Improving group recommendations via detecting comprehensive correlative information

Traditionally, recommender systems are applied to recommending items to individual users. However, there has been a proliferation of recommender systems that try to make recommendations to user groups. Although several approaches were proposed to generate group recommendations, they made recommendations simply through aggregating individual ratings or individual predicted results, rather than comprehensively investigating the inherent relationships between members and the group, which can be used to improve the performance of group recommender systems. For this reason, these approaches continue to suffer from data sparsity and do not work well for recommending items to user groups. Therefore, we proposed a new approach for group recommendations based on random walk with restart (RWR) method. The goal of the work in this paper is describing groups’ preferences better by comprehensively detecting the correlative information among users, groups, and items, in order to alleviate the data sparsity problem and improve the performance of group recommender systems. In the proposed approach, we represent the relationships among users, groups, and items as a tripartite graph. Based on the tripartite graph, RWR can predict the relevance degrees between groups and unrated items by comprehensively detecting their relationships. Using these relevance degrees, we can describe a group’s preferences better so as to achieve a more accurate recommendation. In particular, we devised two recommendation algorithms based on different recommendation strategies. Finally, we conducted experiments to evaluate our method and compare it with other state-of-the-art methods using the real-world CAMRa2011 data-set. The results show the advantage of our approach over comparative ones.