Community Detection Algorithm based on Multi-objective Biogeographical Optimization with Community Belongingness

Abstract

In community detection algorithms based on optimization, although single-objective optimization methods have been widely applied, they only consider a single objective function and ignore other factors related to community structure, making it difficult to effectively reveal complex networks with multiple potential characteristics. In contrast, multi-objective optimization methods can overcome these limitations and have received increasing attention in the field of community detection in recent years. For the problem of community detection in complex networks, this paper proposes a multi-objective biogeographical optimization-based community detection algorithm (MOBBO-CMD) based on community membership. Firstly, by combining non-dominated sorting and crowding distance, the biogeographical optimization (BBO) mechanism is improved, and a BBO model suitable for multi-objective optimization is established. Simultaneously, the proportion of internal connections within the community (NRA) and the proportion of external connections outside the community (RC) are selected as two optimization objectives, modeling the community detection problem as a multi-objective optimization problem to alleviate the limitation of modularity resolution. Secondly, in the migration operation, a one-way migration method based on community membership is proposed to avoid disrupting the existing relatively correct community division. Furthermore, utilizing node division information, a mutation strategy based on boundary nodes is designed to ensure that mutations can produce better solutions. On real network datasets, MOBBO-CMD is experimentally evaluated compared with four contrasting algorithms. The results show that compared to traditional single-objective community detection algorithms, MOBBO-CMD using multi-objective optimization can evaluate community division results more comprehensively and accurately. Compared to other multi-objective contrasting algorithms, MOBBO-CMD improves community detection accuracy and finds communities closer to the true network structure.