Identifying critical nodes in complex networks via semiclustering weighted leverage centrality

Abstract

Identifying key nodes in complex networks is of great significance for optimizing information dissemination, containing epidemics, and analyzing network robustness. The use of clustering coefficients to identify key nodes is a current research focus; however, existing studies have not yet applied this approach to leverage centrality (LC). Furthermore, in the identification of nodes based on LC, negative-leverage nodes are simply regarded as unimportant followers, overlooking the bridging role that nodes with low clustering coefficients may play. This paper proposes a new node centrality—semi-clustering weighted LC (SWLC)—for identifying critical nodes, from the perspective of synergizing node’s own attributes with its neighborhood structure. This method uses the clustering coefficient to apply differential weighting to negative-leverage nodes; it comprehensively evaluates a node’s importance by summing the transformed leverage values of the node itself and all nodes within its two-hop neighborhood through a two-hop aggregation strategy. The rationality analysis and ablation experiments conducted on the proposed method demonstrate that SWLC achieves stable and excellent performance, and that the two innovative modules need to work together to attain the desired effectiveness. To verify the effectiveness and applicability of the proposed method, this paper compares SWLC with nine other centrality methods using the susceptible–infected–recovered model on nine real-world networks of varying sizes and domains. Experimental results demonstrate that SWLC achieves favorable performance in terms of discriminability, overall node ranking accuracy, and critical node ranking accuracy, and is capable of effectively identifying critical nodes in complex networks.