基于agent的多重社会网络中破产传染模型研究

Research of bankruptcy contagion model in multiplex network based on multiagent

社会网络中的信息扩散是目前国际上关注的研究方向,而破产传染作为社会金融中一类特殊而又重要的传播现象,也是当前社会经济领域非常关心的问题。然而,已有的金融传播研究中通常只基于单层的交互关系来分析传播行为,忽略了真实金融主体之间多重社会关系的交叉性。针对该问题,基于多智能体对多重社会网络中的破产传染现象进行建模,构建了纵向复合和并行独立多重网络传染模型,前者关注不同关系网络层次之间的跨层间影响,后者偏重于单层网络的独立性和并行性。设置不同的约束条件,对比实验了两种模型不同的传染效果特征。实验结果揭示了两种模型不同的适用范围:模型选择应由实验目标是否存在明显的不同社会网络层次间的相互影响决定,纵向复合多重网络破产传染模型更适用于社会网络层次之间间接影响明显的环境。

Having the issues of information diffusion in social networks and computing been intensively explored by many researchers,bankruptcy contagion,a special type of information diffusion in social networks,is especially a critical problem that attracts many researchers in the societal economy domain.However,existing financial communication studies usually only analyze communication behaviors based on single-layered interactions,ignoring the intersectionality of multiple social relationships between financial entities.Aiming at the problem above,this paper proposed a novel bankruptcy contagion framework for multiplex networks,which included the vertical composite multiplex network contagion model and the concurrent independent multiplex network contagion model,with the former focusing on the cross-layer influence among multiple social networks while having the latter on the independence and parallelism of a single-layer network.Furthermore,by setting different constraints,the experiment is set to compare and contrast the different infection effects between the two models.The result reveals of the experiment that the two models have different scopes in applications:the vertical combination multiplex network contagion model is more suitable for the environment with obvious correlations among varying social networks.