An Effective Approach to Group Recommendation Based on Belief Propagation

Abstract

Recommender systems have been an active research topic for the past decade. Previous studies have primarily focused on recommendations to a single user. Recently, several interesting approaches to make group recommendations have been proposed. However, the accuracy of existing approaches is significantly affected by the size and cohesiveness of a group. In this paper, we present a novel approach that makes effective group recommendations regardless of the group size or cohesiveness. We first model the relationships between a set of users and a set of items as a bipartite graph from the ratings information. On this graph, we employ the belief propagation to determine probabilistically the target user group’s preference on items. We also propose a new group type that reflects real-life groups effectively and helps better evaluation of group recommendation approaches. Through extensive experiments on a real-life data set, we show that the proposed approach is more accurate than the existing ones up to 20%.