Influential nodes identification in complex networks based on information entropy and discrete curvature

Abstract Identifying influential nodes is a critical task in the study of complex networks. While previous approaches have provided effective strategies for identifying influential nodes, they often suffer from weak connections between local structures and global connectivity. To overcome this limitation, this study presents a novel hybrid method that integrates information entropy with discrete curvature. Specifically, information entropy is introduced to quantify the cycle and tree structures around neighborhoods, providing a precise description of local structures. Meanwhile, discrete curvature is employed to measure the distribution of edges connected to influential nodes, thereby highlighting the importance of preserving the global connectivity of the complex network. The approach modifies the impact of local structures and global connectivity influences by considering heterogeneity within networks. This leads to a more precise ranking of influential nodes for achieving optimization control objectives. Experimental results demonstrate its effectiveness in networks datasets and indicate improved performance in detecting influential nodes.