Abstract: 【Purpose/significance】Exploring the multi-attribute evolution law of online public opinion in major epidemics can provide a reference for the practice of online public opinion governance in major epidemic prevention and control.【Method/process】This paper constructed a multi-attribute evolution analysis model of online public opinion in a major epidemic based on a sociological perspec⁃tive. Then, it selected the Hubei Red Cross event Weibo public opinion data as the research object, and used the influence evaluation of opinion leaders, the LDA topic model and the Snownlp sentiment analysis method to analyze the Internet opinion leaders, topic dis⁃tribution and sentiment trends in major epidemics. It revealed the evolutionary law of multiple attributes of online public opinion in major epidemics from the three social attributes of crowd, content and emotion.【Result/conclusion】The results show that the online
public opinion of major epidemics can be divided into four stages: sudden period, outbreak period, cooling period and out-of-focus pe⁃riod. At each stage, there are obvious differences in opinion leaders, topic discussion content, and emotional orientation. The changes in the emotional stages of netizens are consistent with the evolutionary law of the theme attributes of public opinion.【Innovation/limita⁃tion】The model proposed in this study can effectively meet the requirements for in-depth mining of the evolutionary characteristics of major epidemics online public opinion, and provide a reference for the governance practices of major epidemics online public opinion. Future research can select a wider range of public opinion events and data sources to verify this model.