Method of user modeling based on overlapping communities detection

With the development of social networks, it is important to building the user model which can reflects users´ interests. However traditional methods of user modeling aren´t well detect users´ interests. We review the traditional user modeling method, then presented an idea, it´s that the interest of user in social networks is overlapping of multi-community´ interest. Based on this idea, overlapping communities detection algorithm (OCD) is introduced into user modeling, a user modeling method based on overlapping communities detection algorithm is presented, and used it in personal recommendation. Experimental results show that this method can effectively reduce the prediction errors of collaborative filtering algorithms based on the user.