Disassortative Degree Mixing and Information Diffusion for Overlapping Community Detection in Social Networks (DMID)

Abstract

In this paper we propose a new two-phase algorithm for overlapping community detection (OCD) in social networks. In the first phase, called disassortative degree mixing, we identify nodes with high degrees through a random walk process on the row-normalized disassortative matrix representation of the network. In the second phase, we calculate how closely each node of the network is bound to the leaders via a cascading process called network coordination game. We implemented the algorithm and four additional ones as a Web service on a federated peer-to-peer infrastructure. Comparative test results for small and big real world networks demonstrated the correct identification of leaders, high precision and good time complexity. The Web service is available as open source software.