An Adaptive Group Recommender Based On Overlapping Community Detection

The Internet is experiencing a revolution from the interconnection of physical networks by a collection of protocols towards what is considered the future Internet: a network of applications, information and content. People are growing engaging in a more complicated relationship network. They have to identify their interests from a potentially overwhelming set of choicesTo solve these problems, recommendation system is proposed for not only filtrating confusing information but also mining the real needs of users. But with the popularity and maturity of Web2.0, a fact is that people have gradually accepted that decision making becomes even more difficult when a group, instead of an individual consumer, will consume the product, as is regularly the case for hedonic offerings, group recommenders was proposed and thus sparked researchers’great interests and discussion to identify items that appeal to a group as a whole.In this paper, a kind of modified adaptive group recommender based on overlapping community detection is proposed. Different from existing recommenders, this recommender takes both of group members’ preferences and their complex internal interactions into account. In this research, both of overlapping community integration strategy and contribution-based collaborative filtering are employed to explore group members’ interests and provide the predicted group ratings on movies. The results show that the proposed recommender can achieve comparatively accurate prediction with a comparatively low computation complexity.