Identifying multiple influential spreaders in complex networks based on spectral graph theory

Abstract

One of the hot research topics in propagation dynamics is identifying a set of critical nodes that can influence maximization in a complex network. The importance and dispersion of critical nodes among them are both vital factors that can influence maximization. We therefore propose a multiple influential spreaders identification algorithm based on spectral graph theory. This algorithm first quantifies the role played by the local structure of nodes in the propagation process, then classifies the nodes based on the eigenvectors of the Laplace matrix, and finally selects a set of critical nodes by the constraint that nodes in the same class are not adjacent to each other while different classes of nodes can be adjacent to each other. Experimental results on real and synthetic networks show that our algorithm outperforms the state-of-the-art and classical algorithms in the SIR model.