摘要
【目的】自然灾害网络舆情的治理是综合减灾的重要内容。为探究重大暴雨洪涝灾害引发的网络舆情特征,总结相关舆情应对经验,【方法】以2021年发生的三场重大暴雨洪涝灾害为例,搜集了发布于新浪微博平台上的“山西暴雨”“河南暴雨”“湖北暴雨”话题相关的微博数据。利用基于BERT微调模型的机器学习方法进行了时间序列上的情感分析,梳理了灾害舆情议题的热度和情感强度变化;结合以关键词共现语义网络为核心的知识图谱方法,探寻了不同灾害的舆论热点变化特征。【结果】结果表明:(1)同类型暴雨洪涝灾害舆情通常具有相似的时序特征,即通常会持续较长时间,且舆情高峰都与关键性事件有关。(2)同类型暴雨洪涝灾害舆情情感色彩分布相似,但会因灾害严重程度的不同而产生相应变化。山西暴雨中正向情感占比为36.83%,负向占比为28.81%;河南暴雨则为34.23%和43.25%;湖北暴雨则为45.91%和27.07%。(3)类似暴雨洪涝灾害舆情具有相类似的主题分布,可分为对致灾因子和灾害本身的关注、对致灾过程的关注和应急抢险救援的关注、对洪涝灾害对于社会经济生态影响方面的关注。【结论】暴雨洪涝灾害引发的网络舆情的演变特征大致相近,但因灾害严重程度的差异,而呈现出一定的差异。针对这一特征,提出了如下建议:(1)做好日常舆情管控措施,及时做好突发灾害应对;(2)重视官方灾害信息及时发布,合理疏导和调控舆情;(3)建立舆情管控联动机制,吸取优秀管理经验。研究成果对于研判灾害舆情变化趋势,提升灾害舆情治理体系和能力具有积极意义。
[Objective] The governance of public opinion related to natural disasters is an important element of comprehensive disaster reduction. In order to explore the characteristics of network public opinion triggered by extraordinary rainstorm and flooding disasters, and summarize the experience of relevant public opinion response, [Methods] the three extraordinary rainstorm and flooding disasters that occurred in 2021 were taken as examples and Weibo data related to the topics of “Shanxi rainstorm”, “Henan rainstorm” and “Hubei rainstorm” posted on the Sina Weibo platform were collected. A machine learning method based on the BERT fine-tuning model was used for sentiment analysis on time series to sort out the changes in the hotness and sentiment intensity of disaster public opinion issues;and a knowledge graph method with keyword co-occurrence semantic network as the core explored the characteristics of changes in public opinion hotspots for different disasters. [Results] The results show that(1) public opinion on the same type of rainstorm and flooding disasters usually has similar chronological characteristics, it usually lasts for a long time and the peaks of public opinion are associated with critical events.(2) The emotional color distribution of public opinion on the same type of rainstorm and flooding disasters is similar but will change accordingly depending on the severity of the disaster. The percentage of positive emotion in the Shanxi rainstorm in this study is 36.83% and 28.81% of negative emotion;in the Henan rainstorm, it is 34.23% and 43.25%;in the Hubei rainstorm, it is 45.91% and 27.07%.(3) Public opinion on similar rainstorm and flooding disasters has similar thematic distribution, which can be divided into concerns about the disaster-causing factors and the disaster itself, concerns about the disaster-causing process and emergency rescue and relief, and concerns about the socioeconomic and ecological impacts of the floods. [Conclusion] The evolutionary characteristics of online public opinion triggered by rainstorm and flooding disasters are roughly similar but show some differences due to the difference in the severity of the disasters. In response to this characteristic, the following recommendations are made:(1) daily public opinion control measures and timely response to sudden disasters should be done well;(2) the timely release of official disaster information and reasonably channel and regulate public opinion should be taken seriously;(3) a public opinion control linkage mechanism should be established and excellent management experiences should be learned. The research results are of positive significance to study and judge the changing trends of disaster public opinion and improve the disaster public opinion governance system and capacity.
作者
张谱
张豪
孔锋
孔赟珑
ZHANG Pu;ZHANG Hao;KONG Feng;KONG Yunlong(College of Humanities and Development Studies,China Agricultural University,Beijing 100083,China;College of Information and Electrical Engineering,China Agricultural University,Beijing 100083,China;Center for Crisis Management Research,Tsinghua University,Beijing 100084,China;Aerospace Information Research Institute,Chinese Academy of Sciences,Beijing 100101,China)
出处
《水利水电技术(中英文)》
北大核心
2023年第2期47-59,共13页
Water Resources and Hydropower Engineering
基金
国家自然科学基金项目(61901471,41801064,71790611)
中国高等教育学会2022年度高等教育科学研究规划课题(22DL0302)
国家重点研发计划政府间/港澳台重点专项项目(2018YFE0109600)
中国气象局软科学重点项目(2022ZDIANXM20)
北京市社科基金研究基地项目(19JDGLA008)
中央高校基本科研业务费专项资金项目(2021TC062)。
关键词
暴雨洪涝灾害
灾害舆情
应急管理
灾害比较
自然语言处理
降水
城市内涝
rainstorm flooding disasters
disaster public opinion
emergency management
disaster comparison
natural language processing
precipitation
urban waterlogging