A novel memorizing single chromosome evolutionary algorithm for detecting communities in complex networks

Many real-world systems such as social networks, transportation networks, and the Internet can be modeled as complex networks. An important aspect of such networks is community structures. In fact, community detection can help extract vital information in such networks. Recently, various methods have been proposed for community detection, although there are apparently many unsolved problems which must be addressed. Therefore, this paper proposes a single-chromosome memorizing evolutionary algorithm for detecting calibrated communities. This algorithm benefits from a unique operator called modification topology operator and tries to promote the quality of detected communities by memorizing positions of better solutions. Other contributions of this paper include proposing a novel extended Jaccard index to better measure node similarity and introducing an innovative method for determining the connected components. For this purpose, a special directed graph is developed in this method based on the solution_vector to identify the poorly connected components, which are ignored in order to determine other components. Finally, the detected communities were merged in an agglomerative way to increase community detection correctness. The proposed algorithm was compared with several state-of-the-art methods. According to the experimental results on real networks, the proposed algorithm significantly ranked first among them. There were also considerable improvements on LFR datasets.