services. The portability and dynamicity of the mobile devices in mobile environment is the challenge where the changes of context and location information may change rapidly or unpredictably. Thus, work in [16] shows the examples that consider this issue. They consider the user context such as location properties as one of the elements in the discovery process. Message Exchange: Performance is the most significant issue in the message exchange for mobile environment due to its characteristics in mobile computing. To process XML (Extensible Markup Language) and SOAP (Simple Object Access Protocol) messages in MWS is quite challenging whereby XML messages
A Comparative Evaluation of WSD Approaches for Mobile Computing
241
generate a larger overhead that is not suitable for mobile devices with limited processing power. Work in [2],[13]and[20] try to reduce SOAP messages exchange and to reduce size of messages by eliminating unnecessary message information in order to optimize the performance. Another approach in [14] proposes a getaway-based approach in message exchange and evaluates the response time performance. It is important to identify which part of the message is essential in MWS, which is still an open issue. Security: Security is an important aspect in mobile environment because mobile devices are portable and can be carried around. The IP address of the users device may change as the person moves to another network. Furthermore, with the limitation of network connectivity where communication can be disengaged, important applications such as mobile payment need to be considered thoroughly in the context of security if the network fails. Data and peer authentications are the important aspects in MWS and work in [17] considers the security aspect in their implementation. To summarize, the following mobile requirements as research challenges for the performance enhancement in MWS are identified: – Device specifications that can support available web services. – Heterogeneity and diversity of mobile devices with different capabilities and platform. – The changes of user context and web services context. – Unstable network connectivity requires migration of web services. – Requirement for lightweight message exchange for mobile device. – Security and authentications are important aspects in mobile computing. From the discussion above, MWS has to be designed within the limits of the available device constraints. Although the technologies of mobile devices are evolving, the development of MWS has to meet the mobile challenges. Therefore, in order to deploy web services in mobile computing environment, these mobile requirements should be taken into consideration.
3
Mobile Web Service Discovery
Web services have become one of the most promising approaches to establish service-oriented concept in mobile computing environment. Current research trends mostly focus on the entire web service life cycle, which are developing, deploying, publishing, discovering, composing, monitoring and optimizing the access to web services [27]. Nowadays, the overwhelming proliferation in web services which contributes to the similarity of web services functionality are increasing. This will cause the task to differentiate between several web services to become very complex and time consuming [5]. Therefore, finding relevant web services that match user request and what service provider offers is a challenge. Web service discovery has become essential to offer effective mechanism in order to discover the most relevant web services in mobile computing.
242
N.A. Saadon and R. Mohamad
According to the W3C (World Wide Web Consortium), web service discovery is defined as ”an act of locating a machine-processable description of a Web service that may have been previously unknown and that meets certain functional criteria”[26]. The typical interactions during web service discovery process involve publishing, finding and binding process. Service providers publish and store their service descriptions in public service registry such as UDDI (Universal Description Discovery and Integration). Then, the service requestors or client will search the desired web service by looking up for the service descriptions stored in the public service registry. Candidates of appropriate web services are listed based on the user requirement. Service requestors then invoke the web service that matches their needs. This paper concentrates on the web service discovery for mobile computing and provides a classification of mobile web service discovery approaches in the next section.
4
Classification of Mobile Web Services Discovery Approaches
As discussed in the previous section, conventional web services do not meet the key aspects to be deployed in mobile devices. Therefore, research efforts in MWS have to be designed within the limits of the available device constraints. The existing discovery approaches based on their considered matching elements specifically in mobile computing environment are classified into three categories as in the following: 4.1
Context-Aware Discovery
Context information represents a part of non-functional properties in web services and can be interpreted differently from various perspectives. Different definition of context gives different impacts, specifically in mobile computing environment. For example, the context definition by Dey [9] is quiet general for this study in order to classify the web services approaches. This paper concentrates on mobile web service discovery and adapted Doulkeridis’s definition [10] as the basis to describe the context as ”the implicit information related both to the requesting user and service provider that can affect the usefulness of the returned results”. Context-aware discovery can be defined as the ability to make use of the context information to discover the most appropriate web services to the client against the service offered. Generally, two types of context are identified, which are the user context and web service context. User context can be the user profile, device capabilities, user preferences, location, temporal constraints and time, which are characterised by the users current situation. On the other hand, the properties for web service context can be the provider identity, location of the service, cost and payment method. All these context properties are considered in this paper to classify the approaches of MWS discovery to be in this group.
