基于主题-情感融合分析的突发公共卫生事件网络舆情演化研究

Research on the Evolution of Online Opinion on Public Health Emergencies Based on Topic-Sentiment Fusion Analysis

[目的/意义]针对突发公共卫生事件,借助自然语言处理技术,快速挖掘舆论热点和舆情演化特征,提升政府部门的应急管理能力。[方法/过程]将新冠肺炎疫情作为研究案例,首先搜集了疫情相关的微博文本,在划分为潜伏期、爆发期、衰退期三个阶段的基础上,使用LDA主题模型和语义规则构建的方法进行主题-情感的融合分析,并结合疫情期间的新闻事件探究了网络舆情的情感演化情况和负面舆论的关注热点。[结果/结论]研究发现,关于疫情的负面情绪微博大多集中在前中期,且主要来源于对疫情信息的不确定性;而后期积极态度随国内疫情好转而成为主流。此外,民众对新冠肺炎的传播、成因、境外输入和官方信息发布等子话题的情绪起伏较大。

[Purpose/significance]In response to public health emergencies natural language processing technology can quickly dig out hot topics and evolutionary features of public opinion to enhance the emergency management capacity of government departments.[Method/process]COVID-19 was used as a study sample.We firstly collected microblogs about COVID-19.After dividing them into three stages:latency outbreak and recession we used LDA topic model and semantic rule construction to conduct a topic-sentiment fusion analysis.We also explored the emotional evolution of online public opinion and the hot spots of negative public opinion in relation to news events during the epidemic.[Result/conclusion]Negative opinions were mostly concentrated in the early and middle stages of the epidemic mainly stemming from uncertainty about epidemic information while positive opinions became main stream in the later stages as the epidemic situation in China improved.In addition public sentiment scores on the sub-topics of pneumonia transmission causation imported cases and official information release fluctuated greatly.