关键词: 复杂网络, 链路预测, 三元闭包, 节点权重

Abstract:  Link prediction is a fundamental method for analyzing complex network which has been widely used in many domains. Link prediction in complex network based on solely topological information is a challenging problem. As one of the basic local structure, triadic closure has the characteristics of structural balance and stability. This paper proposes a link prediction algorithm to measure the similarity of nodes based on triadic closure. By calculating the weight of each node in the network according to the triadic closure, and using the weights in the node similarity index, this paper proposes three similarity indexes of TWCN, TWAA, TWRA and the other three similarity indexes of TWCN*, TWAA*, TWRA* with adjustment parameter. The experimental results on ten real network datasets demonstrate that the new method can improve the accuracy of link prediction. Moreover, by analyzing the experiment results, this paper finds that the network with more triadic closure nodes tends to be more stable. In other words, a network with less triadic closure nodes is less stable, and it is more likely to establish a new link with others. This phenomenon is also consistent with some related phenomenon in sociology on weak relations to generate links.

Key words: complex networks, link prediction, triadic closure, node weight