A Comparative Evaluation of WSD Approaches for Mobile Computing
243
Among the approaches that consider the capability of mobile devices limitation is MobiEureka approach [4][5]. It considers device capabilities and user preferences by exploiting the Composite Capabilities/Preference Profile (CC/PP) standard defined by W3C. By using the information extracted from the device profile, relevant MWS are ranked based on the degree of relevance between the service offer and clients’ device. Another approach by Peng et al., [21] takes a step further in discovery mechanism by performing semantic matching between the service requester and service provider. It extends the service profile with the user profile information. A similar approach, namely AIDAS [25] also supports the semantic-based matchmaking by exploiting the user profile, device profile and service profile properties in the discovery process. AIDAS relies on its own ontology as it focuses on the service capabilities compared to the existing matchmaking algorithms and languages which are more focused on the service input and output. Approach in [13] proposes framework for hosting web services in mobile devices with the discovery manager as one of the components. Nonetheless, this approach is not focused specifically in discovery methods but still takes into account the device limitation. Work in [24] optimises the reasoning process to discover the most relevant web services in mobile devices. This approach presents semantic matching in mobile device by developing mTableaux algorithm but does not consider the ranking of web services or any methods of similarity calculation. 4.2
QoWS (Quality Of Web Service) -Aware Discovery
This category of matching elements concerns Quality of Service (QoS) aspects which are non-functional properties in web services. QoS refers to how well a service performs its behaviour to the customer. As the numbers of available web services are increasing, there will be a lot of competitors in service provisions that provide similar functionalities. Thus, when user requests for service, it will result in a long listing of available services in the users mobile devices. Nevertheless, with the constraints of mobile devices such as limited screen display, this will cause the user to scroll through the long list of web services candidates on mobile device. QoWS-aware discovery seems to be the most significant mobile service research recently, which can help the users to discover the most appropriate web services based on QoS properties. Alongside, Quality of Web Service (QoWS) is introduced in [27] to adapt the concept of QoS in web services. Generally, both QoS and QoWS refer to the similar aspects of quality as well. Work in [27] divides the QoWS into two types of quality: runtime quality and business quality. Runtime quality represents the response time, reliability, availability, accessibility and integrity. Business quality refers to the cost, reputation, and regulatory. This paper uses the quality parameters in [27] as a basis to classify the mobile web service approaches to be in this category. The approach in [3] introduces Web Service Relevancy Function (WsRF) for measuring the relevancy of requested service with QoWS properties. The input parameters from the clients are based on graphical user interface and then the
244
N.A. Saadon and R. Mohamad
relevant services are ranked based on the WsRF. They adapted quality parameters such as response time, throughput, available, accessibility, interoperability analysis and cost. However, this study does not explain clearly on how the graphical user interface input method functions for this framework. Another approach in [6] proposes the Evolutionary Computing using Particle Swarm Optimization (PSO) technique in order to discover the most relevant services as QoWS properties as part of the query parameters. The considered QoWS are execution cost, execution time, availability, successful rate, reputation and frequency. This approach considers the semantic service discovery process plug-in together in composition phases by extending the semantic matching algorithm with simple PSO algorithm. QoWS score are calculated by similarity matching and relevant services are ranked based on QoWS score. Work in [28] proposes semantic matching by using vector space model to calculate semantic distance followed by an ontology hierarchical matching. Then, the concept of semantic metric matching for QoWS properties is introduced. Nevertheless, the authors of this paper did not mention the QoWS parameters that they considered. They developed upper ontology QoS-based web services that complement OWL-S but did not describe in details about the ontology on how QoWS properties will help the process of searching the relevant web services to suit the clients need. Another similar approach by Zhu and Meng [29] utilises OWL-S as a service description language and designs a prototype of service discovery which adopts Three-layer Pervasive Semantic Service Matching Algorithm (TPSSMA). The algorithm matches web services based on the category of service, input/output parameters and QoWS properties. Authors in this paper also did not provide the QoWS parameters that they considered. In addition, they did not explain how they will attach QoWS properties to the web service specifications. 4.3
Hybrid-CQ (Context-QoWS) Discovery
Another category that takes a step further, namely Hybrid-CQ (Context-QoWS) concerning both of the matching elements as discussed above: context and QoWS. However, taking both non-functional properties into consideration in service discovery is a challenging issue. There is a trade-off between the query response time and performance of relevant web services discovered as a lot of mechanisms are used to cater both properties. Work by Niazi and Mahmoud [19] considers the user preferences, device capabilities and also QoWS properties for discovering the most relevant MWS. Their approach is similar with MobiEureka approach [5] by using device capabilities and user preferences as the matching elements that exploit the Composite Capabilities/Preference Profile (CC/PP) specification. However, they enriched it by proposing ontology-based namely, Profile-Based Context-aware Ontology (PBCO). It is defined in OWL (Web Ontology Language) and it has been integrated with an ontology released by W3C recently, as
A Comparative Evaluation of WSD Approaches for Mobile Computing
245
Delivery Context Ontology. QoWS properties that they considered are the availability, response time, throughput, latency and reliability which are attached in the PBCO. The advantage of this approach is the type of output format that can be displayed on mobile devices is considered. EASY approach in [16] introduces EASY-L (Language) that extends the language in OWL-S as the service capabilities for matching the functional and non-functional properties, and EASY-M (Matching) as for matching the service functional capabilities comparing two ontology concepts. Then, they calculate and rate the service between the matching service requests with the service advertisement. Their approach shows that combination of functional and nonfunctional properties (context and QoWS) provides the most relevant web services to the user and is even capable with the device constraint. ConQo[7] is another approach in this category that combines context and QoWS properties for the enhancement of web service discovery. ConQo utilises WSMO and introduces WSMO-QoS and WSMO-Context. Another approach in [12] proposes a personalised web service framework for discovering and selecting web services considering the context and QoWS. However, both papers, [7] and [12] did not expose the matching strategy and similarity calculation to provide ranking in details. They only provide general mechanisms to apply in service description and consider the subset of context and QoWS properties only.
