Abstract:
Networks consist of interconnected nodes and edges that depict entities and their relationships. In social network clustering, nodes are
grouped into clusters based on their connectivity, to identify communities. However, community detection methods have not yet
leveraged the Weight-based Fish School Search algorithm, which is one of the promising approaches to finding community structure.
In this paper, we aim to apply a specific class of FSS-Based algorithm, which is weighted FSS, to network clustering. We have
developed a unique hierarchical network clustering method that leverages the Weight-based Fish School Search algorithm (WFSSC).
This methodology focuses on maximizing weights to enhance the modularity function, leading to the identification of community
structures in unipartite, undirected, and weighted networks. The process involves iterative network splitting and the construction of a
dendrogram, with the optimal community structure determined by selecting the cut that maximizes modularity. Our method employs
the modularity function for an objective assessment of the community structure, aiding in optimal network division. We evaluated our
methodology on known and unknown network structures, including a network generated using the LFR model to assess its adaptability
to different community structures. The performance was measured using metrics such as NMI, ARI, and FMI. The results demonstrated
that our methodology exhibits robust performance in identifying community structures, highlighting its effectiveness in capturing
cohesive communities and accurately pinpointing actual community structures.
