An Efficient Centrality-Based GNN for Community Detection in Dynamic Networks

Network communities play a critical role in understanding the organization of dynamic networks. Detecting communities in such networks is vital for comprehending their functionalities and evolution. This is evident in areas like social networks, biology, and information systems. Effective methods for community detection contribute to a deeper understanding of complex systems and their behaviors. This paper presents PDM-GCN, a centrality and preferential deletion-based graph neural network (GNN) for community detection in dynamic networks. The proposed approach calculates the edge betweenness centrality to determine the importance of nodes in the network and applies the preferential deletion model to simulate network dynamics. GNN architecture is built to predict connections between network nodes. The performance of the proposed method is evaluated through node classification accuracy measures and the comparison of network properties between the original and predicted networks. Comparative evaluations with state-of-the-art methods are conducted to assess the effectiveness of the proposed approach. The results demonstrate the potential of this proposed approach for community detection in dynamic networks. The proposed method is generic and can be applied to any network for analysis, irrespective of the underlying domain.