WWW.DISSERTATION.XLIBX.INFO
FREE ELECTRONIC LIBRARY - Dissertations, online materials
 
<< HOME
CONTACTS



Pages:     | 1 || 3 | 4 |   ...   | 19 |

«Learning Implicit User Interest Hierarchy for Web Personalization by Hyoung-rae Kim A dissertation submitted to Florida Institute of Technology in ...»

-- [ Page 2 ] --

7.3. Machine Learning ----------------------------------------------------------------------- 152 7.3.1. Symbolic Methods of Learning----------------------------------------------- 154 7.3.1.1. Semantic Networks -------------------------------------------------------- 156 7.3.1.2. Learning Decision Trees-------------------------------------------------- 156 7.3.1.3. Learning Sets of Rules ---------------------------------------------------- 156 7.3.2. Numerical Methods of Learning---------------------------------------------- 157 7.3.2.1. Hidden Markov Models--------------------------------------------------- 157 7.3.2.2. Naïve Bayes Classifier ---------------------------------------------------- 158 7.3.2.3. Artificial Neural Networks ----------------------------------------------- 158 7.3.2.4. Instance-based Learning -------------------------------------------------- 159 7.3.3. Clustering Techniques --------------------------------------------------------- 159 7.3.4. Correlation Functions ---------------------------------------------------------- 164

8. Conclusions ------------------------------------------------------------------------------------ 166

8.1. Summary of Contributions ------------------------------------------------------------- 167

8.2. Ethical Issues in User Modeling------------------------------------------------------- 172 8.2.1. Privacy --------------------------------------------------------------------------- 172 8.2.2. Confidence on the Results----------------------------------------------------- 173

8.3. Limitation and Future Work ----------------------------------------------------------- 173 References------------------------------------------------------------------------------------------- 175 Appendix -------------------------------------------------------------------------------------------- 188

ix List of Figures

Figure 1. Five research parts in the framework of web personalization ----------------------- 6 Figure 2.

Learning user interest hierarchy in the framework of web personalization-------13 Figure 3. Sample user interest hierarchy ----------------------------------------------------------15 Figure 4. DHC algorithm----------------------------------------------------------------------------19 Figure 5. An example of DHC algorithm (1)-----------------------------------------------------21 Figure 6. An example of DHC algorithm (2)-----------------------------------------------------22 Figure 7. Shown in a Histogram -------------------------------------------------------------------27 Figure 8. Find the widest and deepest valley -----------------------------------------------------27 Figure 9. UIH with words ---------------------------------------------------------------------------38 Figure 10. UIH with words and phrases-----------------------------------------------------------38 Figure 11. Finding phrases in the framework of web personalization ------------------------41 Figure 12. VPF algorithm ---------------------------------------------------------------------------45 Figure 13. Example of VPF-------------------------------------------------------------------------46 Figure 14. Desirable properties in the framework of web personalization -------------------62 Figure 15. Venn diagram ----------------------------------------------------------------------------63 Figure 16. Contingency matrix ---------------------------------------------------------------------69 Figure 17. Contingency matrix of desirable properties over all functions--------------------72 Figure 18. Contingency matrix of desirable properties over property 1 ----------------------73 Figure 19. Contingency matrix of desirable properties over property 6 ----------------------74 Figure 20. Devising a scoring function in the framework of web personalization ----------79 Figure 21. Diagram of Scoring ---------------------------------------------------------------------80 Figure 22. Average and SD of precision with Google ------------------------------------------96 Figure 23. Precision/recall graph for interesting web pages------------------------------------96 Figure 24. Precision/recall with personal score weight c ---------------------------------------97 Figure 25. Up to 20% recall of Figure 24 ---------------------------------------------------------97 Figure 26. Average and SD of precision with Google ---------------------------------------- 100 Figure 27. Precision/recall graph for potentially interesting web pages -------------------- 100 x Figure 28. Precision/recall with personal score weight c ------------------------------------- 101 Figure 29. Up to 20% recall of Figure 28 ------------------------------------------------------- 101 Figure 30. Implicit indicator for interesting web pages --------------------------------------- 106 Figure 31. A retrieved image --------------------------------------------------------------------- 115 Figure 32. Detected three dots-------------------------------------------------------------------- 116 Figure 33. Three dots in an image --------------------------------------------------------------- 116 Figure 34. Diagram of web information retrieval---------------------------------------------- 131 Figure 35. Diagram of user modeling ----------------------------------------------------------- 147 Figure 36. Diagram of machine learning-------------------------------------------------------- 153 Figure 37. Diagram of classification------------------------------------------------------------- 155 Figure 38. Diagram of clustering----------------------------------------------------------------- 160





