As an attractive area of study in complex networks, link prediction of directed networks aims to mining unknown links in networks. Network evolution mechanism is the theoretical source and basis of many link prediction methods. However, the existing similarity-based methods neglect the node's own topological stability and network structure stability when calculating the similarity between nodes. In order to solve this problem, this paper proposed a prediction method based on topological stability. Firstly, it analyzed the topological stability of nodes, and then calculated the similarity between the stable node pairs. Under the three metrics of AUC, precision and ranking score, the experimental results in nine real networks show that the proposed method has higher prediction accuracy.

[2020-12-05] Printed Article /doc/2019.09.0550/v2