Identifying Key Nodes in Multiplex Networks Based on Evidence Theory

Identifying key nodes in multiplex networks is a fundamental problem in the field of network science, which has attracted widespread attention from many scholars. In recent decades, various centrality methods have been proposed for identifying key nodes in multiplex networks. However, these existing methods only utilize the single structural information of multiplex network, resulting in unsatisfactory performance of algorithms. In this paper, we define a simple and effective multiplex centrality based on Dempster -Shafer(MCBDS) evidence theory, which combines local and semi-local information of networks. The proposed centrality trades off between the multi-degree centrality and Supra-Laplacian centrality values of every node in multiplex networks. The influences of both the multi-degree and Supra-Laplacian of each node are represented by basic probability assignment(BPA), which consists of elementshighandlow.Highandloware the absolute values of the difference between the centrality value of nodeiand the maximum and minimum centrality values of all nodes, respectively. Therefore, BPA can represent the degree of trust in nodeibeing an important node and the degree of trust in nodeinot being an important node when using multi-degree or Supra-Laplacian centrality to measure node importance. Then, the MCBDS is proposed by combining these BPAs. In the simulation experiment, we compare the performance of our proposed centrality with other existing centrality methods on three real-world multiplex networks by introducing Linear Threshold(LT) model and resolution entropy(RE) index. The experimental results indicate that our proposed centrality algorithm is a competitive and effective centrality algorithm.