User Profiling based on Folksonomy Information in Web 2.0 for Personalized Recommender Systems Huizhi (Elly) Liang Supervisors: Yue Xu, Yuefeng Li, Richi.

Slides:



Advertisements
Similar presentations
Recommender System A Brief Survey.
Advertisements

Recommender Systems & Collaborative Filtering
Google News Personalization: Scalable Online Collaborative Filtering
Location-Based Social Networks Yu Zheng and Xing Xie Microsoft Research Asia Chapter 8 and 9 of the book Computing with Spatial Trajectories.
Music Recommendation by Unified Hypergraph: Music Recommendation by Unified Hypergraph: Combining Social Media Information and Music Content Jiajun Bu,
Prediction Modeling for Personalization & Recommender Systems Bamshad Mobasher DePaul University Bamshad Mobasher DePaul University.
Suleyman Cetintas 1, Monica Rogati 2, Luo Si 1, Yi Fang 1 Identifying Similar People in Professional Social Networks with Discriminative Probabilistic.
A Graph-based Recommender System Zan Huang, Wingyan Chung, Thian-Huat Ong, Hsinchun Chen Artificial Intelligence Lab The University of Arizona 07/15/2002.
Creating Collaborative Partnerships
Content Management & Hashtag Recommendation IN P2P OSN By Keerthi Nelaturu.
Collaborative Filtering Sue Yeon Syn September 21, 2005.
COMP423 Intelligent Agents. Recommender systems Two approaches – Collaborative Filtering Based on feedback from other users who have rated a similar set.
Query Dependent Pseudo-Relevance Feedback based on Wikipedia SIGIR ‘09 Advisor: Dr. Koh Jia-Ling Speaker: Lin, Yi-Jhen Date: 2010/01/24 1.
Learning to Recommend Hao Ma Supervisors: Prof. Irwin King and Prof. Michael R. Lyu Dept. of Computer Science & Engineering The Chinese University of Hong.
Explorations in Tag Suggestion and Query Expansion Jian Wang and Brian D. Davison Lehigh University, USA SSM 2008 (Workshop on Search in Social Media)
The Social Web: A laboratory for studying s ocial networks, tagging and beyond Kristina Lerman USC Information Sciences Institute.
Personalized Ontologies for Web Search and Caching Susan Gauch Information and Telecommunications Technology Center Electrical Engineering and Computer.
Overview of Web Data Mining and Applications Part I
Nonnegative Shared Subspace Learning and Its Application to Social Media Retrieval Presenter: Andy Lim.
Creating Collaborative Partnerships CHAPTER 15 Copyright © 2013 by The McGraw-Hill Companies, Inc. All rights reserved. McGraw-Hill/Irwin.
Business Driven Technology Unit 4
Item-based Collaborative Filtering Recommendation Algorithms
Introduction The large amount of traffic nowadays in Internet comes from social video streams. Internet Service Providers can significantly enhance local.
Personalization in Local Search Personalization of Content Ranking in the Context of Local Search Philip O’Brien, Xiao Luo, Tony Abou-Assaleh, Weizheng.
MediaEval Workshop 2011 Pisa, Italy 1-2 September 2011.
An Integrated Approach to Extracting Ontological Structures from Folksonomies Huairen Lin, Joseph Davis, Ying Zhou ESWC 2009 Hyewon Lim October 9 th, 2009.
Learning to Classify Short and Sparse Text & Web with Hidden Topics from Large- scale Data Collections Xuan-Hieu PhanLe-Minh NguyenSusumu Horiguchi GSIS,
1 Formal Models for Expert Finding on DBLP Bibliography Data Presented by: Hongbo Deng Co-worked with: Irwin King and Michael R. Lyu Department of Computer.
A Hybrid Recommender System: User Profiling from Keywords and Ratings Ana Stanescu, Swapnil Nagar, Doina Caragea 2013 IEEE/WIC/ACM International Conferences.
A Simple Unsupervised Query Categorizer for Web Search Engines Prashant Ullegaddi and Vasudeva Varma Search and Information Extraction Lab Language Technologies.
Improving Web Search Ranking by Incorporating User Behavior Information Eugene Agichtein Eric Brill Susan Dumais Microsoft Research.
RecSys 2011 Review Qi Zhao Outline Overview Sessions – Algorithms – Recommenders and the Social Web – Multi-dimensional Recommendation, Context-
Xiaoying Gao Computer Science Victoria University of Wellington Intelligent Agents COMP 423.
