Empirical Performace Determination on Community Detection Techniques in Social Networks

Abstract

Community identification is the high common and extending field of interest in social and real-time network applications. In recent years, many community detection methods have been developed. This paper describes various community discovery methods such as InfoMap, Clique Guided, Louvain, Newman and Eigen Vector that have already been developed and also compares the experimental results of those proposed techniques. The proposed work in this paper experiments these community mining algorithms on the two real-world datasets Twitter and DBLP (Computer Science Bibliography) networks. The identified communities by all the community mining algorithms for these two data sets are described in this proposed work. The quality of the derived communities is evaluated by using standard Extended Modularity metric. The experiment results show that the InfoMap algorithm produces a good modularity score than other community mining algorithms for different sizes of communities on both data sets.