Group Drgs ch Recommender systems Evolution of Collaborative filtering Recommender interfaces John riedl For the grouplens research group University of minnesota riedl@cs. umn. edu http://www.cs.umn.edu/research/grouplens Yahoo, Summer 2004
Yahoo, Summer 2004 1 Recommender Systems: Evolution of Collaborative Filtering Recommender Interfaces John Riedl For the GroupLens Research Group University of Minnesota {riedl}@cs.umn.edu http://www.cs.umn.edu/Research/GroupLens
Rescarch Challenges in Recommenders Recommending for Groups(Cosley CsCW 02) Community Recommenders ludford chI 04) Confidence and Explanation(Herlocker CSCW 00)(MCNee Interact 03) Social Navigation(Hook02, 05 Reducing rating effort New Items and Users McNee IUI 02) Shilling lam WWw04) Small Device recommenders( Miller Tois 04) Yahoo, Summer 2004 2
Yahoo, Summer 2004 2 Challenges in Recommenders Recommending for Groups (Cosley CSCW ’02) Community Recommenders (Ludford CHI ’04) Confidence and Explanation (Herlocker CSCW ‘00) (McNee Interact ’03) Social Navigation (Hook ’02, ‘05) Reducing Rating Effort New Items and Users (McNee IUI ’02) Shilling (Lam WWW ’04) Small Device Recommenders (Miller TOIS ’04)
Rescarch Challenges in Recommenders Recommending for Groups(Cosley CsCW 02) Community Recommenders ludford chI 04) Confidence and Explanation(Herlocker CSCW 00)(MCNee Interact03) Social Navigation(Hook02, 05 Reducing rating effort New Items and Users (MCNee IUI 02) Shilling (Lam WWW04) Small Device recommenders( Miller Tois 04) Yahoo, Summer 2004
Yahoo, Summer 2004 3 Challenges in Recommenders Recommending for Groups (Cosley CSCW ’02) Community Recommenders (Ludford CHI ’04) Confidence and Explanation (Herlocker CSCW ‘00) (McNee Interact ’03) Social Navigation (Hook ’02, ‘05) Reducing Rating Effort New Items and Users (McNee IUI ’02) Shilling (Lam WWW ’04) Small Device Recommenders (Miller TOIS ’04)
Group Drgs rch Messages Don't Make Me look stupid Dont be boring Measure what matters Yahoo, Summer 2004
Yahoo, Summer 2004 4 Messages “Don’t Make Me Look Stupid” “Don’t Be Boring” Measure what Matters
Grofb lgescarch Recommenders Tools to help identify worthwhile stuff Filtering interfaces E-mail filters, clipping services Recommendation interfaces Suggestion lists, top-n, offers and promotions Prediction interfaces Evaluate candidates, predicted ratings Yahoo, Summer 2004 5
Yahoo, Summer 2004 5 Recommenders Tools to help identify worthwhile stuff Filtering interfaces E-mail filters, clipping services Recommendation interfaces Suggestion lists, “top-n,” offers and promotions Prediction interfaces Evaluate candidates, predicted ratings
S Rescarch C01e0 Recommenders Purely Editorial Recommenders Content Filtering recommenders Collaborative Filtering recommenders Hybrid recommenders Yahoo, Summer 2004 6
Yahoo, Summer 2004 6 Scope of Recommenders Purely Editorial Recommenders Content Filtering Recommenders Collaborative Filtering Recommenders Hybrid Recommenders
Gro rcscarch Wide Range of Algorithms Simple Keyword vector Matches Pure Nearest-Neighbor Collaborative Filtering Machine learning on content or ratings Yahoo, Summer 2004 7
Yahoo, Summer 2004 7 Wide Range of Algorithms Simple Keyword Vector Matches Pure Nearest-Neighbor Collaborative Filtering Machine Learning on Content or Ratings
Collaborative Rescarch Filtering Premise Information needs more complex than keywords or topics: quality and taste Small community: Manual Tapestry -database of content &z comments Active CF-easy mechanisms for forwarding content to relevant readers Yahoo, Summer 2004
Yahoo, Summer 2004 8 Collaborative Filtering Premise Information needs more complex than keywords or topics: quality and taste Small Community: Manual Tapestry – database of content & comments Active CF – easy mechanisms for forwarding content to relevant readers
Group Drgs rch Automated ce The Grouplens Project( CsCW94 ACF for usenet News users rate items users are correlated with other users personal predictions for unrated items Nearest-Neighbor Approach find people with history of agreement assume stable tastes Yahoo, Summer 2004
Yahoo, Summer 2004 9 Automated CF The GroupLens Project (CSCW ’94) ACF for Usenet News users rate items users are correlated with other users personal predictions for unrated items Nearest-Neighbor Approach find people with history of agreement assume stable tastes
Grou dowdy 特拼排#排#排 SUec 排#排 Operations apply to current selection or cursor position Quit Next unread Nex ][o[ripe[ Next grou甲 From:"Art Poe"<apoeeuni com. net) Creme Brule Categories Desserts Amount Measure Ingredient -- Preparation Method Pinch lespoons Sugar Tablespoons Brown sugar This article is great, I'd like to see more like this one! Save Reply Forward Followup Followup g Reply Cancel Translate ogle header Print郾 atel artRate2artRate3 artRateq artRat Yahoo, Summer 2004 10
Yahoo, Summer 2004 10 Usenet Interface