中文摘要：

链路预测作为复杂网络的一个重要研究方向，基于节点相似性指标进行预测是最为常用的一种方法．传统的链路预测方法通常使用共同邻居数目或节点的度来衡量节点之间的相似性．节点对之间的关系不仅与邻居节点数目和度有关，节点的聚类系数体现了节点的聚集能力，对产生链接会起到一定的作用．基于这个观点，提出一种结合节点度和聚类系数的链路预测算法．利用共同邻居节点的度和聚类系数计算被预测节点对之间的相似性．不仅充分利用网络局部结构信息，还能够体现出共同邻居节点之间的差异性．在十组实际数据集上的实验结果表明，提出的链路预测算法与传统的五个算法（CN，AA，RA，PA，Jaccard）和基于聚类系数的CCLP算法相比具有很好的预测效果．

英文摘要：

Link prediction is an important research direction in complex network area, in which the forecasting methods based on the similarity in-between two nodes is kind of usual link prediction strategies. Traditional link prediction approach is to measure the similarity between two nodes through using the number of theft common neighborhoods and their degrees. The relationship between two nodes not only is related to the properties of their common nodes and their degrees, but also may be related to node clustering coeffi- cient. Node clustering coefficient reflects the node＇s clustering ability and may play a role on the formation of links. Based on this idea, a new link prediction algorithm based on node degree and node clustering coefficient is proposed in this paper. The similarity between two nodes is calculated by the degree and clustering coefficient of common neighborhoods. The method not only make the best of networks＇ local structure ,but also can reflect the difference between common neighborhood nodes. Experiments on 10 real-networks show that our proposed algorithm is more accurate compared with five traditional link prediction algorithms （Common Neighbors, Adamic-Adar, Resource Allocation, Preferential Attachment, and Jaccard）, and the recently introduced CCLP algorithm based on node clustering coefficient.