Hybrid algorithms to customize and optimize diversity and accuracy of recommendations

Preprint

24 March 2009

preprint
Published by arXiv in arXiv

Abstract

Recommender systems use data on past user preferences to predict possible future likes and interests. A key challenge is how to obtain accurate recommendations while avoiding suggestions that are too obvious or insufficiently personalized. In this paper we take four recommendation algorithms and test them on three distinct datasets for accuracy and diversity of their results. Based on this we show how two of the methods can be combined into a hybrid that enables simultaneous gains in both accuracy and diversity of recommendations.

Keywords

RECOMMENDATIONS RECOMMENDER
ACCURACY AND DIVERSITY
ALGORITHMS AND TEST
OPTIMIZE DIVERSITY
HYBRID ALGORITHMS

All Related Versions

Version 1, 2008-08-20, ArXiv
Version 2, 2009-03-24, ArXiv
Version 3, 2010-03-12, ArXiv
Published version: Proceedings of the National Academy of Sciences, 107 (10), 4511.

This publication has 0 references indexed in Scilit: