Recognizing number of communities and detecting community structures in complex networks

TL;DR

This paper presents a new approach combining block models, an optimization algorithm, and a fuzzy boundary method to accurately recognize the number of communities and detect community structures in complex networks.

Contribution

It introduces a novel fuzzy boundary algorithm and an optimization process for block models to improve community detection accuracy in complex networks.

Findings

The proposed algorithm effectively recognizes the number of communities.

Experimental results validate the feasibility of the method.

The approach outperforms traditional community detection techniques.

Abstract

Recognizing number of communities and detecting community structures of complex network are discussed in this paper. As a visual and feasible algorithm, block model has been successfully applied to detect community structures in complex network. In order to measure the quality of the block model, we first define an objective function WQ value. For obtaining block model B of a network, GSA algorithm is applied to optimize WQ with the help of random keys. After executing processes AO (Adding Ones) and RO (Removing Ones) on block model B, the number of communities of a network can be recognized distinctly. Furthermore, based on the advantage of block model that its sort order of nodes is in correspondence with sort order of communities, so a new fuzzy boundary algorithm for detecting community structures is proposed and successfully applied to some representative networks. Finally, experimental results demonstrate the feasibility of the proposed algorithm.