5
Comparative Evaluation Criteria
This section discusses the comparison of above MWS discovery approaches with respect to some criteria. This work is mostly inspired by [8][15][18] and aims to consider, select and integrate the most appropriate criteria by those works for discovering MWS. In this paper, the criteria considered are: matching elements, matchmaking strategy, support similarity calculation, several stage discovery/matching and support QoWS. Matching Elements: The elements between the service requestor and service provider are considered for matching. The followings are the description of elements: – IO: Functional input and output for the web services. – PE: Preconditions and Effects of the web services. – Non-Functional Properties: User Context (device capabilities, user preferences, user profile, location, time, temporal constraints), Web Service Context (name, business category, provider identity, location of the service, cost, payment method) and Quality of Web Service (QoWS) – Other: Interface parts (operation names, operation numbers), WSDL descriptions and textual descriptions for the function of the service. Matchmaking Strategy: The matchmaking strategy uses the matching elements as the input or output of matching. The classified matching strategy below is based on the work in [22] and is categorised into four categories:
246
N.A. Saadon and R. Mohamad
– Keyword-based: The query to find relevant web services based on a list of keywords as an input from client. Keyword-based matchmaking is easy to be implemented but it suffers a huge list of result as it cannot understand users actual intention. – Syntactic-based: The query uses keyword as a part of filter mechanism and then any algorithms are used to refine the list of web service candidates based on the web services function, interface, operation or structure. – Semantic-based: This strategy of matching exploits the advantage of ontology. The web service description file such as OWL-S, WSDL-S and WSMO are important in order to match the service request and service offer based on the ontology. – Pragmatic-based: This kind of strategy considers the context properties as a part of matching process. The matching strategy is any of the above strategy or combination of those three matching strategies. Support Similarity Calculation: The similarity calculation is based on the partially matching between the service requestor and service offer. With this partial match, the score to rank the appropriate services are calculated according to the desired input score. Therefore, the most relevant web services to the user are listed as the high score ranking. Any of the approaches that enable ranking of the matching results based on the user request is assigned to this criterion. Several Stages Discovery/Matching: Web service discovery with several stages leads to the meaningful and effective result. Any of discovery approaches that perform several stages of matching is assigned to this criterion. Accuracy: This criterion is defined as the accuracy of matching requirement during the discovery process between the service requestor and service offer. This criterion is used as a benchmark for the approaches that are being compared in this paper. The overall results of four criteria previously are used as the input score in the followings: – Matching Elements: The more elements supported for an approach, the more accurate it is. One score is attributed to it if any of the elements (IO, PE, Non-Functional Properties, Other) are supported by the approach. – Matchmaking strategy: The more matching strategy takes into account for an approach, the more accurate it is. One score is attributed to it if any of the strategies (keyword-based, syntactic-based, semantic-based, pragmaticbased) are considered by the approach. – Support similarity calculation: If an approach supports similarity calculation, it is more accurate compared to the approach without this support. Thus, one score is attributed to it. – Several stages discovery/matching: If an approach supports several stages matching, it is more accurate compared to the approach without this support. Thus, one score is attributed to it accordingly.
A Comparative Evaluation of WSD Approaches for Mobile Computing
247
To determine the degree of Accuracy, it is evaluated as ’High’ if an approach in MWS discovery contributes a score between 8-10. It is evaluated as ’Average’ if the score is between 4-7 and set as ’Low’ if the score is between 1-3. Support QoWS Properties: Quality of Web Service (QoWS) refers to the non-functional properties that characterise web services overall behaviour. It becomes significant to measure the degree of desired web services with the combination of QoWS parameters. QoWS properties are adapted from the work in [27] as a basis of quality parameters in our work. It is divided into runtime quality and business quality [27]. – Runtime quality: It indicates the measurement of quality properties during the runtime of web services operation which are response time, reliability, availability, accessibility and integrity. If any of the approaches consider at least one of these parameters, it is assigned to this criterion. – Business quality: It represents the assessment of web service operation in the business perspective which are cost, reputation, and regulatory. If any of the approaches consider at least one of these parameters, it is assigned to this criterion. Quality: This criterion is defined as what reflects to the above Support QoWS properties. To determine the degree of Quality, if an approach in mobile web service discovery supports both of the above quality parameters: runtime quality and business quality, it is evaluated as ’High’. It is evaluated as ’Average’ if it supports either one of them and set as ’Low’ if none of the above parameters are considered.
