Community and Local Information Preserved Link Prediction in Complex Networks

Abstract

With the popularity of social network, link prediction that aims to predict missing links in complex networks has attracted much attention. Among the existing link prediction methods, the method based on local information shows considerable competitiveness due to its simplicity and efficiency. However, the majority of local information based methods only perform the task by exploring low order neighbor information to make the prediction of new links, while the higher order information and community information are neglected. In this paper, we propose a community and local information preserved link prediction algorithm named CLLP for accurate link prediction in complex network. Specifically, we design a novel local similarity method based on probability propagation algorithm to get the local information of both low-order and high-order neighbor information, and propose a community information fusion strategy to integrate the community information into the suggested similarity method in order to obtain better prediction effect. Experimental results on real complex networks with different characteristics demonstrate the superiority of the proposed algorithm over several representative link prediction algorithms.