Community detection in complex networks using vertex cover and an adaptive community transformer engine

In recent years, the study of complex networks, such as biological, social, and economic systems, has attracted significant attention due to the complicated interactions among their entities. Community detection plays a key role in uncovering the hidden structure of these networks by dividing them into distinct subgroups, or communities. Research in community detection aims to extract relatively distinct sub-networks, known as communities, from the complex network structure to better understand its topology and functional organization. In this paper, we propose a novel community detection method based on a defined architecture composed of a Data Repository and three main components: Pre-Processing, Primary Communities Composition, and an Adaptive Community Transformer. In the first component, similarity measures are identified and stored, and appropriate weights are assigned to network links. The second component identifies significant network nodes using vertex cover along with the defined measures and weights. Primary community structures are then composed by considering vertex cover nodes as the centers of communities. After refining the primary community structure, an adaptive engine in the third component decides whether to merge or split communities to achieve the optimal community structure. We evaluate our method across various network sizes–small, medium, and large–including both real-world and artificial network scenarios. Compared to other approaches, the community structures detected by our method are size-independent and demonstrate strong evaluation metrics across all network types, especially in large-scale networks. Therefore, our proposal effectively detects community structures that resemble those found in real-world networks.