计算机工程 ›› 2017, Vol. 43 ›› Issue (12): 267-273,277.doi: 10.3969/j.issn.1000-3428.2017.12.048

• 开发研究与工程应用 • 上一篇    下一篇

改进的异构链路协同预测算法研究

刘鑫   

  1. (吉林建筑大学 城建学院,长春 130118)
  • 收稿日期:2016-10-26 出版日期:2017-12-15 发布日期:2017-12-15
  • 作者简介:刘鑫(1978—),女,讲师、硕士,主研方向为数据挖掘、网络安全。

Research on Improved Cooperative Prediction Algorithm of Heterogeneous Link

LIU Xin   

  1. (School of Urban Architecture,Jilin Jianzhu University,Changchun 130118,China)
  • Received:2016-10-26 Online:2017-12-15 Published:2017-12-15

摘要:

现有的链路预测方法无法保证预测的可靠性,应用局限性较大。为此,针对源节点相似节点和目标节点相似节点之间的当前链路信息,提出同质连接原理,设计不同类型节点的相关性指标,用于描述不同类型节点间的链路存在概率,并将其与传统的邻近性指标相结合,用于异构链路预测。融合异构信息网络中的被标记数据和无标记数据,给出一种异构链路协同预测算法,通过获得不同类型链路间的各种复杂关系,结合互补性预测信息,实现多种链路类型的协同预测。实验结果表明,该链路协同预测算法可有效提升异构信息网络的链路预测性能。

关键词: 异构信息网络, 未知链路, 同质连接, 链路协同预测, 邻近性指标

Abstract:

The existing link estimation methods cannot guarantee the reliability of prediction and limitations large.To solve this problem,according to the similar current link information between the similar nodes of the source node and the similar nodes of the target nodes,the homophily connection principle is introduced,and a relatedness measure between different types of objects is designed to compute the existence probability of a link.It also extends conventional proximity measures to heterogeneous links.Furthermore,the labeled and unlabeled data in heterogeneous information networks are combined,and a heterogeneous collective link prediction algorithm is proposed to predict multiple types of links collectively by capturing the diverse and complex relationships among different types of links and leveraging the complementary prediction information.Empirical studies on real-world tasks demonstrate that the proposed collective link prediction approach can effectively boost link prediction performances in heterogeneous information networks.

Key words: heterogeneous information network, unknown link, homogeneous connection, link collaborative prediction, proximity index

中图分类号: