Leiden Fitness-Based Genetic Algorithm with Niching for Community Detection in Large Social Networks

Community Detection (CD) in large social networks is a highly active research area due to its immense practical value in many real-world applications. Genetic algorithms (GAs) are widely used to solve the CD problem due to their strong ability to explore the global discrete search space. However, existing GA-based algorithms focus more on the effectiveness of the solution rather than the capability of handling large social networks scalably. In this paper, we propose Leiden Fitness-based GA (LeFGA) to tackle the scalability issue, allowing GA to effectively and efficiently process large social networks. This is achieved specifically by using the newly developed individual and the fitness evaluation method. LeFGA further adopts a niching method to maintain its population diversity. Experimental results prove that LeFGA can significantly outperform multiple state-of-the-art algorithms, especially on large real-world social networks.