CDPM: Finding and Evaluating Community Structure in Social Networks

In this paper we proposed a CDPM (Clique Directed Percolation Method) algorithm, which clusters tightly cohesive cliques as cluster atoms and merge the cluster atoms into communities under the direction of a proposed object function, namely Structure Silhouette Coefficient (SSC). SSC could measure the quality of community divisions which allows communities share actors. Experiments demonstrate our algorithm can divide social networks into communities at a higher quality than compared algorithms.