6
Discussion
Table 1 shows the result of the comparison for MWS discovery approaches for each category with respect to the proposed classification of matching elements in MWS discovery. For this comparison, this study relies on the criteria in the matching elements, matchmaking strategy, support similarity calculation, support several stages discovery/matching, and support QoWS properties. The classified mobile web service discovery in the study is divided into three groups: Context-aware Discovery, QoWS-aware Discovery and Hybrid-CQ (Context-QoWS) Discovery. The classification of this study focuses on nonfunctional properties matching elements and the scope is restricted to web service discovery in mobile computing environment. The approaches that apply semantic-based matchmaking strategy rely on service inputs and service outputs (IO) of matching elements as those are the important parts in semantic solution to discover matching services [6][16][19][21][24] [25][28][29]. Service inputs and outputs are defined using ontology concepts semantically where OWL-S is used in these approaches as it is a widely used
248
N.A. Saadon and R. Mohamad Table 1. Classification of Mobile Web Services Discovery Approaches
language. However, the approach in [7] with semantic-based matchmaking strategy does not rely on the service inputs and outputs, as it considers non-functional aspects as the matching elements as well. All of the approaches compared were considered non-functional properties (NFP) as it is necessary in order to discover relevant web services instead of functional properties. Nevertheless, none of the approaches use preconditions and effects (PE) elements because of the optional condition which does not contribute much impact to the discovering process. The approaches in [12] and[19] use other matching elements whereas in [19], it considers interface, while in [12] it considers WSDL (Web Services Description Language) descriptions to express the input matching elements. For the matchmaking strategy, there are four categories that are keyword-based, syntactic-based, semantic-based and pragmatic-based. The discussion details for these groups of matchmaking strategy can be found in [22]. The approach in [12] uses keyword-based as a query. The disadvantage of using keyword-based is it cannot understand user’s real intention and it is a major drawback in matchmaking strategy. However, the work in [19] overcomes the capabilities of keywordbased as a filter mechanism and then using semantic-based and pragmatic-based matchmaking strategy in discovering process. Approaches in [3][5][13] use
A Comparative Evaluation of WSD Approaches for Mobile Computing
249
syntactic-based whereas they use keyword as the first query process and then refining the result by adapting algorithm to discover the relevant web services. Approach in [5] uses device-aware computation algorithm and exploits metric to represent the device specification for discovering process. Similar approach in [3] represents the QoWS properties in metric and uses QoWS-aware computation algorithm to their approach. Semantic-based strategy is the most significant strategy that most of the approaches used [6][7][16][19][21][24][25][28][29]. It uses ontology to represent the matching elements semantically. Most of the approaches use pragmatic-based strategy except the approaches in QoWS-aware category [6],[28] and[29]. Pragmatic-based uses context as matching elements and it is a hybrid of any three matchmaking strategies. The context characterises current situation of the user’s context and web service context in order to make searching strategy efficiently match with the user request. Most of the approaches support similarity calculation except the approaches in [12],[13],[24]and[25]. Similarity calculation takes into account when partial match occurs. The relevant web services are ranked based on the calculated score. For example, the approach in [5] uses Device-aware Ranking Function (DaRF) to measure relevant mobile web services and ranks each of them according to the capability of device-specific features. This is similar to the approach in [3] which uses Web Service Relevancy Function (WsRF) to measure relevant mobile web services and ranks each of them according to the QoWS parameters that they have considered. For the several stages of discovery/matching criteria, most of the approaches consider these in their approach except work in [12],[13] and[25]. Web service discovery with several stages leads to the meaningful and effective result. However, there is a trade-off between the quality and response time of matching web services that leads to the complexity of discovery [18]. As the result, the accuracy criterion is the measurement for all the criteria discussed above. The approach in [19] resulted in a high accuracy in this comparison with most of the matchmaking requirements are fulfilled. Approaches in [3],[12],[13] and[25] are measured as having low accuracy as not many requirements are fulfilled. The remaining approaches, measured as having average accuracy with most of the approaches fall into this degree. For the Support QoWS criterion, it is divided into business quality and runtime quality. Approaches in category Context-aware Discovery obviously do not consider QoWS properties. Nevertheless, approaches in category QoWSaware Discovery mostly take into account the runtime quality but not all approaches in this category are considered having business quality. Approaches in [3],[6],[16],[21] and[28] are considered having business quality. Most of the approaches do not take into account the business quality because not all web services provide that information. However, it can be a benchmark with a lot of available web services recently to differentiate them. Quality is a criterion to measure the QoWS parameters as discussed above. Approaches in [3],[6] and[16] are measured as having high quality as they consider both of the QoWS parameters. All approaches in category Context-aware
250
N.A. Saadon and R. Mohamad
Discovery are measured as having low quality as they do not consider both of the QoWS parameters. The remaining approaches are measured as having average quality as they consider either one of the QoWS parameters.
