Merging user social network into the random walk model for better group recommendation

At present, most recommendation approaches used to suggest appreciate items for individual users. However, due to the social nature of human beings, group activities have become an integral part of our daily life, thus the popularity of group recommender systems has increased in the last years. Unfortunately, most existing approaches used in group recommender systems make recommendations through aggregating individual preferences or individual predictive results rather than comprehensively investigating users social features that govern their choices made within a group. Therefore, we propose a new group recommendation approach, it incorporates user social network into the random walk with restart model and variously detects the inherent associations among group members, which can help us to better describe groups preference and improve the performance of group recommender systems. Besides, on the basis of multifaceted associations incorporation, we apply a partitioned matrix computation method in the recommendation process to save computational and storage costs. The final experiment results on the real-world CAMRa2011 dataset demonstrates that the proposed approach can not only effectively predict groups' preference, but also have faster performance and more stable than other baseline methods.