基于文本挖掘的城市景区密集人群风险感知*

为充分挖掘城市景区密集人群风险信息和隐藏规律，以携程、美团、大众点评3个旅游网站上发表的关于城市景区密集人群的游记和评论作为数据源，运用潜在狄利克雷分配(LDA)主题模型及Word2Vec词向量模型挖掘密集人群风险感知主题与关键词及二者之间的关联关系；基于计划行为理论构建城市景区密集人群风险感知模型，结合共现规律对风险感知主题的关联关系分析，明确景区密集人群对个体风险感知的作用规律。研究结果表明:降低个体在密集人群情景下的风险感知水平，将避免公众产生消极行为，降低踩踏事故发生的可能性；同时通过主题共现分析得出影响个体风险感知的核心主题为“网红表演”主题，可通过提高热门表演管制力度来提高个体主观规范对实际行为的正向影响，使得个体风险感知强度处于稳定水平。研究结果可以完善城市景区密集人群风险管理和风险沟通机制，从而有效管理风险。

Risk perception of dense crowd in urban scenic spots based on text mining

In order to fully explore the risk information and hidden law of dense crowd in the urban scenic spots,taking the travel notes and comments about the dense crowd in urban scenic spots published on Ctrip,Meituan and Public comments as the data source,the potential Dirichlet distribution (LDA) topic model and the Word2Vec word vector model were used to mine the risk perception topics and keywords of dense crowd and their correlation.Based on the theory of planned behavior,a risk perception model of dense crowd in the urban scenic spots was constructed,and combined with the correlation analysis of co-occurrence law on the topic of risk perception,the action law of dense crowd in scenic spots on the individual risk perception was clarified.The results showed that reducing the risk perception level of individuals in the situation of dense crowd would avoid the negative behavior of the public and reduce the possibility of stampede accidents.At the same time,through the topic co-occurrence analysis,it was concluded that the core topic affecting the individual risk perception was the topic of "Internet celebrity performances".The positive impact of individual subjective norms on actual behavior could be improved by improving the control intensity of popular performances,so as to make the intensity of individual risk perception at a stable level.The research results can improve the risk management and risk communication mechanism of dense crowd in the urban scenic spots,so as to effectively manage the risk.