一个有效的动态网络节点影响力模型

网络节点影响力度量对社会网络研究具有重要的价值，静态网络的影响力度量是目前研究的主要问题。然后社会网络的结构经常会随着时间变化，呈现出动态网络。静态网络节点影响力度量模型虽然可以对动态网络不同时间点上的快照进行度量，然后这种机制很难刻画动态网络节点影响力的变化过程。本文将动态网络建模为不同时间点网络的叠加快照，然后构建了动态网络边权重衰减和节点影响力衰减机制，基于衰减机制提出了动态网络节点影响力模型，模型可以应用于加权或无权动态网络节点影响力度量。为了客观衡量本文模型的性能，在一个模拟网络和三个真实网络进行了不同实验。在模拟网络上，将结果与人工标注的结果计算肯德尔系数，针对三个真实网络则进行了不同角度的影响力效果分析。实验结果表明本文模型不仅可以较好的刻画动态网络节点影响力的变化过程，还可以准确度量动态网络节点影响力。

An effective model for measuring node influence on dynamic network

Measuring node influence on network is important to the study of social network research. Currently, the main re-search focus on node influence on static networks. The structure of social network usually changes over time. Essentially, social networks are dynamic networks. Although models for measuring node influence on static network can be used to measure node influence in different snapshots for a dynamic network, it is difficult to demonstrate node influence fluctuation. In this paper, a dynamic network is modeled as a series of snapshots at different time points, and then the mechanisms of edge weight attenuation and node influence attenuation are constructed. Based on the attenuation mechanism, a model for measuring node influence on dynamic network is proposed. The model can be applied to weighted or unweighted dynamic networks. In order to evaluate the performance of the model, comprehensive experiments were conducted on a simulated network and three real networks. In the simulated network, the Kendall coefficient is calculated by comparing the experimental results with the results of manual labeling, and the effectiveness of the model on three real networks is analyzed using different evaluation methods. The experimental results show that the model can not only demonstrate node influence fluctuation, but also effectively measure the node influence in dynamic networks.