统计与数据科学学院知识库

  突发公共卫生事件引发的网络舆情难以预测，且易在社会层面内引发负面情绪，导致舆论环境恶化，进而对管理决策和应对措施产生不利影响。因此，挖掘突发公共卫生事件网络舆情演化规律及情感时空变化特征，可以为政府及相关管理部门在处理相关事件时提供参考，及时识别舆情演化的周期阶段和情感倾向时空特征，更能为引导舆论方向和差异化防控提供科学依据。

  新冠肺炎疫情对经济、社会及人民生活产生了巨大的影响。本文以新冠肺炎疫情为案例，研究突发公共卫生事件网络舆情演化规律及情感时空变化特征。首先从微博中爬取有关新冠肺炎疫情的文本数据。其次，结合生命周期理论和微博转发数划分出网络舆情传播阶段，接着结合TF-IDF算法与LDA主题模型，挖掘事件在时序上的舆情主题演化规律和映射出的情感倾向。最后将用户地理信息和文本中蕴含的地理信息引入网络舆情分析中，进行网络舆情空间分析，获得不同区域网民对新冠肺炎疫情事件的关注程度与情感倾向。

  研究发现，在时间维度上，网络舆情传播周期一般可以分成四个阶段，分别为酝酿期、爆发期、波动期及长尾期。不同阶段下网络舆情主题演化呈现出主题关联、主题易变的规律。在舆情主题演变的背景下情感倾向变化主要可以分为两个阶段，前期的情感主要为紧张、慌恐和焦虑的负面情绪，后期的情感主要为振作、自信、平复及积极的正面情绪；在空间层面上，突发公共卫生事件网络舆情具有较强的空间特性，不仅具有全域覆盖性和全民参与性，而且与突发公共卫生事件本身空间分布特性整体具有相似一致性，但是存在局部差异性，呈现出的区域情感值与疫情事件的严重程度为负向相关性。

     Internet public opinion caused by public health emergencies is difficult to predict, and it is easy to trigger negative emotions at the social level, leading to a deterioration of the public opinion environment, which has a negative impact on management decision-making and response measures. Therefore, mining public opinion evolution rules and emotional spatiotemporal characteristics of public health emergencies can provide a reference for the government and relevant administrative departments when dealing with related events, and timely identify the evolutionary cycle stages of public opinion and the temporal and spatial characteristics of emotional tendencies, which can be more guidance the direction of public opinion and differentiated prevention and control provide a scientific basis.

      The new crown pneumonia epidemic has had a huge impact on the economy, society and people's lives. This paper uses the new crown pneumonia epidemic as a case to study the evolutionary law of public opinion on the Internet and the characteristics of emotional temporal and spatial changes in public health emergencies. Firstly, the paper obtain text data about the novel crown pneumonia epidemic from Weibo. Secondly, combining The Life Cycle Theory and the number of Weibo reposts to divide the network public opinion dissemination stage, and then combining the TF-IDF algorithm and the LDA topic model to mine the evolution law of the public opinion theme and the emotional tendency mapped out in the event. Finally, the user's geographic information and the geographic information contained in the text are introduced into the online public opinion analysis, and the spatial analysis of the online public opinion is carried out to obtain the degree of attention and emotional orientation of netizens in different regions to the new crown pneumonia epidemic.

     The study found that in the time dimension, the network public opinion dissemination cycle can generally be divided into four stages, namely the incubation period, the outbreak period, the volatility period and the long tail period. At different stages, the evolution of online public opinion topics shows the laws of topic relevance and variability. In the context of the evolution of public opinion themes, the change of emotional tendency can be divided into two stages. The early emotions are mainly negative emotions such as tension, panic and anxiety, and the later emotions are mainly refreshing, confident, calm and positive positive emotions. At the spatial level, public opinion on public health emergencies has strong spatial characteristics. It not only has global coverage and public participation, but also has similar consistency with the spatial distribution characteristics of public health emergencies, but there are local differences. The regional emotional value presented is negatively correlated with the severity of the epidemic event.