刘 继 武梦娇

(新疆财经大学统计与数据科学学院 乌鲁木齐 830012)

Assessment and Analysis of Network Public Opinion Situation Based on Bayesian Network：COVID-19 as an Example

Liu Ji Wu Mengjiao

(Institute of Statistics and Data Science, Xinjiang University of Finance and Economics, Urumqi 830012)

Abstract:[Purpose/Significance]Major emergencies put forward new requirements for improving national social governance capacity, and improving the ability of network public opinion situation assessment has become an important content of innovative social governance.[Method/Process]This paper constructs network public opinion situation assessment index from the aspects of the characteristics of Internet public opinion events, the degree of concern, the degree of diffusion and the tendency of Internet users' opinions. A novel coronavirus pneumonia network epidemic situation assessment model is constructed by Bayesian network, and the situation of Internet public opinion is evaluated and analyzed with the case of COVID-19.[Result/Conclusion]The COVID-19 has been tested on the Internet public opinion event. The proposed method has better public opinion situation assessment performance, and suggestions are put forward for the COVID-19 related network public opinion governance.

Keywords：internet public opinion; Bayesian network; situation assessment; COVID-19

0 引 言

随着互联网的高速发展，我国网民的规模逐渐扩大。据互联网络信息中心(CNNIC)发布的第45次《中国互联网络发展状况统计报告》显示，截至2020年3月，我国网民规模为9.04亿，互联网普及率达65.5%。互联网大规模的普及使得网民在获取信息的过程中可以跨越时间和空间的限制，当大量的民众在网络上对社会问题或现象表达其意见、态度、情绪时，便可能会形成网络舆情。若负面、极端等消极信息成为网络舆情的主流，则会产生网络舆情危机风险。2020年的“新冠肺炎疫情”揭示了我国公共安全社会治理能力亟待提升，如何提高网络舆情态势评估能力是当前亟待解决的问题。

在网络态势分析的研究中，学者们主要从社会科学的角度对网络舆情的指标体系构建、趋势发展及治理方式等方面进行分析，但随着网络数据呈指数型增长，单一的宏观定性分析已不能满足研究者对网络舆情的情景感知、传播机制、态势预测等方面的深入探究。随着人工智能、机器学习等理论的不断进步，为舆情分析提供了更丰富的定量分析方法。由于网络舆情发展过程是复杂的又是多成分的，马捷等[1]基于利用时间、预案准备情况、社会影响程度及关注人群数量四个维度构建网络舆情危机等级评价模型，通过对两个真实事件的危机等级判定，证明了该模型是合理的。Gu等[2]则是基于节点间的交互关系构建了舆情态势评估模型，但该模型只能针对某些特定事件进行评估，具有一定的局限性。……