PAUL ALEXANDRU CHIRITA STEFANIA COSTACHE SIEGFRIED HANDSCHUH WOLFGANG NEJDL 1* L3S RESEARCH CENTER 2* NATIONAL UNIVERSITY OF IRELAND PROCEEDINGS OF THE.
UOS 1 Ontology Based Personalized Search Zhang Tao The University of Seoul.
ON INCENTIVE-BASED TAGGING Xuan S. Yang, Reynold Cheng, Luyi Mo, Ben Kao, David W. Cheung {xyang2, ckcheng, lymo, kao, The University.
GAUSSIAN PROCESS FACTORIZATION MACHINES FOR CONTEXT-AWARE RECOMMENDATIONS Trung V. Nguyen, Alexandros Karatzoglou, Linas Baltrunas SIGIR 2014 Presentation:
Chengjie Sun,Lei Lin, Yuan Chen, Bingquan Liu Harbin Institute of Technology School of Computer Science and Technology 1 19/11/ :09 PM.
Wang-Chien Lee i Pervasive Data Access ( i PDA) Group Pennsylvania State University Mining Social Network Big Data Intelligent.
Presented By :Ayesha Khan. Content Introduction Everyday Examples of Collaborative Filtering Traditional Collaborative Filtering Socially Collaborative.
Developing Trust Networks based on User Tagging Information for Recommendation Making Touhid Bhuiyan et al. WISE May 2012 SNU IDB Lab. Hyunwoo Kim.
Automatic Image Annotation by Using Concept-Sensitive Salient Objects for Image Content Representation Jianping Fan, Yuli Gao, Hangzai Luo, Guangyou Xu.
Introduction to Digital Libraries hussein suleman uct cs honours 2003.
Collaborative Information Retrieval - Collaborative Filtering systems - Recommender systems - Information Filtering Why do we need CIR? - IR system augmentation.
Semantics-Based News Recommendation with SF-IDF+ International Conference on Web Intelligence, Mining, and Semantics (WIMS 2013) June 13, 2013 Marnix Moerland.
Evaluation of Recommender Systems Joonseok Lee Georgia Institute of Technology 2011/04/12 1.
WEB 2.0 PATTERNS Carolina Marin. Content  Introduction  The Participation-Collaboration Pattern  The Collaborative Tagging Pattern.
Harvesting Social Knowledge from Folksonomies Harris Wu, Mohammad Zubair, Kurt Maly, Harvesting social knowledge from folksonomies, Proceedings of the.
Advanced Semantics and Search Beyond Tag Clouds and Taxonomies Tom Reamy Chief Knowledge Architect KAPS Group Knowledge Architecture Professional Services.
Pairwise Preference Regression for Cold-start Recommendation Speaker: Yuanshuai Sun
Mining Dependency Relations for Query Expansion in Passage Retrieval Renxu Sun, Chai-Huat Ong, Tat-Seng Chua National University of Singapore SIGIR2006.
Automating Readers’ Advisory to Make Book Recommendations for K-12 Readers by Alicia Wood.
Collaborative Filtering via Euclidean Embedding M. Khoshneshin and W. Street Proc. of ACM RecSys, pp , 2010.
Identifying “Best Bet” Web Search Results by Mining Past User Behavior Author: Eugene Agichtein, Zijian Zheng (Microsoft Research) Source: KDD2006 Reporter:
Social Information Processing March 26-28, 2008 AAAI Spring Symposium Stanford University
The Wisdom of the Few Xavier Amatrian, Neal Lathis, Josep M. Pujol SIGIR’09 Advisor: Jia Ling, Koh Speaker: Yu Cheng, Hsieh.
Collaborative Deep Learning for Recommender Systems
GUILLOU Frederic. Outline Introduction Motivations The basic recommendation system First phase : semantic similarities Second phase : communities Application.
1 Dongheng Sun 04/26/2011 Learning with Matrix Factorizations By Nathan Srebro.
COMP423 Intelligent Agents. Recommender systems Two approaches – Collaborative Filtering Based on feedback from other users who have rated a similar set.
StressSense: Detecting Stress in Unconstrained Acoustic Environments using Smartphones Hong Lu, Mashfiqui Rabbi, Gokul T. Chittaranjan, Denise Frauendorfer,
Neural Collaborative Filtering
Recommendation in Scholarly Big Data
Recommender Systems & Collaborative Filtering
Collective Network Linkage across Heterogeneous Social Platforms
Location Recommendation — for Out-of-Town Users in Location-Based Social Network Yina Meng.
Postdoc, School of Information, University of Arizona
Movie Recommendation System
Recommendation Systems
Bug Localization with Combination of Deep Learning and Information Retrieval A. N. Lam et al. International Conference on Program Comprehension 2017.
Presentation transcript:

User Profiling based on Folksonomy Information in Web 2.0 for Personalized Recommender Systems Huizhi (Elly) Liang Supervisors: Yue Xu, Yuefeng Li, Richi Nayak Queensland University of Technology, Australia

Agenda 4 Introduction The Proposed Approaches Experiments Conclusion Literature Review

1 Introduction

Information overload  Personalization “Personalization is the ability providing content and services tailored to individuals based on knowledge about their preferences and behaviours” (Hagen, 1999)  Recommender systems  User profiling  Explicit user profiles Explicit ratings  Implicit user profiling Web log Other information sources

Web 2.0  Web 2.0: Read and Write web (O’Reilly Media, 2004)  A platform for users to conduct online participation, collaboration and interaction.  Expressing opinions, sharing information, building networks  Wikipedia, Facebook, Delicious, Tweeter  Plenty of new user information  Folksonomy (Tags), reviews, networks, blogs, micro-blogs etc.  Opportunities  Providing possible new solutions to profile users

Folksonomy  Folksonomy= folk + taxonomy  Tags: Typical Web 2.0 information  Keywords given by users to organize and classify items  The wisdom of crowds  Multiple functions Item organizing and sharing Building networks Expressing users’ explicit topic interests and opinions

Tag Cloud

Folksonomy Tags Taxonomy categories  Taxonomy  Given by experts  Standard vocabulary & Structural relationship  Well recognized as common knowledge  Independent with user communities  No users’ personal viewpoints or preferences information  Taxonomy  Given by experts  Standard vocabulary & Structural relationship  Well recognized as common knowledge  Independent with user communities  No users’ personal viewpoints or preferences information  Folksonomy  Given by users explicitly and proactively  Reflecting users’ personal viewpoints and topic preferences  Less intrusive & Multiple function  Lightweight textural information  Contains a lot of noise  Folksonomy  Given by users explicitly and proactively  Reflecting users’ personal viewpoints and topic preferences  Less intrusive & Multiple function  Lightweight textural information  Contains a lot of noise

Literature Review 2

User Profiling  Web User profiling  Web content & structure  Web log & Web usage  Taxonomy & Ontology  User Profiling in Web 2.0  New user information sources Folksonomy, blogs, reviews, micro-blogs Videos, audios, images Friends, trust network, followers, following

User Profiling 2  User Profiling based on folksonomy  Approaches Users’ own tags Associated tags Latent topics of tags Popular tags  Challenges Distinctive features of tags Tag quality problem Semantic ambiguity and synonyms About 60% of tags are personal tags

Recommender system  Recommendation tasks  Top N Recommendation (Precision, Recall, F1)  Rating Prediction (Mean Absolute Error, Root Mean Squared Error)  Recommendation approaches  Content based Term vector model Latent Dirichlet Allocation (LDA)  Collaborative Filtering (CF) Memory based CF: User-KNN & Item-KNN Model based CF: Matrix Factorization techniques  Hybrid

Recommender system 2  Recommender systems based on Taxonomy  Ziegler’s approach (CIKM, 2004)  Recommender systems based on Folksonomy  Tag recommendations Tensor based approach (KDD, 2009) Graph based approach (SIGIR, 2009)  Item recommendations Tso-Sutter’s approach(SAC, 2008) Clustering (RecSys, 2009) LDA approach (HT, 2009) Graph Rank (2010) Special tag rating function(WWW,2009)

Research Problem  Research Gap  Features of folksonomy  Noise of folksonomy  Combining with taxonomy  Research Problem  Profiling users based on folksonomy information in Web 2.0 and enhance recommender systems

