Recommender Systems for Social Tagging Systems

Social Tagging Systems are web applications in which users upload resources (e.g., bookmarks, videos, photos, etc.) and annotate it with a list of freely chosen keywords called tags. This is a grassroots approach to organize a site and help users to find the resources they are interested in. Social...

Full description

Saved in:
Bibliographic Details
Main Authors Marinho, Leandro Balby, Hotho, Andreas, Jäschke, Robert, Nanopoulos, Alexandros, Rendle, Steffen, Sc, Lars
Format eBook Book
LanguageEnglish
Published New York Springer-Verlag 2012
Springer
Springer New York
Springer US
Edition1. Aufl.
SeriesSpringerBriefs in Electrical and Computer Engineering
Subjects
Online AccessGet full text

Cover

Loading…
Table of Contents:
  • Intro -- Recommender Systems for Social Tagging Systems -- Preface -- Contents -- Part I Foundations -- Chapter 1 Social Tagging Systems -- 1.1 Introduction -- 1.2 Folksonomies -- 1.3 Tag Clouds -- 1.4 Data Representation -- 1.4.1 Folksonomies as Tensors -- 1.4.2 Folksonomies as Hypergraphs -- 1.5 Recommendation Tasks in STS -- 1.5.1 User Recommendation -- 1.5.2 Resource Recommendation -- 1.5.3 Tag Recommendation -- 1.6 Recommendations in Social Tagging Systems -- 1.6.1 BibSonomy -- 1.6.2 CiteULike -- 1.6.3 Other Systems -- 1.7 Notation -- 1.8 Further Reading -- References -- Chapter 2 Recommender Systems -- 2.1 Rating and Item Prediction -- 2.2 Rating Prediction as Regression Problem -- 2.3 Item Prediction as Ranking Problem -- 2.4 User and Item Attributes -- 2.5 New User and New Item Problems -- 2.6 Context-aware and Multi-Mode Recommendations -- References -- Part II Recommendation Techniques for Social Tagging Systems -- Chapter 3 Baseline Techniques -- 3.1 Constant Models -- 3.1.1 Tag Recommendation -- 3.1.2 User/Tag-aware Recommendation -- 3.1.3 Remarks on Complexity -- 3.2. Projection Matrices -- 3.3 Projection-based Collaborative Filtering -- 3.3.1 Tag Recommendations -- 3.3.2 Tag-aware Recommendations -- 3.3.3 User Recommendations -- 3.3.4 Remarks on Complexity -- 3.4 Further Reading -- References -- Chapter 4 Advanced Techniques -- 4.1 Factorization Models -- 4.1.1 Higher Order Singular Value Decomposition - HOSVD on Tensors -- 4.1.1.1 From SVD to HOSVD -- 4.1.1.2 HOSVD for Recommendations in STS -- 4.1.1.3 Combining HOSVD with Content-based Methods -- 4.1.1.4 Limitations of HOSVD -- 4.1.2 Scalable Factorization Models -- 4.1.2.1 Tucker Decomposition -- 4.1.2.2 Parallel Factor Analysis (PARAFAC) -- 4.1.2.3 Pairwise Interaction Tensor Factorization (PITF) -- 4.1.2.4 Factorization Machines (FM)
  • 4.1.2.5 Relationship of Factorization Models -- 4.1.3 Learning Tag Recommendation Models -- 4.2 Graph-based Models -- 4.2.1 PageRank-based Recommendations in STS -- 4.2.2 Relational Neighbors for Tag Recommendations -- 4.3 Content and Social-Based Models -- 4.3.1 Exploiting the Content of Resources -- 4.3.2 Exploiting Social Relations -- 4.4 Further Reading -- References -- Chapter 5 Offline Evaluation -- 5.1 Evaluation Metrics -- 5.1.1 Precision and Recall -- 5.1.2 Further Measures -- 5.2 Evaluation Protocols -- 5.2.1 LeavePostOut Methodology -- 5.2.2 Time-based Splits -- 5.3 Comparison of Tag Recommenders -- References -- Part III Implementing Recommender Systems for Social Tagging -- Chapter 6 Real World Social Tagging Recommender Systems -- 6.1 Introduction -- 6.2 Challenges and Requirements -- 6.3 The BibSonomy Social Tagging System -- 6.4 Architecture -- 6.4.1 Overview -- 6.4.2 Recommender Interface -- 6.4.3 Logging -- 6.5 Recommender Implementations -- 6.5.1 Meta Recommender -- 6.5.1.1 First Weighted By Second -- 6.5.1.2 Weighted Merging -- 6.5.1.3 Remote Recommender -- 6.5.2 Multiplexing Tag Recommender -- 6.5.3 Example Recommender Implementations -- 6.5.3.1 Most Popular ρ-Mix (MPρ-mix) -- 6.5.3.2 Title Tags Weighted by User Tags (TbyU) -- 6.5.3.3 Other -- 6.6 Further Reading -- References -- Chapter 7 Online Evaluation -- 7.1 Evaluation Setting -- 7.1.1 Metrics and Protocols -- 7.1.2 Preprocessing and Cleansing -- 7.2 Case Study -- 7.2.1 General Results -- 7.2.2 Influence of the `reload' Button -- 7.2.3 Logged `click' Events -- 7.2.4 Average F1-Measure per User -- 7.3 The ECML PKDD Discovery Challenge 2009 -- 7.3.1 Setting -- 7.3.2 Methods -- 7.3.3 Results -- 7.4 Conclusion -- References -- Chapter 8 Conclusions -- 8.1 Summary -- 8.2 Discussion and Outlook -- References