Centrality properties of directed module members in social networks

TL;DR

This paper explores the centrality of nodes in directed modules within social networks, revealing that overlapping nodes are central and that different network types exhibit distinct centrality-role relationships.

Contribution

It extends the Clique Percolation Method to directed graphs and analyzes the centrality of overlapping module members in social networks.

Findings

Overlapping nodes are central in social network modules.

Different network types show distinct centrality-role relationships.

Directed modules reveal unique structural properties.

Abstract

Several recent studies of complex networks have suggested algorithms for locating network communities, also called modules or clusters, which are mostly defined as groups of nodes with dense internal connections. Along with the rapid development of these clustering techniques, the ability of revealing overlaps between communities has become very important as well. An efficient search technique for locating overlapping modules is provided by the Clique Percolation Method (CPM) and its extension to directed graphs, the CPMd algorithm. Here we investigate the centrality properties of directed module members in social networks obtained from e-mail exchanges and from sociometric questionnaires. Our results indicate that nodes in the overlaps between modules play a central role in the studied systems. Furthermore, the two different types of networks show interesting differences in the relation between the centrality measures and the role of the nodes in the directed modules.