A Multi-View Group Recommender System based on Trust and Ratings

Sometimes, depending on the type of system, it is not possible to offer a list of items for each user individually and the number of items in the recommended lists should be limited, therefore group recommender systems will be used. So far, various group recommender systems have been presented. Most of them have been investigated from the standpoint of user preferences (rating matrix). In this paper, the proposed Multi-View Group Recommender System (MVGRS), investigates from two points of view, i.e., user preferences and social connections (trust). This system, with multi-view, first offers the individual recommendations, then generates the group recommendation by aggregating individual recommendations. The system has been tested with the different numbers of rating clusters and trust clusters. Finally, the error of the MVGRS is compared with the error of the similar single-view group recommender system. The results show that the MVGRS outperforms the single-view group recommender system.