7
Conclusion
In this paper, the mobile requirements as boundary conditions and challenges for the development of web services in mobile computing environment are presented. The aim is to provide an overview of Mobile Web Services (MWS) and compare recent works in Mobile Web Services Discovery approaches. These approaches are based on non-functional properties matching elements and are classified into three categories: Context-aware Discovery, QoWS-aware Discovery and HybridCQ (Context-QoWS) Discovery. In each category, an introduction is given and the selected approaches are discussed. Then, the classified Mobile Web Services Discovery approaches with some criteria related to discovery aspects are compared. However, it cannot be claimed that this comparison is comprehensive and exhaustive. The considered criteria can be used as reference to help generally evaluating or selecting Mobile Web Services Discovery approaches. However, the result shows that there is no one prominent approaches fulfills all the criteria. Therefore, it depends on certain scenarios and specifications in order to decide the approach to be selected either in the development or for future research. Acknowledgments. We would like to thank Universiti Teknologi Malaysia for sponsoring the research through the grant with vote number 00J37 and for providing the facilities and support for the research. In addition, we would like to extend the gratitude to our lab members in EReTSEL Lab of Faculty of Computer Science and Information Systems, Universiti Teknologi Malaysia for their ideas and support throughout this study.
References 1. Portio Research Mobile Factbook (February 2011). Portio Research (2011) 2. Adacal, M., Bener, A.: Mobile Web Services: A New Agent-based Framework. IEEE Internet Computing 10(3), 58–65 (2006) 3. Al-Masri, E., Mahmoud, Q.H.: QoS-based Discovery and Ranking of Web Services. In: 2007 16th International Conference on Computer Communications and Networks, pp. 529–534. IEEE (August 2007) 4. Al-Masri, E., Mahmoud, Q.H.: Device-Aware Discovery and Ranking of Mobile Services. In: 2009 6th IEEE Consumer Communications and Networking Conference, pp. 1–5. IEEE (January 2009) 5. Al-Masri, E., Mahmoud, Q.H.: MobiEureka: An Approach for Enhancing the Discovery of Mobile Web Services. Personal and Ubiquitous Computing 14(7), 609–620 (2009) 6. Bhuvaneswari, A., Karpagam, G.: Reengineering Semantic Web Service Composition in a Mobile Environment. In: 2010 International Conference on Recent Trends in Information, Telecommunication and Computing, pp. 227–230. IEEE (March 2010)
A Comparative Evaluation of WSD Approaches for Mobile Computing
251
7. Braun, I., Strunk, A., Stoyanova, G., Buder, B.: ConQo: A Context- And QoSAware Service Discovery. In: IADIS Intl. Conference WWW/Internet, pp. 432–436 (2008) 8. Cardoso, J.: Semantic Web Services: Theory, Tools, and Application. Idea Group Inc, IGI (2007) 9. Dey, A., Abowd, G., Salber, D.: A Conceptual Framework and a Toolkit for Supporting the Rapid Prototyping of Context-Aware Applications. Human-Computer Interaction 16(2), 97–166 (2001) 10. Doulkeridis, C.: CASD: Management of a Context-Aware Service Directory. Pervasive and Mobile Computing 4 (February 2004), 737–754 (2008) 11. Farley, P., Capp, M.: Mobile Web Services. BT Technology 23(3), 202–213 (2005) 12. Jarir, Z., Boumhamdi, K., Quafafou, M.: Context- and QoS Aware Personalization for Web Services: Web Information Extraction Case Study. In: 2009 Fifth International Conference on Signal Image Technology and Internet Based Systems, pp. 401–406. IEEE (November 2009) 13. Kim, Y., Lee, K.: A Lightweight Framework for Mobile Web Services. Computer Science-Research and Development 24(4), 199–209 (2009) 14. Mahmoud, Q.H., Ngo, T.: A Gateway-based Framework for Finding and Accessing Mobile Web Services. In: 2010 IEEE International Conference on Information Reuse & Integration, pp. 213–218. IEEE (August 2010) 15. Mohebbi, K., Ibrahim, S., Khezrian, M., Munusamy, K., Tabatabaei, S.: A Comparative Evaluation of Semantic Web Service Discovery Approaches. In: Proceedings of the 12th International Conference on Information Integration and Web-based Applications & Services, pp. 33–39. ACM (2010) 16. Mokhtar, S., Preuveneers, D., Georgantas, N., Issarny, V., Berbers, Y.: EASY: Efficient semAntic Service discoverY in Pervasive Computing Environments with QoS and Context Support. Journal of Systems and Software 81, 785–808 (2007) 17. Naumenko, A., Srirama, S., Terziyan, V., Jarke, M.: Semantic Authorization of Mobile Web Services. Journal of Theoretical and