Community detection is a very important task in social network analysis. Most existing community detection algorithms are designed for single-relational networks. However, in the real world, social networks are mostly multi-relational. In this paper, we propose a coalition formation game theory-based approach to detect communities in multi-relational social networks. We define the multi-relational communities as the shared communities over multiple single-relational graphs, and model community detection as a coalition formation game process in which actors in a social network are modeled as rational players trying to improve group’s utilities by cooperating with other players to form coalitions. Each player is allowed to join multiple coalitions and coalitions with fewer players can merge into a larger coalition as long as the merge operation could improve the utilities of coalitions merged. We then use a
to identify communities. Experimental results and performance studies verify the effectiveness of our approach.