基于无标度网络的微博流言传播模型与实证分析

Modeling Rumor Diffusion on Scale-free Network Combined with Empirical Analysis on Microblog

流言在社交媒体中滋生和蔓延已经成为日益凸显的社会问题,特别是社会网络已成为影响流言传播的重要变量。随着计算方法的发展,网络建模作为复杂社会系统的定量描述工具,为探索流言的网络结构特性和传播机制提供了可能性。本研究借鉴了经典的SIR传染病模型,将微博流言传播网络中的个体划分为四种类型:易感染者、积极感染者、消极感染者和免疫者。我们构建了基于无标度网络的微博流言的传播动力学模型(SIpInR模型),并运用多主体仿真平台NetLogo进行了模拟仿真实验。研究发现,网络中个体对于流言的积极感染率越高,流言传播的强度和广度越大;转移率的变化对流言传播强度有影响,但对传播的广度和长度没有影响;相反地,恢复率和免疫率主要影响流言传播的广度和长度,而不影响传播的强度。本研究经过对真实微博数据的实证分析,检验了模型的前提和实验结果的可靠性,对进一步把握社交媒体流言传播规律和遏制流言具有参考意义。

Rumor breeding and spreading in social media are becoming an increasingly serious social problem,especially the social network has become an important variable affecting the process of rumor diffusion.With the development of computational methods,network modeling,which is a quantitative approach for complex social system,provides the possibility to explore the network structural characteristics and dynamics mechanism of rumor.In this study,drawing on the classic SIR infectious disease model,the agents in rumor diffusion network was classified into four types of status,such as susceptible,positively infected,negatively infected,and resistant.We proposed the SIpInR model,a dynamic model of rumor diffusion based on scale-free network and applied the multi-agent simulation platform,i.e.,NetLogo,to conduct simulation experiments.It was found that(a)the higher the positive infection rate of rumor we tuned,the intensively and broadly it spreads;(b)changing the transfer rate had influence on the intensity of rumor diffusion,rather than the breadth and length;(c)on the contrary,recovery and immunization rates mainly affect the breadth and length of rumor,rather than the intensity.Using empirical analysis with real data from microblog,the premise and experimental results of our model were validated.It will help for us to further understand and control the rumor diffusion in social media.