Abstract

Community detection is an NP-Hard problem that concerns itself with partitioning a network into groups such that nodes within a partition enjoy denser network connections compared to nodes in different partitions. The capacity to locate and examine these groups can offer invaluable assistance in comprehending and illustrating the framework of networks. Since the community detection problem is inherently complex, metaheuristic optimization algorithms are extensively employed to address this problem. A recently proposed metaheuristic population-based algorithm, Coronavirus Herd Immunity Optimizer (CHIO), draws inspiration from the COVID-19 herd immunity treatment strategy. We adapt the CHIO algorithm for the problem of community detection in social networks. In this proposal, the Network Modularity value is computed to assess the quality of a community structure.

Tuning the parameters of a metaheuristic algorithm to a given problem at hand is essential for good algorithm performance and is the focus of our proposal. The parameter tuning method developed by Genichi Taguchi is utilized to fine-tune the parameters of the CHIO algorithm in the context of community detection. Experiments on real-world benchmark networks were conducted. The adapted CHIO algorithm is run with parameter values obtained after tuning. It is noticed that the proposed approach is successful in detecting community structures with a high modularity value.

Statistics