一种与聚类系数相结合的链路预测方法

A link prediction method based on clustering coefficient

针对目前的链路预测方法考虑了共同邻居节点的数量,并没有综合考虑局部邻居节点的聚类系数信息,忽略了节点之间的紧密程度对链路预测的影响,故提出一种基于聚类系数的链路预测方法。该方法将共同邻居节点的聚类系数添加到结构相似性的四种链路预测方法中(CN,AA,RA,PA),并将改进的方法在四种数据集上同六种基于结构相似性的链路预测方法进行比较。实验结果证明,基于聚类系数的链路预测方法可以进一步提高链路预测的效果。

The number of common neighbor nodes is considered in current link prediction method,but the clustering coefficient information of local neighbor nodes isn′t considered comprehensively,and the influence of the closeness between nodes on the link prediction is ignored.Therefore,a link prediction method based on clustering coefficient is proposed,by which the clustering coefficient of common neighbor nodes is added to four link prediction methods(CN,AA,RA,PA)based on structural similarity.The improved method is compared with six link prediction methods based on structural similarity by using four datasets.The experimental results show that the link prediction method based on clustering coefficient can further improve the effect of link prediction.