Detecting communities via edge Random Walk Centrality

Abstract

Herein we present a novel approach of identifying community structures in complex networks. We propose the usage of the Random Walk Centrality (RWC), first introduced by Noh and Rieger [Phys. Rev. Lett. 92.11 (2004): 118701]. We adapt this node centrality metric to an edge centrality metric by applying it to the line graph of a given network. A crucial feature of our algorithm is the needlessness of recalculating the centrality metric after each step, in contrast to most community detection algorithms. We test our algorithm on a wide variety of standard networks, and compare them with pre-existing algorithms. As a predictive application, we analyze the Indian Railway network for robustness and connectedness, and propose edges which would make the system even sturdier.