Research on the integration strategy of group recommendation based on user's interactive behaviors

The traditional recommender system is recommends for individuals, but many activities are generated by groups in real life. It is a hot spot that how to recommend for groups. And for recommending better and improving the accuracy of recommender system, it is important that collects and integrates the group members' preferences. The best recommendation focus on reducing the dissatisfaction of the group members to the recommended items as much as possible. On the basis of the traditional recommender system, this paper constructed a preference integration model for group recommender systems. This model obtained the each members' prediction rating on the items and group's predictions on the items based on the individual and group collaborative filtering algorithm. Additionally, this paper put forward a method that obtain the weights of members through the interactive behaviors, then obtain the final prediction of items through the preference fusion method. Finally, this paper put forward the improved GMAE evaluation model, and the fusion model was verified and evaluated by experiments, it demonstrated that the proposed model perform better than pure collaborative filtering method on accuracy and diversity.