Abstract: 【Purpose/significance】Taking four representative negative public security emergencies in China in recent two years
(2018-2019) as examples, this paper clusters their microblog comments, finds out the factors that affect the negative emotion tendency
of microblog users, and provides suggestions for the government to deal with public opinion response.【Method/process】Combined
with social network analysis and LDA, we extract the key elements of the comment text, get the comment attribution dimension, and
then analyze the emotional inclination of each dimension through the emotional analysis software.【Result/conclusion】The results
show that microblog users mainly comment on public security emergencies from seven aspects: event subject, event analysis, event
handling, social relations, news media, empathy and personal experience. Among them, event analysis, event handling, event subject
and social relations are the main influencing factors of microblog users' negative emotion tendency. Accordingly, this paper puts for⁃
ward corresponding suggestions on public opinion persuasion.【Innovation/limitation】Based on attribution theory, this paper innova⁃
tively puts forward the attribution dimension system that affects the emotional orientation of microblog users, but the number of public
opinion events is limited and no more fine-grained sentiment classification analysis is carried out.