The Proposed Approaches 3

 User Profiling Models  User Profiling based on Folksonomy  User Profiling based on Taxonomy  Hybrid User Profiling  Recommender System  Top N item recommendation Recommendation making The Proposed Approaches User Profiling User Profiling-Folksonomy User Profiling-Taxonomy User Profiling-Hybrid

The Relationship Modelling  The Multiple relationships of tagging  Two dimensional relationships User-Item relationship User-Tag relationship Item-Tag relationship  Three dimensional relationship Personal tagging behavior User-Tag-Item relationship (User×Tag)-Item mapping Item-(User×Tag) mapping

 Part 1: User Profiling Approaches based on Folksonomy  Tag representation-Folksonomy  Item representation-Folksonomy  User representation-Folksonomy Tag Representation- Folksonomy Item Representation- Folksonomy User Representation- Folksonomy User Profiling-Folksonomy

Tag representation-Folksonomy  Reduce the noise of tags  Find the personally related tags of each tag  Determine the relevance weight  Relevance weight of two tags with respect to a user  The collected items of a tag  The expectation of the probability of a tag being used for the collected items “apple” “garden” “globalization” “apple” “internet” Number of users used the tag for the item Number of users collected the item

Item representation-Folksonomy  Expand the tags of each item  Find the relevant tags of each item  Determine the relevance weight  The relevance of an item to a tag  User-tag pairs  The relevance of two tags with respect to a user  Inverse item frequency “garden” “apple” “globalization” “internet” “0403”

User Representation-Folksonomy  Find users’ preferences to tags  The preference weight of a user to a tag  Preferences to one tag  The relevance of two tags with respect to a user  Inverse user frequency “garden” “apple” “globalization” “internet” “0403” Number of items collected with the tag by the user Number of items collected by the user

 User  Item preferences Implicit ratings  Topic preferences Tag vocabulary  Item  Tag vocabulary User Profiling-Folksonomy “garden” “apple” “globalization” “internet” “0403” “garden” “apple” “globalization” “internet” “0403”

 Part 2: User Profiling based on Taxonomy  Advantages of Taxonomy Standard vocabulary Well recognized Independent with user communities Experts’ viewpoints  Representations Item representation-Taxonomy Tag representation-Taxonomy User representation-Taxonomy “apple” Tag Representation- Taxonomy Item Representation- Taxonomy User Representation- Taxonomy User Profiling-Taxonomy

 Find the relevant taxonomic topics of each item  The relevance of an item to a taxonomic topic  The average weight of a taxonomic topic in all descriptors The weight of a taxonomic topic in an item descriptor Deploy weight from leaf topic to root topic  Inverse item frequency Item Representation-Taxonomy “programming” “book” “computers” “networks”

 Reduce the noise of tags  Find the personal semantic meaning of each tag  The relevance of a tag to a taxonomic topic with respect to a user  The collected items of a tag  Average relevance weight of a taxonomic topic to the collected items Tag Representation-Taxonomy “computers” “programming” “databases” “networks” “apple” “garden” “flowers” “fruit” “apple”

 Find users’ preferences to taxonomic topics  The preference weight of a user to a taxonomic topic  Preference to a tag  Relevance of a tag to a taxonomic topic with respect to the user  Inverse user frequency User Representation-Taxonomy “databases” “programming” “computers” “book” “0403”

 User  Item preferences Implicit ratings  Topic preferences Taxonomy vocabulary  Item  Taxonomy vocabulary User Profiling-Taxonomy “databases” “programming” “computers” “book” “computers” “programming” “networks”

 Part 3: Hybrid User Profiling  Combine Part 1 and Part 2  Wisdom of crowds Tag vocabulary & Users’ viewpoints  Wisdom of experts Taxonomy vocabulary & Experts’ viewpoints Tag representation-Hybrid Item representation-HybridUser representation-Hybrid

 Personalized Recommendation Making  Top N item recommendation Neighborhood Formation Recommendation Generation User Profiling-Folksonomy User Profiling-Taxonomy User Profiling-Hybrid User Profiling Recommendation Making

Neighbourhood Formation  K-Nearest Neighbourhood  User-KNN Similarity of item preferences Similarity of topic preference Tags Taxonomic topics Linear combination Item Preferences Topic Preferences Tags Taxonomic topics User Similarity

