Overlapping community detection with a maximal clique enumeration method in MapReduce

Abstract

Overlapping community detection is progressively becoming an important issue in social network analysis (SNA). Faced with massive amounts of information while simultaneously restricted by hardware specifications and computation time limits, it is difficult for clustering analysis to reflect the latest developments or changes in complex networks. To meet these demands, this research proposes a novel distributed computation method, which combines MapReduce, a distributed computation framework, and the TTT algorithm, to speed up the discovery of all maximal cliques in large-scale social networks. Then, overlapping community detection is implemented by the Clique Percolation Method (CPM) to incrementally merge adjacent cliques based on k-cliques with k-1 common nodes. Six groups of YouTube datasets (from 50K to 300K nodes with interval 50K) are adopted to evaluate clustering quality and execution time of the proposed method.