Efficient retrieval of recommendations in a matrix factorization framework
Author(s) -
Noam Koenigstein,
Parikshit Ram,
Yuval Shavitt
Publication year - 2012
Publication title -
citeseer x (the pennsylvania state university)
Language(s) - English
Resource type - Conference proceedings
DOI - 10.1145/2396761.2396831
Subject(s) - speedup , computer science , collaborative filtering , dot product , search engine indexing , cluster analysis , matrix decomposition , metric (unit) , simple (philosophy) , set (abstract data type) , recommender system , algorithm , theoretical computer science , mathematics , artificial intelligence , machine learning , parallel computing , philosophy , eigenvalues and eigenvectors , physics , geometry , operations management , epistemology , quantum mechanics , economics , programming language
Low-rank Matrix Factorization (MF) methods provide one of the simplest and most effective approaches to collaborative filtering. This paper is the first to investigate the problem of efficient retrieval of recommendations in a MF framework. We reduce the retrieval in a MF model to an apparently simple task of finding the maximum dot-product for the user vector over the set of item vectors. However, to the best of our knowledge the problem of efficiently finding the maximum dot-product in the general case has never been studied. To this end, we propose two techniques for efficient search -- (i) We index the item vectors in a binary spatial-partitioning metric tree and use a simple branch and-bound algorithm with a novel bounding scheme to efficiently obtain exact solutions. (ii) We use spherical clustering to index the users on the basis of their preferences and pre-compute recommendations only for the representative user of each cluster to obtain extremely efficient approximate solutions. We obtain a theoretical error bound which determines the quality of any approximate result and use it to control the approximation. Both these simple techniques are fairly independent of each other and hence are easily combined to further improve recommendation retrieval efficiency. We evaluate our algorithms on real-world collaborative-filtering datasets, demonstrating more than ×7 speedup (with respect to the naive linear search) for the exact solution and over ×250 speedup for approximate solutions by combining both techniques.
Accelerating Research
Robert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom
Address
John Eccles HouseRobert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom