Identifying overlapping communities and their leading members in social networks

Abstract

With the recent increasing popularity of social networking services like Facebook and Twitter, community structure has become a problem of considerable interest. Although there are more than a hundred algorithms that find communities in networks, only a few are able to detect overlapping communities, and an even smaller number of them follow an approach based on the evolution dynamics of these networks. Thus, we present FRINGE, an algorithm for the detection of overlapping communities in networks, which, based on the ideas of friendship and leadership, not only returns the overlapping communities detected, but also specifies their leading members. We describe the algorithm in detail and compare its results with those obtained by CFinder and iLCD for both synthetic and real-life networks. These results show that our proposal behaves well in networks with a clear social hierarchy, as seen in modern social networks. © 2013 Springer-Verlag.