Transcript
Page 1: Recommender system   introduction

Recommender SystemIntroduction

[email protected]

Page 2: Recommender system   introduction

What is good recommender system?

Page 3: Recommender system   introduction

Outline

• What is recommender system?– Mission– History– Problems

• What is good recommender system?– Experiment Methods– Evaluation Metric

Page 4: Recommender system   introduction

Information Overload

Page 5: Recommender system   introduction

How to solve information overload

• Catalog– Yahoo, DMOZ

• Search Engine– Google, Bing

Page 6: Recommender system   introduction

Mission

• Help user find item of their interest.• Help item provider deliver their item to

right user.• Help website improve user engagement.

Page 7: Recommender system   introduction

RecommenderSystem

Page 8: Recommender system   introduction

Search Engine vs. Recommender System

• User will try search engine if– they have specific needs– they can use keywords to describe needs

• User will try recommender system if– they do not know what they want now– they can not use keywords to describe needs

Page 9: Recommender system   introduction

History: Before 1992

• Content Filtering – An architecture for large scale information

systems [1985] (Gifford, D.K)– MAFIA: An active mail-filter agent for an

intelligent document processing support [1990] (Lutz, E.)

– A rule-based message filtering system [1988] (Pollock, S. )

Page 10: Recommender system   introduction

History: 1992-1998

• Tapestry by Xerox Palo Alto [1992]– First system designed by collaborative filtering

• Grouplens [1994]– First recommender system using rating data

• Movielens [1997]– First movie recommender system– Provide well-known dataset for researchers

Page 11: Recommender system   introduction

History: 1992-1998

• Fab : content-based collaborative recommendation– First unified recommender system

• Empirical Analysis of Predictive Algorithms for Collaborative Filtering [1998] (John S. Breese)– Systematically evaluate user-based

collaborative filtering

Page 12: Recommender system   introduction

History: 1999-2005

• Amazon proposed item-based collaborative filtering (Patent is filed in 1998 and issued in 2001) [link]

• Thomas Hofmann proposed pLSA [1999] and apply similar method on collaborative filtering [2004]

• Pandora began music genome project [2000]

Page 13: Recommender system   introduction

History: 1999-2005

• Lastfm using Audioscrobbler to generate user taste profile on musics.

• Evaluating collaborative filtering recommender systems [2004] (Jonathan L. Herlocker)

Page 14: Recommender system   introduction

History: 2005-2009

• Toward the Next Generation of Recommender Systems: A Survey of the State-of-the-Art and Possible Extensions. [2005] (Alexander Tuzhilin)

• Netflix Prize [link]– Latent Factor Model (SVD, RSVD, NSVD, SVD++)– Temporal Dynamic Collaborative Filtering– Yehuda Koren [link]’s team get prize

Page 15: Recommender system   introduction

History: 2005-2009

• ACM Conference on Recommender System [2007] (Minneapolis, Minnesota, USA)

• Digg, Youtube try recommender system.

Page 16: Recommender system   introduction

History: 2010-now

• Context-Aware Recommender Systems• Music Recommendation and Discovery• Recommender Systems and the Social Web• Information Heterogeneity and Fusion in

Recommender Systems• Human Decision Making in Recommender Systems• Personalization in Mobile Applications• Novelty and Diversity in Recommender Systems• User-Centric Evaluation

Page 17: Recommender system   introduction

History: 2010-now

• Facebook launches instant personalization [2010]– Clicker– Bing– Trip Advisor– Rotten Tomatoes– Pandora– ……

Page 18: Recommender system   introduction

Problems

• Main Problems– Top-N Recommendation– Rating Prediction

Page 19: Recommender system   introduction

Problems

• Top-N Recommendation– Input

– Output

user item

A a

B a

B b

… …

Page 20: Recommender system   introduction

Problems

• Top-N Recommendation– Input

– Output

user item rating

A a

B a

B b

… … …

?

Page 21: Recommender system   introduction

What is good recommender system?

Page 22: Recommender system   introduction

Experiment Methods

• Offline Experiment• User Survey• Online Experiment– AB Testing

Page 23: Recommender system   introduction

Experiment Methods

• Offline Experiment

DataSet

Train Test

• Advantage:• Only rely on dataset•

• Disadvantage:• Offline metric can not reflect business goal

Page 24: Recommender system   introduction

Experiment Methods

• User Survey– Advantage:• Can get subjective metrics• Lower risk than online testing

– Disadvantage:• Higher cost than offline experiments• Some results may not have statistical significance• Users may have different behaviors under testing

environment or real environment• It’s difficult to design double blink experiments.

Page 25: Recommender system   introduction

Experiment Methods

• On line experiments (AB Testing)– Advantage:• Can get metrics related to business goal

– Disadvantage:• High risk/cost• Need large user set to get statistical significant result

Page 26: Recommender system   introduction

Experiment Metrics

• User Satisfaction• Prediction Accuracy• Coverage• Diversity• Novelty• Serendipity• Trust• Robust• Real-time

Page 27: Recommender system   introduction

Experiment Metrics

• User Satisfaction– Subjective metric– Measured by user survey or online experiments

Page 28: Recommender system   introduction

Experiment Metrics

• Prediction Accuracy– Measured by offline experiments– Top-N Recommendation• Precision / Recall

– Rating Prediction• MAE, RMSE

Page 29: Recommender system   introduction

Experiment Metrics

• Coverage– Measure the ability of recommender system to

recommend long-tail items.

– Entropy, Gini Index

||

|),(|

I

NuRCoverage Uu

Page 30: Recommender system   introduction

Experiment Metrics

• Diversity– Measure the ability of recommender system to

cover users’ different interests.– Different similarity metric generate different

diversity metric.

Page 31: Recommender system   introduction

Experiment Metrics

• Diversity (Example)

Watch History Related Items

Page 32: Recommender system   introduction

Experiment Metrics

• Novelty– Measure the ability of recommender system to

introduce long tail items to users.– International Workshop on Novelty and

Diversity in Recommender Systems [link]– Music Recommendation and Discovery in the

Long Tail [Oscar Celma]

Page 33: Recommender system   introduction

Experiment Metrics

• Serendipity– A recommendation result is serendipity if:• it’s not related with user’s historical interest• it’s novelty to user• user will find it’s interesting after user view it

Page 34: Recommender system   introduction

Experiment Metrics

• Trust– If user trust recommender system, they will

interact with it.– Ways to improve trust:• Transparency• Social• Trust System (Epinion)

Page 35: Recommender system   introduction

Experiment Metrics

• Robust– The ability of recommender system to prevent

attack.– Neil Hurley. Tutorial on Robustness of

Recommender System. ACM RecSys 2011.

Page 36: Recommender system   introduction

Experiment Metrics

• Real-time– Generate new recommendations when user

have new behaviors immediately.

Page 37: Recommender system   introduction

Too many metric!Which is most important?

Page 38: Recommender system   introduction

How to do trade-off

• Business goal• Our belief• Making new algorithms by 3 steps

experiments:– Offline testing– User survey– Online testing

Page 39: Recommender system   introduction

Thanks!


Top Related