A degree-related and link clustering coefficient approach for link prediction in complex networks

Link prediction plays a significant role in both theoretical research and practical application of complex network analysis, and thus has attracted much attention. Numerous similarity-based methods have been proposed to solve the link prediction problem, and various topological structure features of the network have been exploited to construct the similarity score. Most methods focus on the topological feature information of nodes rather than that of links. We define a degree-related and link clustering coefficient that can better describe the function of the common neighbor in distinct local areas. Then, the proposed clustering coefficient is applied to determine the similarity of node pairs. In particular, the node degree information of each endpoint is utilized to reflect the influence of the end node when exploring the similarity score. In addition, on small-scale, medium-scale, and large-scale real-world networks from different fields, our method is compared with some representative methods, including local similarity-based methods and graph embedding-based methods , and the performances are evaluated by two commonly used metrics. The experiment results show the feasibility and effectiveness of our method for networks with different scales, and demonstrate that prediction accuracy can be further improved by the novel measure of the degree-related and link clustering coefficient.

Automated summary

Link prediction is important in complex network analysis and has attracted much attention. Researchers have proposed a degree-related and link clustering coefficient to better describe the function of common neighbors in different local areas for determining node pair similarity. Experimental results demonstrate the feasibility and effectiveness of this method in networks of different scales.