A mixed strength decomposition method for identifying critical nodes by decomposing weighted social networks

¹ School of Economics and Management, Harbin Engineering University - Harbin, China
² Centre for Big Data and Business Intelligence, Harbin Engineering University - Harbin, China

^(a) E-mail: houjishan@163.com (corresponding author)

Received: 9 December 2022
Accepted: 30 May 2023

Abstract

Identifying critical nodes is an efficient strategy for preventing the dynamics of risk dissemination. The properties of edges connecting to the removed nodes are assumed to be the same by many decomposition methods. However, the edge weights are always different in weighted social networks since they have certain practical implications. In this study, a mixed strength decomposition (MSD) method is proposed to identify critical nodes in weighted social networks. This method aims to address the issue of not accounting for the information on removed nodes by considering both residual strength and exhausted strength. Three experimental analyses —the monotonicity test, Susceptible-Infected (SI) diffusion simulation, and successive node removal experiments— conducted on six real-world networks demonstrate that the MSD method has a competitive performance in identifying critical nodes, which overcomes the instability of the node strength and the degeneracy of the s-core method.