This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: 1 CS345 Data Mining Recommendation Systems Netflix Challenge Course Projects Anand Rajaraman, Jeffrey D. Ullman Recommendations Items Search Recommendations Products, web sites, blogs, news items, … From scarcity to abundance ¡ Shelf space is a scarce commodity for traditional retailers ¢ Also: TV networks, movie theaters,… ¡ The web enables near-zero-cost dissemination of information about products ¢ From scarcity to abundance ¡ More choice necessitates better filters ¢ Recommendation engines ¢ How Into Thin Air made Touching the Void a bestseller The Long Tail Source: Chris Anderson (2004) Recommendation Types ¡ Editorial ¡ Simple aggregates ¢ Top 10, Most Popular, Recent Uploads ¡ Tailored to individual users ¢ Amazon, Netflix, … Formal Model ¡ C = set of Customers ¡ S = set of Items ¡ Utility function u : C × S → R ¢ R = set of ratings ¢ R is a totally ordered set ¢ e.g., 0-5 stars, real number in [0,1] 2 Utility Matrix 0.4 1 0.2 0.3 0.5 0.2 1 King Kong King Kong LOTR LOTR Matrix Matrix Nacho Nacho Libre Libre Alice Alice Bob Bob Carol Carol David David Key Problems ¡ Gathering “known” ratings for matrix ¡ Extrapolate unknown ratings from known ratings ¢ Mainly interested in high unknown ratings ¡ Evaluating extrapolation methods Gathering Ratings ¡ Explicit ¢ Ask people to rate items ¢ Doesn’t work well in practice – people can’t be bothered ¡ Implicit ¢ Learn ratings from user actions ¢ e.g., purchase implies high rating ¢ What about low ratings? Extrapolating Utilities ¡ Key problem: matrix U is sparse ¢ most people have not rated most items ¡ Three approaches ¢ Content-based ¢ Collaborative ¢ Hybrid Content-based recommendations ¡ Main idea: recommend items to customer C similar to previous items rated highly by C ¡ Movie recommendations ¢ recommend movies with same actor(s), director, genre, … ¡ Websites, blogs, news ¢ recommend other sites with “similar” content Plan of action likes...
View Full Document
This document was uploaded on 03/04/2012.
- Fall '09
- Data Mining