Differential betweenness in complex networks clustering

Abstract

We propose a novel metric for measuring the degree of edge centrality in complex networks clustering, a task commonly called community detection in the analysis of social, biological and information networks. The metric, which has been called differential betweenness, has some unexpected and interesting properties that might help us to create better clustering algorithms. We compare our measure with the shortest path edge betweenness of Girvan and Newman and found that it can be more accurate and robust without requiring the costly recalculation step the other measure needs. © 2008 Springer-Verlag Berlin Heidelberg.