一种基于节点局部相似性的复杂网络链路预测算法  

A COMPLICATED NETWORK LINK PREDICTION ALGORITHM BASED ON LOCAL SIMILARITY OF NODES

作　　者：白桦 马云龙[2] 毕玉 张为子 Bai Hua;Ma Yunlong;Bi Yu;Zhang Weizi(Shanghai Grandhonor Information Technology Co.,Ltd.,Shanghai 200072,China;Tongji University,Shanghai 210804,China)

机构地区：[1]上海高重信息科技有限公司,上海200072 [2]同济大学,上海201804

出　　处：《计算机应用与软件》2020年第5期298-301,308,共5页Computer Applications and Software

摘　　要：链路预测作为复杂网络中的一项基础研究,已经在社交网络分析、商品推荐、疾病预测等众多领域得到了广泛的应用。为了提高复杂网络中链路预测的准确性,提出一种基于Admic-Adar和Resource-Allocation的改进型算法Enhanced-Resource-Allocation。通过五个复杂网络数据集进行实验,采用AUC指标评价其效果,实验表明改进的算法在预测准确率上整体优于其他的局部相似性算法。As a basic research in complicated networks,link prediction has been widely used in social network analysis,merchandise recommendation,disease prediction and many other fields.To improve the accuracy of link prediction in complicated networks,this paper proposes an improved algorithm called Enhanced-Resource-Allocation based on Admic-Adar and Resource-Allocation.It was experimented on five complicated network data sets,and its effect was evaluated by AUC index.The experimental results show that the improved algorithm is better than other local similarity algorithms in predicting accuracy.

关 键 词：复杂网络 局部相似性 链路预测 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]