The present invention relates to a
civil aviation public opinion emotion
analysis method. The method comprises the steps of performing searching, preprocessing and word segmentation on microblog texts comprising keywords about
civil aviation security public opinions in
the internet; building a dictionary; grading each microblog text to obtain a microblog text emotion value; performing objective and subjective judgment on the microblog text according to the emotion value, so as to obtain a
threat value of the microblog text for
civil aviation security; and determining the
threat level of the opinion in the microblog text for civil
aviation security according to the
threat value. The emotion value of the microblog text is determined by combining text
semantics and microblog emoticons, the limitation of the dictionary and
semantics rules are overcome, and the emotion value is determined more accurately. Due to full use of the characteristic of the microblog text, the
threat level is determined more reasonably. The method differs from a
machine learning method and is not trained with a large scale of tagged data, thereby being more suitable for
processing of real-
time data flow.