Efficient Algorithm to Find Influential Nodes Using Community-Based Hybrid Centrality in Complex Networks

Finding influential nodes is a crucial task for understanding and optimizing information spreading in complex networks such as social, biological, transportation, and technological networks, where group-based interactions are common and efficiently represented by using hypergraphs. Traditional centrality measures fail to capture the nuanced relationships in inherent hypergraph structures. To overcome these limitations, we propose a novel centrality measure Isolating Relative Average Shortest Path (IRASP), by combining both isolating centrality and relative average shortest path. It considers both neighborhood structure, path connectivity, and the relative change of shortest path in the hypergraph after removing the node. In addition to our study, we integrated hypergraph community detection to refine and ensure the structural modularity to select diverse influential nodes. IRASP achieves higher efficiency than existing centrality measures by leveraging local structure and relative path changes. When combined with community detection, it significantly reduces computation time, making it well-suited for large-scale hypergraphs. We evaluated our method on six real-world hypergraph datasets, by employing the Kendall Tau correlation and diffusion model on community structures. Integrating community detection significantly reduces the computation time, making IRASP particularly effective for large-scale hypergraphs and improving influence detection up to 0.05%.