A new centrality measure based on random walks for multilayer networks under the framework of tensor computation

摘要

In this paper, we introduce a fourth-order tensor to represent multilayer networks and establish a new centrality for identifying essential nodes based on random walks, referred to as the BORW centrality. We first obtain the hub and authority scores of nodes and layers by solving limiting probabilities arising from multilayer networks. Then, we establish a new centrality by integrating hub and authority scores of nodes. We also propose a novel iterative algorithm to solve tensor equations to get limiting probabilities. The existence and the uniqueness of limiting probabilities were proven by using Brouwer fixed point theorem. The numerical experiments on a simple multilayer network and two real-world multilayer networks(i.e., European Air Transportation and FAO trade multilayer networks) show that the proposed centrality outperforms some existing centrality measures. (C) 2019 Published by Elsevier B.V.