Identifying Influential Nodes in Complex Networks Based on Local and Global Methods

Abstract

Identifying Influential Nodes in complex networks is of great significance in both theory and reality. K-shell decomposition method is a local method which is suitable for increasing scale of complex networks but limited in accuracy because many nodes are partitioned with the same K-shell value. To overcome the coarse result of K-shell, an improved K-shell which considers the number of nodes’ iteration layers and degrees is proposed. Unlike local methods, global methods such as Betweenness Centralities (BC) are accurate but time-consuming. We employed an algorithm framework which combines advantages of both local and global methods where core network is extracted by improved K-shell and then BC is used to quantitatively analyze nodes in the core network. We compare the proposed method with other existing methods on Susceptible-Infective-Removal (SIR) mode. Experiments on three real networks show that the proposed method is more efficient and accurate.