Community detection through topic modeling in social networks

Abstract

The research on communities in social networks takes many paths in the literature, among which: the problematic of accurately detecting communities; modeling the evolution of those communities within the evolving network; and then finding the patterns that characterize this evolution over time. In our work, we focused on the problematic of detecting communities in social networks based on the information disseminated among users of the social network and the type of content shared by these users. The work at hand consists of a brief introduction to the subject and the problem definition, then we move to state the main contribution of our work which consists of a multi-layer model to detect communities of users based on the content shared by users, the lowest layer would detect topics of interest of each user while the upper layer would form communities from generated topics. We conclude the paper stating our perspectives and future works.