Critical Nodes Evaluation in Multiplex Heterogeneous Network Based on Gravity Model

Identifying critical nodes in networks is one of the important ways to explore the structural and dynamic characteristics of networks. The existing evaluation of key nodes in multiple heterogeneous networks often ignores the importance of different relationships. Based on this issue, this study proposes a method for evaluating critical nodes in multiplex heterogeneous networks, called the Gravity Centrality measure based on the Network Embedding (NEGC). The method calculates the importance of different link relationships and constructs a weighted network based on the multiplex heterogeneous networks. It uses graph neural network to extract node features and obtain embedding vectors. Based on the embedding vectors, it calculates the effective distance between nodes and improves the gravity model with the effective distance to complete the evaluation of critical nodes. Experimental results on two real network datasets using the Susceptible Infected Recovered(SIR) model show that this method can effectively and comprehensively evaluate critical nodes in multiplex heterogeneous networks.