A game-theoretic approach for locally detecting overlapping communities in social networks

The study of embedded structure of communities in social and information networks is an extensive studies in this domain and vast variety of community detection methods have been proposed. In this paper we proposed a distributed approach for local and overlapping community detection based on the game theory. In our method, each node is a player and there is an iterative cycle in which players can play their best action from a given set of actions periodically in their turn. Each player decides to become member of a community which has the best influence on it in order to maximize its utility function. According to players' decisions communities will be formed gradually. Therefore, when the game process reaches the Nash equilibrium, the community emerges. We evaluate our method on some common datasets to indicate the performance and sufficiency of it.