xi List of Tables

Table 1. Sample data set-----------------------------------------------------------------------------15 Table 2.

AEMI values -------------------------------------------------------------------------------24 Table 3. AEMI-SP values ---------------------------------------------------------------------------25 Table 4. Distribution of frequency and number of children ------------------------------------27 Table 5. Combination of AEMI, MaxChildren, and entire page-------------------------------36 Table 6. Combination of AEMI-SP, MaxChildren, and entire page --------------------------36 Table 7. Combination of AEMI, Valley, and entire page---------------------------------------36 Table 8. Combination of AEMI, MaxChildren, and 100 words -------------------------------36 Table 9. Use words and phrases --------------------------------------------------------------------38 Table 10. With-pruning vs. without-pruning -----------------------------------------------------52 Table 11. Ranked by average across humans and articles – Exact match --------------------56 Table 12. Exact match across humans-------------------------------------------------------------56 Table 13. Ranked by average across humans and articles – Simple match ------------------58 Table 14. Simple match across humans -----------------------------------------------------------58 Table 15. All possible cases (with the same incremental ratio)--------------------------------65 Table 16. An example for property 4 --------------------------------------------------------------66 Table 17. An example for property 5 --------------------------------------------------------------67 Table 18. Summary of correlation functions' properties ----------------------------------------70 Table 19. Properties of correlation functions -----------------------------------------------------77 Table 20. Precision in Top 1, 5, 10, 15 and 20 for interesting web pages --------------------96 Table 21. Precision in Top 1, 5, 10, 15 and 20 for potentially interesting web pages ---- 100 Table 22. ANOVA test with “visits with maximum duration” data set--------------------- 124 Table 23. ANOVA test with the data set of “all visits” --------------------------------------- 124 Table 24. Results of bookmark, save, print, memo indicators ------------------------------- 127

xii Acknowledgments

I would like to express my sincerest thanks to Dr. Philip K. Chan, my dissertation advisor, for his encouragement, patience, financial support and greatest guidance through this dissertation, including co-authorship of several papers, which were published as a result of this research. Mere words cannot express my profound appreciation for his endless love and support. I extend thanks to my other committee members, Dr. Debasis Mitra, Dr. Marius-Calin Silaghi, and Dr. Alan C. Leonard.

Appreciation is also acknowledged for the following people who helped me during

my dissertation and Ph.D. program:

• Financial support: Dr. Philip K. Chan, Dr. Shirley A. Becker, Dr. Debasis Mitra, Dr. William D. Shoaff;

• Experimental help: Matthew Scripter, Stan Salvador, Matthew Mahoney, Dahee Jung, Timothy, Gaurav Tandon, Rachna Vargiya, Mohammad Arshad, Amanda, Audra, Matt, Turky Alotaiby, Mohsen AlSharif, Akiki, Michel, Ayanna, Jae-gon Park, Ji-hoon, Jun-on, Chris Tanner, Grant Beems;

