Structure and Content based Community Detection in Evolving Social Networks

Research has focused on exploring communities in evolving social networks where rapid changes occur continuously. Most community detection approaches rely only on network structure ignoring its content and thus valuable information. We propose a community detection algorithm that exploits both content and structure so as to form more thematically cohesive communities. In particular, we extend a fast incremental structural community detection algorithm in order to take into consideration the content of the network at each step and detect evolving, overlapping communities. We compare the proposed algorithm against the original one and show that it discovers more cohesive communities through time.