Applied Electronic Commerce Research 1(1), 1–15 (2006) 18. Ngane, L., Goh, A., Tru, C.: A Survey of Web Service Discovery Systems. International Journal of Information Technology and Web Engineering (IJITWE) 2(2), 65–80 (2009) 19. Niazi, R., Mahmoud, Q.H.: An Ontology-Based Framework for Discovering Mobile Services. In: 2009 Seventh Annual Communication Networks and Services Research Conference, pp. 178–184. IEEE (May 2009) 20. Oh, S., Aktas, M.: Efficiency Benefits in Mobile Web Service Architecture Using Context-store. Information Systems 3(2), 1–21 (2009) 21. Peng, R., Mi, Z., Wang, L.: An OWL-S Based Adaptive Service Discovery Algorithm for Mobile Users. In: 2008 4th International Conference on Wireless Communications, Networking and Mobile Computing, pp. 1–5. IEEE (October 2008) 22. Rong, W., Liu, K.: A Survey of Context Aware Web Service Discovery: From User’s Perspective. In: 2010 Fifth IEEE International Symposium on Service Oriented System Engineering, pp. 15–22. IEEE (June 2010) ˇ 23. Srirama, S., Vainikko, E., Sor, V., Jarke, M.: Scalable Mobile Web Services Mediation Framework. In: 2010 Fifth International Conference on Internet and Web Applications and Services, pp. 315–320. IEEE (2010) 24. Steller, L.A., Krishnaswamy, S., Gaber, M.M.: Enabling Scalable Semantic Reasoning for Mobile Services. International Journal on Semantic Web & Information Systems 5(2), 91–116 (2009)
252
N.A. Saadon and R. Mohamad
25. Toninelli, A., Corradi, A., Montanari, R.: Semantic-based Discovery to Support Mobile Context-Aware Service Access. Computer Communications 31(5), 935–949 (2008) 26. W3C: Web service discovery, http://www.w3.org/TR/ws-arch/#wsdisc 27. Yu, Q., Liu, X., Bouguettaya, A., Medjahed, B.: Deploying and Managing Web Services: Issues, Solutions, and Directions. The VLDB Journal 17, 537–572 (2006) 28. Zhou, L.: An Approach of Semantic Web Service Discovery. In: 2010 International Conference on Communications and Mobile Computing, pp. 537–540. IEEE (April 2010) 29. Zhu, Y.A., Meng, X.H.: A Framework for Service Discovery in Pervasive Computing. In: Engineering and Computer Science (ICIECS), pp. 2–5 (2010)
Planning and Project Control with EXCEL Software in Asalouyeh 5th Olefin Project Adnan Keshvari1,*, Hamed Mirzapour1, Jalil Ebrahimpour1, and Nikzad Manteghi2 1 Planning & Project Control, Morvarid Petrochemical, Morvarid P.C, Assalouyeh, Iran Ph. D, Department of Mathematics, Shiraz Branch, Islamic Azad University, Shiraz, Iran {Adnan.keshvari,Mirzapour.hamed,Jalilebrahimpour}@gmail.com, [email protected]
2
Abstract. Project management is the application of knowledge, skills, tools, and techniques to project activities to meet project requirements. Motivating and preparing the necessities for team working, planning, supervising, resource and cost control to accomplish the project in the specified time are some of the main duties of project management in all over the world. Project management is known as a professional and important job. Planning and project control technique can be a very effective method to be used in project management. This technique uses three managerial fields which are scope, time and cost to achieve its goals. In this paper we state main concepts of project management; also we have demonstrated the application of EXCEL software in planning and project control, especially application of this software in Work Breakdown Structure (WBS) and Weight Factor calculation. In order to gain a better realization, these concepts have been surveyed in Morvarid Petrochemical Company. Morvarid.P.C is the executive of 5th olefin plan that produces liquid Ethylene to feed west Ethylene line in second phase of Asalouyeh petrochemical zone. Keywords: Planning, Project Control, Work Breakdown Structure (WBS), Weight Factor, EXCEL.
1 Introduction Planning and project control has a long history. Since the ancient Egyptians applied this techniques in building massive pyramids took profits until Henry Gant - Father of Project Management - WBS and Gant charts introduced, this knowledge can be during their evolutionary. Over time, new techniques were developed like Program Evaluation Review Techniques (PERT) and Critical Path Method (CPM). This knowledge has completed gradually until the year 1981 Project Management Institute (PMI) [1], in order to serve the project management industry, provided PMBOK Guide. The project management from the viewpoint of this standard includes nine knowledge fields that project planning and control is involved within scope, time and cost management directly [2]. *
Corresponding author.
