[Purpose/significance] The formation and evolution of group polarization of network public opinion in emer? gencies are affected by many risk factors from physical space, social space and information space. The identification of group polarization risk is based on the comprehensive sorting and analysis of the formation and evolution mechanism of group polarization risk of network public opinion in emergencies and the judgment of whether the risk is generated. It is the leading link of social risk assessment and control. [Method/process] This paper demonstrates that extreme views are the most important concrete form of group polarization risk, analyzes the characteristics of extreme views, constructs a TCMCR identification model of extreme views. Based on the archetypal concept in self classification theory, extreme views are identified, and the archetypal views are expressed by the value of meta contrast ratio. Taking the public opin? ion of "Pfizer Paxlovid" under the outbreak of COVID-19 as an example, the empirical research was carried out. [Result/conclusion] The model constructed in this paper can effectively identify the extreme viewpoints of group polar? ization of network public opinion in emergencies. The empirical research results show that the phenomenon of group po? larization has occurred in viewpoint O13, which forms the risk of group polarization.