Detecting hierarchical and overlapping community structures in networks

Abstract

Community structure can be observed in many natural, biological and social networks. Studies suggest that these communities may have organized in a hierarchical manner while some communities overlap with others. This paper introduces an algorithm to detect such hierarchical and overlapping community structures in networks based on the concept of maximal cliques. It introduces an alternate modularity for evaluating overlapping community structures. Unlike existing algorithms for detecting hierarchical and overlapping community structures, the new algorithm is free of parameter tuning and random seeds. Experiments conducted on two real-world networks show that this algorithm is capable of providing satisfactory and consistent results.