Using local utility maximization to detect social networks communities

Community detection has recently turned to be one of the most popular research topics in social networks analysis. Majority of community detection methods already considered in the literature try to optimize a global metric through a centralized decision maker. These approaches are too time-consuming in huge networks. Several of methods need initial parameters such as number and size of communities in order to find out the problems; however, they are not always reachable. In this paper, we propose a local utility maximization approach for community identification as a distributed framework in which each community acts as a selfish agent to maximize its utility function based on some predefined actions. Our framework has some crucial characteristic features. The first feature is the local approach that is easily implemented through parallel computing concepts, while the second is an economical interpretation of utility measurement. Experimental results on output benchmark datasets show that our proposed method can perform as well as the existing centralized approaches that already exist in the literature to detect non-overlapping communities.