Evolution and response of flood disaster network public opinion based on an event logic graph: A case study of rainstorms in Henan, China

Social media takes the center stage as the primary information and opinion-sharing platform for the public during disaster events. Related topics quickly spread, and multiple public opinion events exhibit clear coupling phenomena, leading to the formation of network public opinion. By creating a network public opinion event logic graph, our study illustrates the development and evolution of network public opinion during disaster events and provides the government with a crucial foundation for action. For this study, we selected tweets on Sina Weibo regarding the "Henan rainstorm" between July 19, 2021, and August 2, 2021. To construct a network public opinion event logic graph, pattern matching was used to discover event relationships and extract causal event pairs and sequential event pairs. Moreover, the term frequency-inverse document frequency +K-means clustering algorithm (TF-IDF +K-Means) clustering algorithm was used to group event pairs with high similarity into one category. The critical nodes and evolution pathways in the event logic graph were then thoroughly examined. Our findings demonstrated that the topic evolution of network public opinion was multidirectional in nature, covering topics such as disaster events, emergency rescue, public sentiment, media opinion, and livelihoodrelated events. The two main components of the evolutionary pathway of network public opinion are the evolution of the disaster itself and its evolution in the social sphere. Our study provides a technique for determining the dynamic evolution of network public opinion and provides useful consultation for the government's upcoming response to network public opinion during disaster events.