Collaborative Filtering Recommender Systems
- ISSN: 15513955
- ISBN: 9783540720782
- DOI: 10.1561/1100000009
Abstract
Every day, we are inundated with choices and options. What to wear? What movie to rent? What stock to buy? What blog post to read? The sizes of these decision domains are frequently massive: Netflix has over 17,000 movies in its selection 15, and Amazon.com has over 410,000 titles in its Kindle store alone 7. Supporting discovery in informa- tion spaces of this magnitude is a significant challenge. Even simple decisions what movie should I see this weekend? can be difficult without prior direct knowledge of the candidates. Historically, people have relied on recommendations and mentions from their peers or the advice of experts to support decisions and dis- cover new material. They discuss the weeks blockbuster over the water cooler, they read reviews in the newspapers entertainment section, or they ask a librarian to suggest a book. They may trust their local the- ater manager or news stand to narrow down their choices, or turn on the TV and watch whatever happens to be playing.
Collaborative Filtering Recommender Systems
Human–Computer Interaction
Vol. 4, No. 2 (2010) 81–173
c
© 2011 M. D. Ekstrand, J. T. Riedl and J. A. Konstan
DOI: 10.1561/1100000009
Collaborative Filtering Recommender Systems
By Michael D. Ekstrand, John T. Riedl
and Joseph A. Konstan
Contents
1 Introduction 82
1.1 History of Recommender Systems 84
1.2 Core Concepts, Vocabulary, and Notation 85
1.3 Overview 87
2 Collaborative Filtering Methods 88
2.1 Baseline Predictors 89
2.2 User–User Collaborative Filtering 91
2.3 Item–Item Collaborative Filtering 95
2.4 Dimensionality Reduction 101
2.5 Probabilistic Methods 107
2.6 Hybrid Recommenders 111
2.7 Selecting an Algorithm 112
3 Evaluating Recommender Systems 114
3.1 Data Sets 115
3.2 Offline Evaluation Structure 116
3.3 Prediction Accuracy 117
3.4 Accuracy Over Time 119
3.6 Decision Support Metrics 122
3.7 Online Evaluation 125
4 Building the Data Set 128
4.1 Sources of Preference Data 129
4.2 Rating Scales 132
4.3 Soliciting Ratings 134
4.4 Dealing with Noise 136
5 User Information Needs 138
5.1 User Tasks 139
5.2 Needs for Individual Items 140
5.3 Needs for Sets of Items 141
5.4 Systemic Needs 144
5.5 Summary 149
6 User Experience 150
6.1 Soliciting Ratings 150
6.2 Presenting Recommendations 151
6.3 Recommending in Conversation 153
6.4 Recommending in Social Context 154
6.5 Shaping User Experience with Recommendations 157
7 Conclusion and Resources 159
7.1 Resources 162
References 164
Sign up today - FREE
Mendeley saves you time finding and organizing research. Learn more
- All your research in one place
- Add and import papers easily
- Access it anywhere, anytime