• Support & Prayer: Young-ki Kim, Matthew Scripter, Nattawut Sridranop, Jaehyeon Lee, Ji-won Kim, Sun-young Kweon, Seong-won Kim, Gaehlan, Ron, Dana, Kirk, Simon, Paster Luther V. Laite, Paster Warren E. Baker, Dr. Peggy Douglas, Prof. Harry Alston, Se-hoon Kweon, Eun-jeong Lee, Seong-jin Park, Young-chun Bae, Su-bong Ham, Seong-hoon Park, Ali Al-Badi, Marvin Scripter, Faith Scripter, Hun Namgung, Shin-suk Kim, In-suk Gang, Paster Hyoung-woo Park, Paster Hee-youn Lee, etc.;

• Special thanks: Dr. Do-hong Cheon, Dr. Lieberio, Patti Laite.

xiii 저의 논문을 위해 희생하신 아버지와 어머니의 노고에 비하면 저는 아무것도 한 것이 없는 것처럼 느껴집니다. 아버지는 새벽기도를 나가시면서 나무를 심으시면서 매일 기도의 마음을 잊지 않으셨습니다. 어머니는 좋아하시는 미역국도 삼가시면서 온 마음의 정성을 다 하셨습니다. 또한 공부는 길고 지루한 과정이기 때문에 습관화가 되지 않으면 그리고 주변의 격려가 없으면 끝까지 견디기가 어렵다고 생각합니다. 항상 연구하는 자세를 보여주시는 아버지의 생활 습관 그리고 항상 적극적으로 믿고 지원해 주신 어머니의 격려가 아니었다면 이 논문을 마치기 힘들었다고 믿습니다. 제 아버님의 성함은 김항남, 어머님은 권성자십니다. 저의 형제들 또한 마치 바위같이 흔들림없이 묵묵히 참으면서 뒤에서 도와준 것에 감사를 드립니다. 형제분들은 김미영, 김혜정, 김정래, 확장된 형제로는 장성수, 스퇴킹어 도미닉, 정래영입니다. 제 작은 아버님들과 어머님들, 그리고 외삼촌분들과 외숙모님들께도 감사 드립니다. 또한 조카들 장문주, 장혁주, 스퇴킹어 막스에게 앞으로 여유있는 외삼촌이 되어 주고 싶습니다. 나이로 인해 이젠 몸의 여러부분이 불편하신 가운데에서도 저의 박사논문을 염려해 주신 외할아버지에게도 감사드립니다. 마지막으로, 연구가 막힐 때 마다 기도를 통해 아이디어를 하나님으로부터 제공받았기에, 한편으론 하나님과 생산적인 대화를 즐겼었다고 봅니다.

–  –  –

Web personalization adapts the information or services provided by a web site to the needs of a user. Web personalization is used mainly in four categories: predicting web navigation, assisting personalization information, personalizing content, and personalizing search results. Predicting web navigation anticipates future requests or provides guidance to client. If a web browser or web server can correctly anticipate the next page that will be visited, the latency of the next request will be greatly reduced (Eirinaki et al., 2004; Kim et al., 2004; Shahabi and Banaei-Kashani, 2003; Cadez et al., 2000). Assisting personalization information helps a user organize his or her own information and increases the usability of the web (Maarek and Ben-Shaul, 1996; Li et al., 1999). Personalizing content focuses on personalizing individual pages, site-sessions (e.g., adding shortcut), or entire browsing sessions (Anderson, 2002). Personalized web search results provide customized results depending on each user’s interests (Jeh and Widom, 2003; Haveliwala, 2002; Liu et al., 2003; Bharat and Mihaila, 2001). Information access through a search engine has become an essential part of our daily lives. We use a search engine to find various information from a cloth to technical references. However, the accuracy of search engines is still as low as 55% (Delaney, 2004). In this work, we focus on personalizing web search by ordering search engine results based on the interests of each individual user, which can greatly aid the search through massive amounts of data on the Internet.

1.1. Motivation When a user browses the web at different times, s/he could be accessing pages that pertain to different topics. For example, a user might be looking for research papers at one time and airfare information for conference travel at another. That is, a user can exhibit different kinds of interests at different times, which provides different contexts underlying a user's behavior. However, different kinds of interests might be motivated by the same kind of interest at a higher abstraction level (computer science research, for example). That is, a user might possess interests at different abstraction levels — the higher-level interests are more general, while the lower-level ones are more specific. During a browsing session, general interests are in the back of one's mind, while specific interests are the current foci.

We believe identifying the appropriate context underlying a user's behavior is important in more accurately pinpointing her/his interests. Unlike News Dude (Billsus and Pazzani, 1999), which generates a long-term and a short-term model of interests, we propose to model a continuum of general to specific interests (web browsing interests of a user). The model provides concept hierarchical clusters called a user interest hierarchy (UIH), while suffix tree clustering (STC) algorithm (Zamir and Etzioni, 1998) provides flat clusters.

We can improve the UIH by using phrases in addition to words. A composed term by two or more single words (called “phrase”) usually has more specific meaning and can disambiguate related words. Statistical phrase-finding approaches have been used for expanding vector dimensions in clustering multiple documents (Turpin and Moffat, 1999;



Pages:     | 1 || 3 | 4 |   ...   | 19 |


Similar works:

«Partial melting and melt segregation within the contact aureole of the Sudbury Igneous Complex (Ontario, Canada) and their significance in hydrothermal Cu-Ni-PGE mineralization of the footwall ATTILA PÉNTEK M.Sc. Geology Dissertation submitted to the Ph.D. program for Geology and Geophysics at the Ph.D. school of Earth Sciences, Eötvös Loránd University, Budapest in partial fulfilment of the requirements for the degree of DOCTOR OF PHILOSOPHY Supervisors: FERENC MOLNÁR, Ph.D. associate...»

«Temporal Signature Modeling and Analysis by Jiangqin Sun B.S. Soochow University, 2007 M.S. Harbin Institute of Technology, 2009 A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Imaging Science Chester F. Carlson Center for Imaging Science College of Science Rochester Institute of Technology December 6, 2014 Signature of the Author Accepted by Coordinator, Ph.D. Degree Program Date CHESTER F. CARLSON CENTER FOR IMAGING SCIENCE...»

«HOW TO TEACH SO THAT PEOPLE LEARN POSTCARDS FROM CORINTH • CHAPTER EXCERPT Postcards is the Users Guide for personal discipleship. While The Compass provides content for one-on-one discipleship, Postcards tackles the practice, philosophy, and difficult challenges that are a part of discipleship. Here, for example, is a partial list of the topics covered: Habitual Sin, Authority Issues, Christian Counseling, Fasting, Theological Conflicts, Coaching Through Trials, Challenging to Conferences,...»

«Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 901 Query Processing for Peer Mediator Databases BY TIMOUR KATCHAOUNOV ACTA UNIVERSITATIS UPSALIENSIS UPPSALA 2003 ˚ Dissertation at Uppsala University to be publicly examined in Siegbahnsalen, Angstr¨ m Labo oratory, Tuesday, November 11, 2003 at 13:00 for the Degree of Doctor of Philosophy. The examination will be conducted in English Abstract Katchaounov, T. 2003. Query Processing for Peer Mediator...»

«University of Calgary From the Steppe to the Stable: Horses and Horsemanship in the Ancient World by Carolyn Willekes A DISSERTATION SUBMITTED TO THE FACULTY OF GRADUATE STUDIES IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY DEPARTMENT OF GREEK AND ROMAN STUDIES CALGARY, ALBERTA APRIL, 2013 © Carolyn Willekes 2013 UNIVERSITY OF CALGARY FACULTY OF GRADUATE STUDIES The undersigned certify that they have read, and recommend to the Faculty of Graduate Studies for...»

«Development, Experimentation, and Implementation of a Gap Damper System to Limit Displacements in Extreme Ground Motions by Taylor A. Rawlinson A dissertation submitted to the Graduate Faculty of Auburn University in partial fulfillment of the requirements for the Degree of Doctor of Philosophy Auburn, Alabama May 10, 2015 Keywords: Base-isolation, Earthquake Engineering, Energy Dissipation, Seismic Pounding Copyright 2015 by Taylor A. Rawlinson Approved by Justin D. Marshall, Chair, Associate...»

«MULTIOBJECTIVE DESIGN AND INNOVIZATION OF ROBUST STORMWATER MANAGEMENT PLANS A Dissertation Presented by Karim Jeffrey Chichakly to The Faculty of the Graduate College of The University of Vermont In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy Specializing in Computer Science May, 2013 Accepted by the Faculty of the Graduate College, The University of Vermont, in partial fulfillment of the requirements for the degree of Doctor of Philosophy, specializing in...»

«ABSTRACT Title of dissertation: HOW L2 LEGAL WRITERS USE STRATEGIES FOR SCHOLARLY WRITING: A MIXED METHODS STUDY Donna Bain Butler, Doctor of Philosophy, 2010 Dissertation directed by: Professor Rebecca L. Oxford Department of Curriculum and Instruction Second Language Education and Culture This dissertation research fills existing gaps regarding the practices and processes of teaching second language (L2) writers at higher ranges of proficiency in law school context. It is a mixed methods,...»

«School Travel Mode Choice Behaviour in Toronto, Canada by Raktim Mitra A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Program in Planning, Department of Geography University of Toronto © Copyright by Raktim Mitra 2011 School Travel Mode Choice Behaviour in Toronto, Canada Raktim Mitra Doctor of Philosophy Program in Planning, Department of Geography University of Toronto Abstract Interest in school transportation has emerged in response to concern...»

«APRES MOI LE DELUGE: INDIVIDUALS, INCENTIVES, AND CONFLICT TERMINATION A Dissertation submitted to the Faculty of the Graduate School of Arts and Sciences of Georgetown University in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Government By John Furman Daniel, III, B.A. Washington, D.C. August 31, 2010 Copyright 2010 by John Furman Daniel, III All Rights Reserved ii APRES MOI LE DELUGE: INDIVIDUALS, INCENTIVES, AND CONFLICT TERMINATION John Furman Daniel,...»

«AN EXPERIMENTAL TECHNIQUE FOR DETERMINING CYCLERESOLVED PRE-COMBUSTION IN-CYLINDER COMPOSITION AND ITS APPLICATION TOWARDS THE UNDERSTANDING OF DIESEL ENGINE EMISSIONS DURING TRANSIENT OPERATION by Jonathan Ray Hagena A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Mechanical Engineering) in The University of Michigan 2008 Doctoral Committee: Professor Dionissios N. Assanis, Co-Chair Research Associate Professor Zoran S. Filipi,...»

«THE EVANGELICAL LUTHERAN CHURCH IN NAMIBIA (ELCIN) AND POVERTY, WITH SPECIFIC REFERENCE TO SEMI-URBAN COMMUNITIES IN NORTHERN NAMIBIA A PRACTICAL THEOLOGICAL EVALUATION by Gideon Niitenge Dissertation Presented for the Degree of Doctor of Philosophy in PRACTICAL THEOLOGY (COMMUNITY DEVELOPMENT) at the University of Stellenbosch Promoter: Prof Karel Thomas August March 2013 i Stellenbosch University http://scholar.sun.ac.za DECLARATION I, the undersigned, hereby declare that the work contained...»





 
<<  HOME   |    CONTACTS
2016 www.dissertation.xlibx.info - Dissertations, online materials

Materials of this site are available for review, all rights belong to their respective owners.
If you do not agree with the fact that your material is placed on this site, please, email us, we will within 1-2 business days delete him.