An Improved Link Prediction Algorithm of Complex Network

Abstract

In recent years, link prediction has become a hot spot in complex network research area. Its goal is to calculate the possibility of future connection between the currently unconnected nodes through the known topology and other information in the network. At present, most prediction algorithms are based on local similarity and focus on the common neighbors and degree of nodes without considering the impact of compactness between nodes on the prediction. For this reason, in this paper an improved link prediction algorithm, named as CCRA(Resource Allocation with Clustering Coefficient) is proposed by combining the local similarity property with clustering coefficient. Furthermore, the accuracy of this algorithm's prediction is verified through experiments on real data sets. Consequently, conclusion on this algorithm is derived thereby.