摘要
突发事件应急管理中社交媒体数据质量参差不齐、难以直接为应急救援机构或志愿者的现场救援活动提供帮助,探究有助于从突发事件的社交媒体数据中快速挖掘出应急救援信息的关键特征,从而提升社交媒体数据的严谨性,推动社交媒体数据纳入正式的应急决策过程具有重要意义。以“微博”平台为例,通过对“微博”平台的分析和相关研究文献的总结,确定了8个潜在影响微博内容能否支撑应急救援行动的特征。基于“#河南暴雨互助#”话题下的微博内容、传播和用户维度抽取8个特征,以决策树模型为基准模型,通过CART算法评估各个特征对区分应急救援信息的贡献度。结果表明,信息内容地址信息特征、信息内容语言特征、信息主体特征是社交媒体中的应急救援信息分类的关键特征。
Social media data contains a lot of low-quality information in emergency information management,which makes it difficult to directly provide instant help for the on-site rescue activities of emergency management agencies or volunteers.In order to improve the preciseness of social media data and promote the inclusion of social media data into the formal emergency decisionmaking process,it is necessary to explore the features that can provide help to quickly mine high-quality emergency rescue information from the social media data in emergencies.Taking the“Weibo”platform as an example,this research,through the analysis of the“Weibo”platform and the summary of relevant research literature,determines eight characteristics that potentially affect whether the microblog content can support emergency rescue operations.Based on the content,communication and user dimensions of Weibo under the topic of“Henan rainstorm mutual assistance”,eight features are extracted.Taking the decision tree model as the benchmark model,the contribution of each feature to the help of distinguishing emergency rescue information is evaluated by CART algorithm.The results show that information content address information characteristics,information content language characteristics,and information subject characteristics are the key features of emergency rescue information classification in social media.
作者
沈洪洲
居玥
SHEN Hongzhou;JU Yue(School of Management,Nanjing University of Posts and Telecommunications,Nanjing 210003,China;Research Center for Information Industry Integration,Innovation and Emergency Management,Nanjing University of Posts and Telecommunications,Nanjing 210003,China)
出处
《智能计算机与应用》
2023年第7期19-26,共8页
Intelligent Computer and Applications
基金
国家自然科学基金(71974102)
江苏省研究生科研与实践创新计划项目(KYCX21_0833)研究成果之一。
关键词
微博
应急救援信息
基本特征
数据挖掘
决策树
Weibo
emergency rescue information
basic features
data mining
decision tree
作者简介
沈洪洲(1980-),男,博士,副教授,主要研究方向:社会化媒体、信息资源管理;居玥(1998-),男,硕士研究生,主要研究方向:社会化媒体。
