Label propagation algorithm for community detection based on Coulomb’s law

Abstract

Community detection is an important field in social network analysis; it provides a higher level of structure and greater understanding of the network. In this paper, we develop an improved label propagation algorithm for solving the community detection problem based on Coulomb’s Law abbreviated as LPA_CL. Coulomb’s Law in analogy with particles in the physics domain captures the importance of a node within a network by computing a proximity index using the geodesic distance between nodes as punishment. In other words, the adopted proximity index between two nodes of a network integrates both the local and the global structural information of a given network. Moreover, experimental results on real-world networks and artificial networks indicate that the proposed algorithm (LPA_CL) is efficient and effective to be used for community detection of medium and large networks. It has better accuracy and stability and converges LPA in shorter iteration.