Link Prediction of Complex Network Based on Eigenvector Centrality

Abstract

Abstract As one of the important methods connecting complex network and computer science, Link prediction deals with the most basic problems in information science. Therefore it is of great importance to probe into it. But how to improve the prediction accuracy is one of the focus problems we are facing. Most of the current link prediction methods are related to the indicators based on the similarity of nodes, and the importance of the neighbor nodes of nodes in the network is often determined by the similarity of nodes. indicators are ignored. Considering the aforementioned problems, we propose a link prediction algorithm based on eigenvector centrality calculated by node importance based on the eigenvector. The algorithm mainly uses the information of eigenvector centrality and considers Common Neighbor (CN), Adamic-Adar (AA) The similarity index of and Resource Allocation (RA), and the AUC value and the exact value are used as a reference for the pros and cons of the index, The results of simulation experiments are reported on two different network data sets, and the final results indicate that the algorithm based on eigenvector centrality is more accurate than the algorithm based on node importance in the link prediction of complicated networks.