基于LDA-CNN-BiLSTM与四象限法则的突发公共事件舆情主题演化研究

Research on the Theme Evolution of Public Opinion of Public Emergencies Based on LDA-CNN-BiLSTM and Four Quadrant Rule

[目的/意义]挖掘突发公共的舆情主题演化规律,探究大众情绪的时序趋势,提供政府相关部门决策与监测的参考模型。[方法/过程]以新冠疫情期间(2020年1月1日—2020年3月24日)的公共卫生事件为例,基于舆情分析的理论框架,综合运用LDA主题模型、TFIDF、CNN-BiLSTM、四象限法则等理论和技术,结合舆情发展的生命周期理论,动态挖掘不同阶段下的突发公共舆情主题关注度演化,分析社会公众情感演化趋势,最后基于四象限法则讨论不同阶段下的主题分布态势。[结果/结论]该模型能够追踪突发公共舆情不同阶段下社会大众的关注热点与情绪波动,并提供政府相关部门处理类似舆情时的决策参考与监测框架,具有一定的现实意义与理论价值。

[Purpose/significance] The paper discovers the theme evolution laws of public opinion of public emergencies,explores the time sequence trend of public sentiment,and provides a reference model of decision-making and monitoring for relevant government departments [Method/process] Taking the public health events during the period of COVID-19(January 1,2020-2020 March 24 th) as an example,based on the theoretical framework of public opinion analysis,the paper uses synthetically the theories and techniques of LDA thematic model,TFIDF,CNN-BiLSTM and four quadrants rule,combining with the life cycle theory of public opinion development,excavates dynamically the theme evolution of public opinion in different stages,analyzes the trend of social public sentiment evolution,and finally discusses the theme distribution situation in different stages based on the four quadrants rule.[Result/conclusion] The model above can track the hot spots and emotional fluctuations of public opinion in emergencies at different stages,and provide decision-making references and monitoring framework for relevant government departments to deal with similar public opinion,which has certain practical significance and theoretical value.