基于三元闭包的节点相似性链路预测算法

发布时间：2018-12-28 16:56

【摘要】：链路预测作为复杂网络分析的基本方法被应用到很多领域,完全基于拓扑结构信息的复杂网络链路预测仍然是一个具有挑战性的问题。三元闭包作为网络中最小局部结构,具有结构平衡和稳定的特征。提出了一种基于三元闭包的节点相似性链路预测算法,通过计算出每个节点在网络中所占三元闭包的权重,并将该权重用于节点相似性指标中,提出了3个相似性指标TWCN、TWAA、TWRA和具有调节参数的3个相似性指标TWCN*、TWAA*、TWRA*。在10个不同的网络数据集上的实验结果表明,所提算法能够提高链路预测的精度。不仅如此,通过分析实验结果,发现在社交网络中拥有较多三元闭包的节点具有局部稳定性,不倾向于建立更多的新链接;相反,拥有较少三元闭包的节点具有局部不稳定性,倾向于建立更多的新链接。这种现象也符合社会学中有关弱关系产生链接的现象。
[Abstract]:As a basic method of complex network analysis, link prediction is applied to many fields. It is still a challenging problem to predict complex network links based on topology information. As the smallest local structure in the network, the ternary closure has the characteristics of structural equilibrium and stability. A node similarity link prediction algorithm based on ternary closure is proposed in this paper. By calculating the weight of each node in the network, and applying the weight to the similarity index of nodes, three similarity indexes TWCN,TWAA, are proposed. TWRA and TWCN*,TWAA*,TWRA*. with three similarity parameters Experimental results on 10 different network datasets show that the proposed algorithm can improve the accuracy of link prediction. Moreover, by analyzing the experimental results, it is found that the nodes with more ternary closures in social networks have local stability and do not tend to build more new links. On the contrary, nodes with fewer ternary closures have local instability and tend to create more new links. This phenomenon is also in line with the sociological phenomenon of weak relations to produce links.
【作者单位】：安徽大学计算机科学与技术学院;
【基金】：国家自然科学基金Nos.61175046,61402006 安徽省自然科学基金No.1508085MF113 教育部人文社科基金项目No.1508085MF113~~
【分类号】：O157.5

【相似文献】