Identifying influential nodes in complex network based on weighted semi-local centrality

Aiming at the problem that it is difficult to identify the influential nodes in complex network when risk comes, a weighted semi-local centrality measure which conquers the defect of semi-local centrality is proposed, and the method not only synthesizes the degree and weight of a node but also considers information of multiple layer neighbors of nodes. To evaluate the performance, the Susceptible-Infected(SI) model is used to estimate the spreading influence of the top-ranked nodes by different centrality. The experimental results on two simple weighted network and a real network show that our method can well identify influential nodes, and much better than degree and betweenness centrality ones, and almost as good as the ESC method and the closeness centrality measure while with much lower computational complexity.