Operations Research Center, Massachusetts Institute of Technology and The Charles Stark Draper Laboratory, Cambridge, Massachusetts 02139, USA.
College of Environmental Design, University of California, Berkeley and Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA.
Phys Rev E. 2019 Jan;99(1-1):012323. doi: 10.1103/PhysRevE.99.012323.
We define a structural property of real-world large-scale communication networks consisting of the recurring patterns of communication among individuals, which we term persistent cascades. Using methods of inexact tree matching and agglomerative clustering, we group these patterns into classes which we claim represent some underlying way in which individuals tend to disseminate information. We extend methods from epidemic modeling to offer a way to analytically model this recurring structure in a random network, and comparing to the data, we find that the real cascading structure is significantly larger and more recurrent than the random model. We find that the cascades reveal a habitual hierarchy of spreading, alternative roles in weekday vs weekend spreading, and the existence of hidden spreaders. Finally, we show that cascade membership increases the likelihood of receiving information spreading through the network through simulation on the real order of communication events.
我们定义了一个真实世界中大型通信网络的结构属性，该属性由个体之间的通信重复模式组成，我们称之为持久级联。使用不精确树匹配和凝聚聚类的方法，我们将这些模式分组到类中，我们声称这些类代表了个体传播信息的某种潜在方式。我们从传染病模型扩展了方法，提供了一种在随机网络中分析模型这种重复结构的方法，并且与数据进行比较，我们发现真实的级联结构比随机模型显著更大且更频繁。我们发现，这些级联揭示了传播的习惯性层次结构，工作日和周末传播的替代角色，以及隐藏传播者的存在。最后，我们通过在真实的通信事件顺序上进行模拟，展示了级联成员身份增加了通过网络接收信息传播的可能性。
