Hybrid Recommender System for Joining Virtual Communities

The variety of social networks and virtual communities has created problematic for users of different ages and preferences; in addition, since the true nature of groups is not clearly outlined, users are uncertain about joining various virtual groups and usually face the trouble of joining the undesired ones. As a solution, in this study, we introduced the hybrid community recommender system which offers customized recommendations based on user preferences. Although techniques such as content based filtering and collaborative filtering methods are available, these techniques are not enough efficient and in some cases make problems and bring limitations to users. Our method is based on a combination of content based filtering and collaborative filtering methods. It is created by selecting related features of users based on supervised entropy as well as using association rules and classification method. Supposing users in each community or group share similar characteristics, by hierarchical clustering, heterogeneous members are identified and removed. Unlike other methods, this is also applicable for users who have just joined the social network where they do not have any connections or group memberships. In such situations, this method could still offer recommendations.