“4·29”特别重大房屋倒塌事件舆情主题聚类及演化研究  

Research on the theme clustering and evolution of public opinion on the“4.29”particularly major house collapse incident

作　　者：晋良海[1,2,3] 王昕煜 张文 王抒情 JIN Lianghai;WANG Xinyu;ZHANG Wen;WANG Shuqing(Hubei Provincial Key Laboratory of Construction and Management of Hydropower Engineering,China Three Gorges University,Yichang 443002,Hubei,China;College of Hydraulic&Environmental Engineering,China Three Gorges University,Yichang 443002,Hubei,China;Safety Production Standardization Evaluation Center of China Three Gorges University,Yichang 443002,Hubei,China;The Seventh Geological Brigade of Hubei Provincial Geological Bureau,Yichang 443002,Hubei,China)

机构地区：[1]三峡大学水电工程施工与管理湖北省重点实验室,湖北宜昌443002 [2]三峡大学水利与环境学院,湖北宜昌443002 [3]三峡大学安全生产标准化评审中心,湖北宜昌443002 [4]湖北省地质局第七地质大队,湖北宜昌443002

出　　处：《安全与环境学报》2024年第7期2787-2796,共10页Journal of Safety and Environment

基　　金：国家自然科学基金面上项目(52179136);教育部人文社科基金项目(21YJA630038)。

摘　　要：研究房屋倒塌突发事件舆情主题及演化规律,可为应急主管部门引导和调控舆情提供实践指导。以长沙“4·29”特别重大房屋倒塌事件为对象,收集新浪微博平台上事件发生后8 d内的网民评论,运用词频逆文本频率(Term Frequency-Inverse Document Frequency,TF IDF)算法提取关键词,计算共现频率并刻画共现网络关系,耦合困惑度与K均值聚类(K Means)算法进行舆情主题聚类;采用对应分析方法,识别舆情主题的相关性,探究舆情主题随时间的热度变化趋势,揭示房屋倒塌事件舆情主题演化规律。结果表明:通过聚类得到的7个舆情主题依据相关性可耦合为救援善后T_(1)、调查追究T_(2)、安全防范T_(3)三大主题;以流言传播、官方召开发布会为最佳节点,将舆情传播划分为演化特征不同的前期、中期和后期。其中,前期主题T_(1)占优,中期主题T_(2)、T_(3)占优,后期主题T_(1)再次占优。通过类比分析化工爆炸事件舆情演化特征,验证了主题T_(1)、T_(2)、T_(3)对突发公共事件舆情演变起关键作用,满足公众对主题T_(1)、T_(2)和T_(3)的信息诉求是舆情平缓的关键。研究成果可为类似突发公共事件舆情治理提供参考。To mitigate the impact of public opinion during building collapse emergencies,this paper provides emergency authorities with practical guidance for regulating and guiding public opinion using the theme clustering and evolution research model.Focusing on the“4.29”particularly major house collapse incident in Changsha,a Python crawler collected netizen comments on Sina Weibo within 8 days of the incident.The Term Frequency-Inverse Document Frequency(TF IDF)algorithm is used to extract keywords,determine co-occurrence frequency,and describe the co-occurrence network relationship.Public opinion topics are analyzed through cluster analysis using the coupling confusion degree and K Means clustering algorithm.The applied analysis method identifies correlations among public opinion themes,explores their changing trends over time,and reveals the evolution law of public opinion themes during the house collapse event.The results suggest a divided public opinion on the collapse event,with both an aggregation effect and a series effect.The seven themes of public opinion that are clustered can be grouped into three primary themes.T_(1) is dedicated to rescue and rehabilitation,T_(2) to investigation and inquiry,and T_(3) to security and prevention.These themes allow for speculation about the incident s cause and suggest potential management strategies.Using rumor propagation and official press conferences as key points,the evolution of public opinion can be divided into early,middle,and late stages,each with distinctive characteristics.During the initial stage,T_(1) is dominant,followed by T_(2) and T_(3) in the middle,and T_(1) once again towards the end.An analysis of public opinion dynamics during chemical explosions indicates that T_(1),T_(2),and T_(3) are crucial factors in their evolution.It is crucial to meet the public s information requirements concerning these topics for stable public opinion.On the contrary,it is easy to create mistrust in officials and lead to the spread of widespread rumors,which presents grea

关 键 词：安全社会工程 建筑物倒塌 网络舆情分析 聚类 演化规律 

分 类 号：X915.2[环境科学与工程—安全科学]