基于真实数据集的动态网络时间中心性分析

Real trace driven temporal centrality analysis for dynamic networks

针对动态网络中时间中心性的适用性问题进行实证性研究.对动态网络建模方法进行了描述,在分析时间可达图构建方法的基础上,定义了与传统的度值、接近数、介数等中心性相对应的时间中心性度量.进而基于公开发布的第三方数据集以及来自本校校园网的用户在线数据,对时间中心性和静态中心性之间的相关性进行了实证性研究和分析.实验结果表明:在合适的时间跨度下,时间度中心性和静态度中心性具有较高且稳定的相关度,而接近数和介数在不同的度量方法下不存在显著的相关性.

Empirical studies were carried out with regards to the applicability of temporal centrality in the dynamic network.Description was made about the modeling approaches of the dynamic network,and on the basis of analyzing the methods of constructing temporal reachability graph,temporal centrality measurements corresponding to traditional degree,closeness and betweenness were defined.Then based on openly released third-party dataset and real trace data from SEU campus network,the correlation between static centralities and temporal centralities was examined and analyzed empirically.Experiment results show that,under an appropriate time span,degree centrality has relatively high and stable correlation when observed in temporal and static perspective,while closeness and betweenness centralities do not show significant correlation.