基于物理信息神经网络方法的群体极化现象动态分析研究  

A Dynamic Analysis of Group Polarization Phenomenon Based on Physical Information Neural Network Methods

作　　者：刘先博 窦逸飞 杜彦辉[1] 冯俊又 陈李舟 徐煊翔 Liu Xianbo;Dou Yifei;Du Yanhui;Feng Junyou;Chen Lizhou;Xu Xuanxiang(School of Information and Cyber Security,People’s Public Security University of China,Beijing 100038;School of Criminology,People’s Public Security University of China,Beijing 100038)

机构地区：[1]中国人民公安大学信息网络安全学院,北京100038 [2]中国人民公安大学犯罪学学院,北京100038

出　　处：《情报探索》2025年第3期16-21,共6页Information Research

基　　金：“中国人民公安大学研究生科研创新计划项目”(项目编号:2023yjsky012);中央高校建设世界一流大学(学科)和特色发展引导专项“中国人民公安大学网络空间安全执法技术双一流创新研究专项”(项目编号:2023SYL07)的研究成果之一。

摘　　要：[目的/意义]旨在研究群体极化影响下的网络谣言传播过程,防范治理群体极化现象出现后产生的一系列非理性行为。[方法/过程]基于物理信息神经网络方法,融合犹豫机制、直接免疫机制、遗忘机制及媒体报道的辟谣机制等因素,探究网络谣言群体极化现象的动态传播过程,构建了CE-USDR谣言传播模型并对“河南郑州7·20暴雨”事件进行数据拟合,有效验证了模型在群体性舆情事件中的拟合性能。[结果/结论]针对群体极化阈值、谣言初始传播率、辟谣开始时间等因素采取应对策略有利于降低谣言传播规模,进而实现群体性事件谣言的有效管控引导。[Purpose/significance]This paper intends to study the process of network rumor propagation under the influence of group polarization,and to prevent and control a series of irrational behaviors after the emergence of group polarization.[Method/process]Based on the physical information neural network method,this paper integrates the factors of hesitation mechanism,direct immune mechanism,forgetting mechanism and rumor refuting mechanism of media reports,explores the dynamic propagation process of group polarization phenomenon of network rumors,constructs the CE-USDR rumor propagation model and carries out the data fitting of the“7·20 rainstorm in Zhengzhou,Henan”event,which effectively verifies the fitting performance of the model in mass public opinion events.[Result/conclusion]Adopting coping strategies for factors such as group polarization threshold,initial rumor spread rate,and rumor refutation start time is conducive to reducing the scale of rumor spread,thereby achieving effective control and guidance of group event rumors.

关 键 词：网络谣言 谣言传播 群体极化 物理信息神经网络 群体性事件 

分 类 号：G206[文化科学—传播学]