Communities detection and the analysis of their dynamics in collaborative networks

Abstract

The analysis of graphs like collaborative networks aims at studying the relationships between individuals, instead of individual attributes or properties. One of the interesting substructures of such a graph is a community structure, which is a subset of nodes that are more densely linked when compared with the rest of the network. Such dense subgraphs gather individuals who share similar interests depending on the type of relation encoded in the graph. In this paper we tackle the problem of identifying communities in dynamic networks. We propose an approach based on the random walk to identify communities in evolving graphs like collaborative networks. We apply this approach to the Infocom co-authorship network to determine stable collaborations and evolving communities. We use such information, combined with other Digital Bibliography & Library Project (DBLP) co-authorship network topology features, to analyse the formation of the programme committee board of a conference.