摘要：

Abstract:

【Purpose/significance】 Provide ideas for public opinion guidance personnel in terms of the evolution of public opinion emo⁃

tions and themes at various stages, in order to further improve public opinion guidance strategies and ensure the security of the online

society.

【

Method/process】 Based on the theory of crisis life cycle, the "Chongqing Starbucks Driving Out Police Officers" event that

occurred on February 13, 2022 was selected as the research object. This article uses Python crawler method to obtain Weibo comment

data source, and uses TF-IDF combined with SVM support vector machine to build a network public opinion comment sentiment clas⁃

sifier to study the emotional evolution of the "Chongqing Starbucks Driving Out Police Officers" event, Finally, combined with case

analysis and reference from domestic research, a network public opinion guidance model for sudden public events was successfully

constructed.【Result/conclusion】 The emotional evolution of public opinion events is influenced by multiple factors, among which self

media creators, opinion leaders, and authoritative organizations are the main sources of influence. At the same time, public opinion

guidance should focus on strengthening public opinion monitoring during the incubation period, guiding public emotions during the

outbreak and diffusion periods, making information public and transparent. In the recovery period, attention should be paid to the re⁃

currence of public opinion events, strengthening public education, and improving public information literacy.【Innovation/limitation】

Only a single case will be used as the research object, and in the future, research samples will be added to better analyze the guiding

strategies of online public opinion.