数据与计算发展前沿 ›› 2023, Vol. 5 ›› Issue (1): 115-127.

CSTR: 32002.14.jfdc.CN10-1649/TP.2023.01.011

doi: 10.11871/jfdc.issn.2096-742X.2023.01.011

• 技术与应用 • 上一篇    下一篇

网络舆情SIR模型优化与干预研究

杨超波1(),谢卫红1,2,王力纲2,3,*()   

  1. 1.广东工业大学，管理学院，广东 广州 510623
    2.广东工业大学，经济学院，广东 广州 510623
    3.广东工业大学，教务处，广东 广州 510006
  • 收稿日期:2022-08-03 出版日期:2023-02-20 发布日期:2023-02-20
  • 通讯作者: * 王力纲（E-mail: 50344950@qq.com）
  • 作者简介:杨超波，广东工业大学，博士研究生，主要研究方向为大数据、网络舆情算法优化等，代表论文《食品安全网络舆情的重复感染SIR模型研究》。
    在本文中负责设计方案、分析数据，撰写论文。
    YANG Chaobo is a Ph.D. candidate in the School of Man-agement at the Guangdong University of Technology. His research interests cover Big Data and algorithm optimization of network public opinion. His representative paper titled “Research on repeated infection SIR model of network public opinion about food safety” was published in the Journal of Systems Engineering (Issue 2, 2022).
    In this paper, his contribution is to design the scheme, analyze the data, and write the paper.
    E-mail: 1506473800@qq.com|王力纲，广东工业大学，博士研究生，主要研究方向为数字经济系统优化等，代表论文《基于数据包络分析的高校教育资源投入产出效率评价》。
    在本文中负责收集实验数据，修改论文。
    WANG Ligang is a Ph.D. candidate in the School of Economics at the Guangdong University of Te-chnology. His research interests cover optimization of digital economy system. His representative paper titled “On the eva-luation of input-output efficiency of the higher education resou-rce based on data envelopment analysis” was published in the Journal of Guangdong University of Technology (Social Sciences Edition) (Issue 4, 2011).
    His contribution is to collect experimental data and revise the paper.
    E-mail: 50344950@qq.com
  • 基金资助:
    国家自然科学基金项目“大数据背景下的网络隐私顾虑影响因素及行为效应研究：基于多维发展理论视角”(71672043);广东省自然科学基金项目“数字化创新对制造企业绩效影响机理研究：基于重组创新与资源编排视角”(2020A1515010971)

Research on Optimization and Intervention of SIR Model of Network Public Opinion

YANG Chaobo1(),XIE Weihong1,2,WANG Ligang2,3,*()   

  1. 1. School of Management, Guangdong University of Technology, Guangzhou, Guangdong 510623, China
    2. School of Economics, Guangdong University of Technology, Guangzhou, Guangdong 510623, China
    3. Department of Education, Guangdong University of Technology, Guangzhou, Guangdong 510006, China
  • Received:2022-08-03 Online:2023-02-20 Published:2023-02-20

RichHTML

23

PDF

97

摘要：

【背景】网络舆情对企业的健康发展产生越来越重要的影响作用。SIR传染病模型是常用的网络舆情传播研究模型，目前关于网络舆情传播的研究大部分是基于SIR模型及其变种。但现有SIR模型没有将感染者细分，不利于网络舆情的传播研究和精准化监控。【目的】通过优化SIR模型，使其更能反映企业网络舆情的真实情况，并提升监控效果。【方法】将SIR模型的感染者细分为积极的感染者、中性的感染者和消极的感染者等三类感染者。对于不同类型的感染者，其发帖率不同。设置差异化的发帖率，以提升企业网络舆情的监控和预测精准度。根据不同的网络舆情级别，设计三个不同干预级别的监控措施，提升监控效果。【结果】将改进模型应用到“海底捞大肠菌群不合格”的真实企业网络舆情，从效果对比得知，改进模型的监控效果比SIR模型的更理想。【结论】细分研究对象、考虑感染者的发帖率、制定不同监管力度的干预级别，有利于提升企业网络舆情监管的精准度、监管成效和预测准确度。

关键词: 网络舆情, 传染病模型, 传播

Abstract:

[Background] Network public opinion has an increasingly important influence on the healthy development of the enterprise. The SIR infectious disease model is a commonly used research model for network public opinion dissemination. At present, most researches on network public opinion dissemination are based on the SIR model and its variants. The SIR model does not subdivide infected persons, which is not conducive to propagation research and the precise monitoring of network public opinion. [Objective] By optimizing the SIR model, it can better reflect the real situation of enterprise network public opinion and improve the monitoring effect. [Methods] The work presented in this article divides the infected persons under the SIR model into three types: positively infected, generally infected, and negatively infected. For different types of infected people, the posting rate is different. Differentiated posting rates are set to improve the monitoring and prediction accuracy of enterprise network public opinion. According to the different levels of network public opinion, three monitoring measures with different intervention levels are designed to enhance the monitoring effect. [Results] The improved model is applied to the real enterprise network public opinion of the "unqualified coliform group of Haidilao". we found that the monitoring effect of the improved model is more ideal than that of the SIR model. [Conclusions] Subdividing the research objects, considering the posting rate of infected persons, and formulating intervention levels with different supervision strengths will help us to improve the accuracy, supervision effectiveness, and prediction accuracy of enterprise network public opinion supervision.

Key words: network public opinion, infectious disease model, dissemination