Link Prediction Algorithm Based on Node Structure Similarity Measured by Relative Entropy

Abstract

Abstract To solve the problem that the link prediction method based on local information ignores the influence of neighbor structure information on the similarity measurement of nodes, a link prediction method based on relative entropy and local structure of nodes is proposed. Firstly, the second-order local network is introduced to describe the local structure of nodes; then, the structural similarity between nodes is described by redefining the relative entropy; finally, the structural similarity of nodes is measured based on relative entropy, and the structural similarity index of the node structure is proposed considering the structure information of the neighbor. Experimental results on 7 real network data sets show that the proposed method can achieve better results and can be applied to networks with a small average aggregation coefficient compared with other similarity indexes based on local and global information, and also have better performance on large-scale networks.