基于元路径的异构信息网络的影响力建模

Modeling of Influence of Heterogeneous Information Network Based on Meta-path

对影响力的评估不仅有助于识别有影响力的用户,而且有助于对社交平台和应用程序的设计提供方案。然而,现有的社会影响分析工作大多集中在同质信息网络上。很少有研究系统地研究如何挖掘异构信息网络中节点间的影响强度。为此提出了一种基于元路径的信息熵模型来模拟异构信息网络中的社会影响。通过设置元路径,不仅可以灵活地集成异构信息,还可以获取潜在的链路信息来度量节点的影响。实验结果表明,提出方法的MPIE的Kendall’s tau值为0.38,说明具有较好性能。

Evaluation of influence not only helps to identify influential users,but also helps to provide solutions for the design of social platforms and applications.However,most of the existing social impact analysis works focus on homogeneous information networks.Few studies have systematically studied how to mine the intensity of influence among nodes in heterogeneous information networks.This paper proposes a meta-path-based information entropy model to simulate the social impact in heterogeneous information networks.Setting meta-paths not only can integrate heterogeneous information flexibly,but also can obtain potential link information to measure the influence of nodes.The experimental results show that the Kendall’s tau value of MPIE is 0.38,which indicates that the proposed method has good performance.