An Efficient Community Detection Method Based on Path Proximity and Local Edge Betweenness Centrality

Abstract

Centrality is a measure index evaluating node and edge importance, and edge betweenness centrality is the most important among those. The length of shortest path between two nodes shows their relationship. When the path is longer, their relationship is weaker. Based on this idea, this paper think that two nodes have no relationship beyond some range. This paper introduces local edge betweenness centrality within K steps and weight of path length, and puts forward a new centrality evaluation method. Based on this, our method utilizes the core idea of GN algorithm, and puts forward a new community detection algorithm called Dist-K. Its performance and result are compared with other existing methods, as applied to different well-known instances of complex networks. Our method has better modularity, normalized mutual information, adjusted rand index and accuracy, which are widely used to measure community structures.