a review of community detection methods in social networks with a focus on the integration of deep learning and the grasshopper algorithm

Community detection in social networks is a fundamental problem in complex network analysis, aimed at identifying cohesive groups of nodes with strong intra-group connections and weak inter-group connections. classical methods, such as spectral clustering and modularity-based algorithms, face challenges including dependence on initial parameters, sensitivity to data noise, and poor scalability in large and dynamic networks. in the past decade, the emergence of deep learning-based methods and their combination with metaheuristic algorithms such as grasshopper, whale, crow, and particle swarm optimization has brought significant improvements in the accuracy and stability of community detection. in this review article, the latest research on community detection using hybrid deep learning and evolutionary optimization models is examined and compared. the findings show that integrating autoencoders and graph neural networks (gnns) with nature-inspired algorithms leads to significant improvements in metrics such as modularity, accuracy, and nmi. furthermore, employing these hybrid frameworks enhances the ability to uncover hidden structures and overlapping communities in complex networks. finally, existing challenges—including scalability, network dynamics, high computational cost, and model interpretability—are discussed, and future directions for developing more efficient, interpretable, and adaptive methods for dynamic data are suggested.