A. Abd Manaf et al. (Eds.): ICIEIS 2011, Part IV, CCIS 254, pp. 253–264, 2011. © Springer-Verlag Berlin Heidelberg 2011
254
A. Keshvari et al.
Although MSP and Priimavera are the main software of planning and proj oject control, EXCEL software application a should not be overlooked in this knowledge and the major problem in reseearch where face to big projects that have many hum man resources, machineries, matterials and to much various overhead costs, how we cann do calculation weight factors, performance and scheduling complete by using EXC CEL he complex was introduced as the top project in Bushhehr computing power. Now, th province from Ministry of Industries and Mines, in 2008. The complex, also, gaiined I prize, in 2009. It gained the IMPA award as the Iran Project Management Institute second top project in the world, w in 2010. It gained the first place in Comprehenssive Project Management Maturrity Model (CPM3) among Iran’s petrochemical projectss.
2 Process Planning and a Project Control Processes for planning, seq quencing and paralleling necessary for implement a proj oject is set regarding to time req quired for implementation of each activity and the quallity. In fact, the first step in preeparing the plan is complete knowledge and understandding of an economical state off activities to balance three factors inclusive time, ccost and quality. Indeed, projecct control, accurate and complete implementation of pllans designed to maintain the paath in the project to achieve an economic balance betw ween the three factors described by using the own tools and techniques. Project plannning b showed in the form of Figure 1 that is similarr to and control process can be PDCA1 [3].
Fig. 1. 1 PDCA in Planning and Project Control
3 Project Triple Con nstraints Projects likes any other afffair have their own limitations. These constraints in proj oject management traditionally include i scope, time and cost. These three factors togetther are called the project manag gement triangle.
1
Planning, Doing, Checking, Acting. A
Planning and Project Control with EXCEL Software in Asalouyeh 5th Olefin Project
255
Time constraint is about the amount of time available to complete the project, the cost constraint, the amount of funds available to complete the project and scope constraint is processes that must be done to project the final results to be obtained. These three constraints are interacting together. It should be considered that changes in project scope changes two other constraints, while the change in time or cost constraints, do not necessarily cause changes in two other constraints. There is another point of view which introduces quality as another constraint. However, planning and project control looks for management of the three major constraints for projects.
4 Scope Management Project scope management contains necessary processes to ensure all work carried out needed for the successful completion of project is done. Scope management is generally related on what project includes what project does not included. In scope management, preparation of WBS is very important. WBS is the process of dividing project into deliverables items and smaller components of work for more management capacity. Work Breakdown Structure, is an analysis of hierarchical work items which must be performed by the project team to provide project objectives and deliverables required to create items to be implemented in which each lower level of WBS, indicating more detailed definitions of project works. WBS organizes and defines the entire project scope [2]. 4.1 Steps to Create WBS • • • • •
Careful study of the contract to determine the Contractor Scope of Work exactly Modeling from WBS made from previous projects (If the project WBS model is close to the project in hand in terms of the nature, the work will be easier) Consulting professionals and experts (Expert Judgment) Studying drawings Determining the zero level to work package level based on information obtained from previous steps. In Primavera software work package level as the last level of WBS is quite observable.
Important point is that work breakdown rate depends on the data collected from the project, the amount of time that project requires later and also the importance of precision and control required on the project. Therefore, projects should always develop work breakdown level to the extent that the project control and information is available since if the level is low, may be loss any activities and over than it also cause very difficult to control. Another point is that the WBS is not unique; that is, two or more types of Work Breakdown Structure could be correct. After completing the WBS, activities should be defined. This phase is also of great importance because those activities are assigned the resources and in planning phase assigned duration and relations. Another point that should be mentioned about
256
A. Keshvari et al.
milestone, as an activity with zero time, is that important events sometimes project expressed an important prerequisite for their activities is after. 4.2 WBS in Terms of Appearance Used in the Morvarid Petrochemical In Morvarid projects two WBS types is generally used. •
Cascade WBS
In this type of WBS all levels are located below each other and the distinctions between levels are determined by their color. By the introduction of Office 2007 and increasing the number of EXCEL rows and columns amount to 1,048,576, 16,384, the problem for large projects such as petrochemical in order to use this structure was solved. The advantage to this structure is the reporting ease and elegance [4]. This type of structure is used in MSP and Primavera software and in Figure 2 a view of the WBS structure in EXCEL.
Fig. 2. Cascade WBS
•
Engineering WBS
In this case, usually the work breakdown of each level is stored in an EXCEL sheet and each branch of the work breakdown in a separate file. They are all linked together and any final data files are transferred to zero level of main file (Figure 3).
