A Hybrid Framework for Community Detection Integrating Clauset Newman Moore (CNM), Graph Aggregation, and Cuckoo Search

The prime focus of community detection is to reveal groups of nodes that are densely connected within the network (internally) and sparsely connected to the rest of the network. Although, challenge remains to identify meaningful community structures in large and complex networks due to hierarchical organization, heterogeneous connectivity, and resolution-limit effects. The greedy modularity-based methods such as the Clauset–Newman–Moore (CNM) algorithm is computationally efficient but often suffer from immutable merge decisions and poor performance at multiple structural scales. This paper proposes a hybrid multi-scale community detection framework that combines CNM-based hierarchical aggregation with a Cuckoo Search–based refinement strategy to address these limitations. At first, CNM is applied to obtain an initial fine-scale partition, then community-based graph aggregation is followed to reduce the search space and mitigates resolution bias. At final step, a Cuckoo Search–based optimization mechanism employed Levy-flight-inspired node perturbations is used to escape local optima and refine community assignments by maximizing modularity. Experimental analysis conducted on multiple real-world and stimulate network, like Karate, Dolphin, Football, Facebook, Polbooks, Les-mis (Les Misérables), Jazz, and LFR benchmarks. The proposed hybrid approach indicates consistently improvement in modularity and clustering accuracy over standard CNM and other baseline methods.