Social Event Recommendation Based on Social Relationship and Attention Mechanism

Event recommendation based on social relationship is a common method in EBSNs. This kind of algorithm can solve the problem of cold start and data sparsity to a certain extent, but it doesn‘’t consider the influence degree of diffrent friends and the influence degree of diffrent groups on events. In this paper, a novel method for social event recommendation based on social relationship and self-attention (AtSoRec) was developed. The algorithm gets the trust weights among different friends and users by the attention model training. In the process of training, the information of groups and events is integrated into the module. Then, the final Top-k recommendation is obtained by Matrix factorization algorithm. Through the experiments on meetup and plancast data set, it is proved that the algorithm has a good performance on solving the problem of cold start.