Deep Group Recommender System Model Based on User Trust

Today’s group recommender systems rarely consider the trust relationship between users, but the interaction between users often has a significant impact on each other’s preferences. A group recommendation model based on deep learning and user interaction is proposed, which has stronger representational capability than previous models. In addition, an improved weight fusion method based on user implicit vector is applied to the preference fusion of group members. Epinions data set shows that this model is superior to other comparison algorithms in RMSE and hit ratio.