Revisiting neighbourhood proximity based algorithm for overlapping community detection in weighted networks

A network is an indispensable tool for studying complex systems and large data. Any systems or data could be partitioned into components based on traits. This could be interpreted as the community in a network, and detection is essential for comprehending real-world systems and revealing profound insights from diverse realms. Over the past decade, much attention has been given to studying overlapping community structures without considering the weights of connections. However, the current frontier in network science involves expanding this understanding to encompass overlapping community structures with weighted connections. This introduces a new layer of complexity due to the nuanced nature of edge weights, making it an intriguing challenge for modern network science. In this paper, we proposed a refined version of the existing neighbourhood Proximity based Community Detection algorithm and call it ”neighbourhood Proximity based Community Detection using Weighted Centrality (NPCD-WC)” algorithm for extracting overlapping community structures in undirected weighted networks. This proposed algorithm is based on the weighted node centrality concept which makes it more stable and includes improved neighbourhood proximity measures. Subsequently, the proposed algorithm is based on expansion where a chosen node into a community is facilitated through an innovative adaptation of neighbourhood proximity. Two input threshold parameters ρ _0, θ _0 are needed to output cover of membership communities. Rigorous experimentation on real-world and artificial networks, employing well-established quality-based metrics compelling evidence that NPCD-WC outperforms other baseline algorithms. The algorithm evidenced time complexity better than most baseline algorithms and was highly used for applicable world.