A Novel Link Prediction Algorithm Based on Deepwalk and Clustering Method

Abstract

Link prediction has attracted more and more attention due to its wide application in social network analysis, bioinformatics, and personalized recommendation. One of the methods for judging whether two nodes have connections in the network is to calculate the similarity. This method not only has low computational complexity, but also can achieve better prediction results, and it's more suitable for large-scale networks. There are many similarity indexes proposed until now, but most of them only consider degree of the node and its common neighbors. With the proposal of Deepwalk, many people applied it to link prediction. However,the method of judging the similarity between two nodes simply by their distance is also one-sided. In this paper, we propose a new similarity index, called Deep Affinity (DA) index, through combining the traditional similarity index with the distance index obtained by Deepwalk and introducing the idea of clustering at the same time. After conducting experiments on different network datasets, the results show that DA-based link prediction algorithm has greatly improved the prediction accuracy, especially for large-scale network datasets.