SOCIAL: A Self-Organized Entropy-Based Algorithm for Identifying Communities in Networks

The identification of communities in complex networks is important to many fields including medicine, social science, national security, and marketing. A community structure facilitates the identification of hidden relations in networks that go beyond simple topological features. Current detection algorithms are centralized and scale very poorly with the number of nodes and edges present in the network. The use of these algorithms is prohibitive when applied to large-scale networks. In this paper, we propose a Self-Organized Community Identification Algorithm (SOCIAL) based on local calculations of node entropy that enables individual nodes to independently decide the community they belong to. In our context, node entropy is defined as the individual node's satisfaction with its current community. As nodes become more "satisfied'' (entropy decreases) the community structure of a network emerges. Our algorithm offers several advantages over existing approaches including near-linear performance, identification of community overlaps, and localized management of dynamic changes in the network.