Research on Nodes Similarity in Complex Networks Based on Multistage Path Distribution Relative Entropy

In the complex network system, the similarity measure of nodes in the network is an important topic in the current research, which is of great significance for the study of complex network structure. In view of the current research situation, this paper proposes a method of relative entropy of multi-order path distribution to measure the similarity of nodes, and calculates the non-repeatable multi-order paths of edges and points in the network. This method can choose the order at will. Compared with other methods, it not only has less calculation, but also considers more multilateral information, making the similarity more sufficient. The comparison test of the real network data set shows that the method performs better in the mutual similarity and propagation influence index and the average aggregation coefficient evaluation index.