摘要：

Abstract:

【Purpose/significance】It is of great practical significance to clarify the laws of public opinion propagation in the Weibo com⁃

munity to achieve public opinion early warning.【Method/process】Construct a dynamic evolution model of SIRS public opinion propa⁃

gation, analyze the threshold characteristics of public opinion under different evolution states, and obtain sufficient conditions for pub⁃

lic opinion early warning.The gray prediction method and Markov model are used to predict the threshold future trend to realize the ear⁃

ly warning of public opinion risk, and classify based on the early warning results.Finally, use Python software to mine historical data of

hot cases in the Weibo community,and perform model fitting to test its validity.【Result/conclusion】The results show that: the propor⁃

tion of infected netizens has different evolutionary trends according to different threshold characteristics. Netizens are naturally re⁃

moved based on the diminishing value of information transmission, and government departments can reduce the trust coefficient

through rumors and other methods, so that the public opinion gradually disappear.【Innovation/limitation】Construct a SIRS public opin⁃

ion propagation model under linear infection conditions and introduce natural removal rates, in order to grasp the law of emotional in⁃

fection of netizens in the Weibo community to achieve preliminary data prediction and classification of public opinion levels, and then

establish a sound public opinion early warning mechanism. However, there are still some shortcomings in the current research, that is,

the types of netizens' emotions need to be further expanded.