Planning and Project Contro ol with EXCEL Software in Asalouyeh 5th Olefin Project
Branch
257
Main File
Fig. 3. 3 How to Link File in Engineering WBS
This structure form is th he first type of the WBS which was very good accordding to the row and column restriictions in older versions of EXCEL. On the other hand beeing separate files, this feature provides to each unit to update its files related to the action and WBS will be up pdated when transfer the sub-files to main file, (in ordeer to update WBS can be passed d across the following in EXCEL: (Edit →Link →Upddate Values) [5]. This type of WBS also has h disadvantages that can be pointed out a lot of files and high volume of Print and weakness w in providing integrated report. Figure 4 show ws a part of the pre-commissioniing WBS of Morvarid petrochemical. 4.3 Methods of Weight Factor F Calculation After recognizing WBS, acttivities and relations between them, duration for complettion of activities and resources such as human resources, machinery and materials, shoould be assigned. In fact, the value v of each activity depends to the time and resourrces required for completion. But, normally, degree of difficulty is hidden in thhese two categories. In huge projjects in Iran, because of lack of personnel for prediction and allocation of resources, weight factor is used to determine the value of each activvity which is very helpful in callculating the project progress. Different ways of calculatting the weight factor based on o experiences of the fifth olefin project are introduuced below. •
Based on manpoweer and machinery
In this method, W.F of the head h branches or high levels can be determined by assiggned manpower and machinery to all the activities and valuation of each source and calculating the total value of project - which is sum product resource unit costt to amount of resource- and thrrough the formula (1) in Excel and with roll up down too up weight factor of higher leveel calculated. % %Wf
V V
100
(1)
258
A. Keshvari et al.
Fig. 4. Engineering WBS
This method is used for valuation of the activity level to higher level. This type of weight value is calculated in lump sum contracts, and calculating basics are the same as software Primavera and MSP. This method has been used in commissioning stage in Morvarid petrochemical company.
Planning and Project Control with EXCEL Software in Asalouyeh 5th Olefin Project
259
Table 1. Calculating W.F Based on manpower and machinery
•
Based on contract price and consumed sources
This type of W.F is usually used in the unit rate of contracts that are more common in civil works. The calculation method is the same as above method, but the amount of consumed sources for each activity is estimated on execute plans and then is multiplied by the unit price and subtotal is calculated for the entire project. With this method, all activities unit will be changed to currency and will be comparable. Then W.F will be calculated through the formula (2) in Excel. %Wf
V V
(2)
For example, Table 1 shows how to calculate W.F of a building foundation. In the civil works such as yard building, official and industrial buildings in Morvarid petrochemical Co. this method has been used. Table 2. Calculating W.F Based on contract price and consumed sources
260
A. Keshvari et al.
•
Based on agreement between owner and contractor
This kind of W.F is more common in fixed price or lump sum contracts and called weight value that last experiences affect in it. To calculate this type of W.F, project manager’s signature and minutes of meeting and verify weight factor of first level and sometimes second level. Then planning and project control team calculate W.F of lower levels by using mathematical appropriateness and identification material take off. In this method, unlike two previous methods, valuation is performed from high levels to lower levels. For instance, in the above example, if the W.F of foundation is 68% and W.F of wall is 32%, the calculation would be as Table 2. At first, the price of each subcategory will be determined and then, their value will be calculated in according to branches (higher level; note column 2 of Table). In next stage, numbers obtained will be multiplied to the percentage of W.V in order to get management applications. This method has been used in mechanical installation process unit of Morvarid petrochemical company. Table 3. Calculating W.F Based on agreement between owner and contractor
5 Time Management There are different methods for analysis and also estimating the required time of a project. One of these methods is to identify the required activities for a project and set the WBS. At first, time of each activity must be estimated and finally total time of the project will be identified. Prioritization of activities and the dependencies between them must specify and this information will be recorded in the project schedule. Dependencies between activities can effect on the duration of project; as access to resources (resource constraints) also influence the duration of the project. The time is a constraint that isn’t a cost nor a resource, so can’t the project manager control its
Planning and Project Control with EXCEL Software in Asalouyeh 5th Olefin Project
261
changes. This property, make time distinct from other resources. It should be noted that cost reducing is the most important effort with time project control can prevent losing most of sourcing. It should be considered that some very profitable projects such as petrochemical Projects have $1/500/000 daily gains averagely by investment $200 billion. Because of high investment return rate, projects by controlling and completing on the time can return consumer spending. And delay will be cause to lose of profit. Therefore, in order to control the time, should be compare actual progress with plan progress. 5.1 Calculating Plan Progress in According to W.F Planning and project control software calculates plan progress easily by tracking, but should be know if don’t assign resources or WF calculating progress isn’t accuracy, those results wouldn’t be reliable, because activates value calculate base on time during to accomplish them. Calculating plan progress by using WF with Primavera and MSP software is somewhere difficult. These calculations will be done easily by using Excel. Calculating base is in the following. Suppose: G1 is a date that you want to calculate the progress up to this time; (Cut of Date) M13 is start time of desired activity; (Start Activity) N13 is finish time of desired activity; (Finish Activity) P is a plan progress; 1) If XB then P=%100 3) If A