Group Recommendations Based on Comprehensive Latent Relationship Discovery
Shanshan Feng, Jian Cao, Jie Wang, Jing He · 2016
In recent years, due to an increasing overload of information on the Internet, there are many scenarios where Recommender Systems (RSs) are employed to provide suggestions to user groups. However, most proposed approaches of group recommendations simply aggregate individual ratings or individual prediction results, rather than comprehensively investigating the hidden correlative information between members and the group, which results in inferior recommendation performance. In this paper, we propose a new approach, RWR-UTM, for group recommendations based on the combination of an integrated probabilistic topic model - a User Topic Model (UTM) and the Random Walk with Restart (RWR) method. The UTM provides a latent framework of users, groups, and items by exploiting both the users' preference profiles and the items' content information, which together can describe group interests and item features in a more complete manner. This latent framework is then combined with RWR to predict the preference degrees of groups to unrated items by detecting comprehensive latent relationships. In particular, we devised two group-based recommendation algorithms on the basis of different recommendation strategies. Finally, we conducted experiments to evaluate our approach and compare it with other state-of-the-art approaches using the real-world CAMRa2011 data-set. The results demonstrate the advantage of our approach over comparative ones.