Accuracy and Diversity Improvements for Multi-Criteria Recommender Systems

Abstract

Various recommendation algorithms have been developed to improve the accuracy of recommendations; however, the diversity of recommendations has often been overlooked. Intuitively, it may be possible to achieve improvements in one of these two metrics at the expense of the other. For example, higher accuracy may sometimes be obtained by safely recommending to users the most popular items, which can lead to the reduction in aggregate recommendation diversity, i.e., less personalized recommendations. Conversely, higher diversity can be achieved by trying to uncover and recommend highly personalized items for each user, which are inherently more difficult to predict and, thus, may lead to a decrease in recommendation accuracy. To overcome this accuracy-diversity tradeoff, this work builds on the following two ideas: incorporating multi-criteria rating information and applying different ranking methods. Experiments using movie rating data empirically demonstrate simultaneous improvements in accuracy and diversity.