Neighbourhood Formation 2  K-Nearest Neighbourhood  Item-KNN Similarity of Tags Similarity of Taxonomic topics Linear combination TagsTaxonomic TopicsItem similarity

Recommendation Generation  Candidate items  Neighbour items & Not tagged by the target user  User based recommendation  Item based recommendation Prediction ScoreUser Similarity Content matching Tags Taxonomic Topics Prediction Score Item Similarity

Experiments 4

Datasets  D1: Amazon.com  4112 users, tags, items, 9919 taxonomic topics  D2: CiteULike “Who-posted-what” dataset  7103 users, tags, items  Power Law Distributions Tags Items

Experiment setup  Top N item recommendation  Experiment setup 5-folded 80% training & 20% testing  Evaluation Metrics Precision, Recall, F1 Measure  Comparisons  Proposed Models Folksonomy Model: FM-User, FM-Item Taxonomy Model: TM-User, TM-Item Hybrid Model: FTM-User, FTM-Item  Baseline Models

 Tag Noise Removing Approaches (Dataset D1)  Parameter setting  FM-User: : ,  1 :  FM-Item:  1 : Results-I Folksonomy Model

 The Comparison of the State-of-the-art approaches (Dataset D1) Results-I

 Comparison results of Dataset D2 Results-I

 Parameter setting (Dataset D1)  TM-User:  : ,  1 :  TM-Item:  1 : Results-2 Taxonomy Model

 Parameter setting (Dataset D1)  FTM-User:  FTM-Item:  1 =0.3,  Hybrid Models v.s. Single Models  Folksonomy Model v.s. Taxonomy Model Results-3 Hybrid Models

Results-3  The influence of personal tags  D1 personal tags: 67%,   10: 4.8%  D2 personal tags: 70%,   10: 5.2%  Findings  Personal tags can improve the precision results  Precision values decreased dramatically when large number (i.e., 90%) of tags (i.e.,  5) was removed. TM-User, D1 (9919, 0.24)

Discussions  The proposed approaches outperformed other related work  The Hybrid Model performed the best  Each tag counts  Folksonomy can be used as quality information source (rich personalization information)

Conclusions 5

 Web 2.0  New user information  Modelling the relationships of tagging behaviour  Tag quality problem  The wisdom of crowds & experts  Proposed three user profiling models  User profiling based on folksonomy  User profiling based on taxonomy  Hybrid user profiling  Utilized the proposed user profiles to improve recommender systems  User based  Item based  Evaluation Experiments

Contributions  Advantages  Domain free  Language free  Information overload  User profiling and web personalization  Recommender systems  Web 2.0

Future Work  Time factor  Cross folksonomy recommendations  Mobile platform application  Integrate with other user information  Explicit ratings  Tweets  Friendship network

Published Work  Liang, H. et al. (2010). Personalized Recommender System Based on Item Taxonomy and Folksonomy. CIKM  Liang, H. et al. (2010). Connecting Users and Items with Weighted Tags for Personalized Item Recommendations. Hypertext  Liang, H. et al. (2010). A Hybrid Recommender System based on Weighted Tags. SDM Workshop  Liang, H. et al. (2010). Mining Users’ Opinions based on Item Folksonomy and Taxonomy for Personalized Recommender Systems. ICDM Workshop  Liang, H. et al. (2010). Parallel User profiling based on folksonomy for Large Scaled Recommender Systems- An implementation of Cascading MapReduce. ICDM Workshop  Liang, H. et al. (2009). Collaborative Filtering Recommender Systems based on Popular Tags. ADCS  Liang, H. et al. (2009). Tag Based Collaborative Filtering for Recommender Systems. RSKT  Liang, H. et al. (2009). Personalized Recommender Systems Integrating Social tags and Item Taxonomy. WI  Liang, H. et al. (2008). Collaborative Filtering Recommender Systems Using Tag Information. WI Workshop  Bhuiyan, T., Xu, Y., Jøsang, A., & Liang, H. (2010). Developing Trust Networks Based on User Tagging Information for Recommendation Making. WISE

Acknowledgements Time Supervisor Team HPC group Penal Members ISS Anonymous Reviewers Papers Staffs Colleagues Friends Google Books Sunshine CSC Trees Stars Music Trips Blogs Beaches Family …

Questions & Answers