200 likes | 421 Views
Efficient Retrieval of Recommendations in a Matrix Factorization Framework. Motivation. In the field of Recommender System , Matrix Factorization (MF) models have shown superior accuracy for recommendation tasks. E.g., The Netflix Prize, KDD-Cup’11, etc.
E N D
Efficient Retrieval of Recommendationsin a Matrix Factorization Framework
Motivation • In the field of Recommender System, Matrix Factorization (MF) models have shown superior accuracy for recommendation tasks.E.g., The Netflix Prize, KDD-Cup’11, etc. • Training is fast. Computing test scores is fast.But… Retrieval of Recommendations (RoR) is s--l--o--w ! • This problem is well known in the industry, yet never been approached before in academia!
I T E M S Yahoo! Music: 1M Users 625K Items 6 Tera elements ~300 multiplications ~5 days CPU Naïve Multithreading:High latency + wasteful Yahoo! Music: 1M Users 625K Items 6 Tera elements ~300 multiplications ~5 days CPU U S E R S
Reduction to Inner Product Core problem: Given a user vector and a set of item, find an item vector that will maximize
Best Matches Algorithms • Metric Space • Cosine Similarity • Locality Sensitive Hashing
Metric Trees R R
Approximate Solution Users vectors can be normalized Users can be clustered based on their spherical angle!
Relative Error Bound What is the error when recommendations are retrieved based on an approximate user vector?
Experimentations Set-up Yahoo! Music Recommendations: Modeling Music Ratings with Temporal Dynamics and Item Taxonomy Gideon Dror, Noam Koenigstein, Yehuda Koren (RecSys-11`)
Conclusions • We introduce a new and relevant research problem • An exact solution with limited speedup • An approximate solution with a tradeoff between accuracy and speedup • Much room for further research…