To meet the demands of government crisis response in the evolution of network public opinion,an internet users’ negative emotion model based on SOAR(State,Operator and Result)model is proposed,where internet users is treated as Agent,behavior transformation process of internet user group in network public opinion is considered as continuous transformation process of state with time in the corresponding public opinion space in this paper.Then,working memory,long-term memory,decision-making process,learning mechanism of Agent are designed,internet user behavior transformation rule repository is constructed,and SOAR Agent model to reveal behavior transformation rules of internet user group is developed ultimately.Moreover,an simulation experiment is designed,typical public opinion events are seleted as samples,group behavior transformation process of Weibo users under the influence of different government emergency measures is conducted,and the SOAR Agent model of internet users’ negative emotion is validated.As result of this research,the collective behavior transformation rules of behavior transfer of internet users in the different environments could be analyzed and predicted based on the modelling of internet users’ negative emotion,such as different period of internet public opinion,different government crises response strategy.