李劲松¹,
彭建华¹,
刘树新^1,,,
季新生^{1, 2}
1.中国人民解放军战略支援部队信息工程大学 郑州 450001
2.清华大学计算机科学与技术系 北京 100084
基金项目:国家自然科学基金(61803384)

详细信息

作者简介:李劲松：男，1992年生，博士生，研究方向为复杂网络，链路预测，网络安全
彭建华：男，1966年生，研究员，研究方向为网络安全，云安全，复杂网络
刘树新：男，1987年生，助理研究员，研究方向为复杂网络，链路预测，网络演化
季新生：男，1969年生，教授，研究方向为网络安全，云安全，复杂网络

通讯作者:刘树新　liushuxin11@126.com

中图分类号:N92; TP393

计量

文章访问数:1256
HTML全文浏览量:472
PDF下载量:77
被引次数:0

出版历程

收稿日期:2019-09-20
修回日期:2020-05-25
网络出版日期:2020-06-01
刊出日期:2020-10-13

A Link Prediction Method in Directed Networks Via Linear Programming

Jinsong LI¹,
Jianhua PENG¹,
Shuxin LIU^1,,,
Xinsheng JI^{1, 2}
1. Information Engineering University, People’s Liberation Army Strategic Support Force, Zhengzhou 450001, China
2. Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China
Funds:The National Natural Science Foundation of China (61803384)


摘要
摘要:大多数有向网络链路预测方法在计算节点相似性时没有充分考虑有向网络的结构特点，未区分不同有向邻居对连边形成具有的贡献差异，导致预测性能受到局限。鉴于此，该文提出一种基于线性规划的有向网络链路预测方法。该方法对3种有向邻居的信息贡献进行量化分析，结合结构特点建立线性规划模型，进而通过求解贡献矩阵的最优解构建相似性指标。9个真实有向网络中的实验结果表明，所提方法相比于9种现有方法在两种衡量标准下表现出较高的预测性能与良好的鲁棒性。
关键词:有向网络/
链路预测/
节点相似性/
线性规划
Abstract:Most existing link prediction methods in directed networks fail to consider the structural properties of directed networks when calculating node similarity, nor do they differentiate the contributions of directed neighbors on link formation, resulting in the limitation on prediction performance. To solve these problems, a novel link prediction method in directed networks based on linear programming is proposed. The contributions of three types of directed neighbors are quantified, then the linear programming problem is established based on network topological property. The similarity index is deduced by solving the optimal solution of the linear programming problem. Experimental results on nine real-world directed networks show that the proposed method outperforms nine benchmarks on both accuracy and robustness under two evaluation metrics.
Key words:Directed network/
Link prediction/
Node similarity/
Linear programming



PDF全文下载地址:

https://jeit.ac.cn/article/exportPdf?id=a09f5fad-40a1-480b-9018-8fff59b3dfa4