Modelling of Overlapping by Community Detection Algorithms in Social Networks: A Review

A social network consists of some people who are related to each other through some similarities. The emergence and evolution of these networks and increasing rate of using them is the major cause for social network analysis to be a hot research topic. Using various algorithms, each network can be divided into some communities. So, each community includes some members of the social network. Community detection is one of the most important and fundamental tasks in network analysis. It is a step towards understanding the patterns and characteristics of the complex systems they represent. In this paper, the state of the art algorithms for community detection are categorized into six categories (spectral clustering and centrality, quality function, Label propagation, Structure, Closeness, link clustering) based on their definition of the community and modelling the concept of overlapping (existence of the nodes with membership in multiple communities). Next, these methods are implemented on various datasets and compared to each other. It is obvious from the results of performance measures, even in this small collection of data sets, no algorithm can be considered as the best community detection method for all kinds of networks.