Abstract: Firstly, based on the pointwise mutual informationinformation retrieval (PMIIR) algorithm, we proposed Laplace smoothing
sentiment opinion (LSSO) algorithm, which automatically expanded the emotional dictionary and the emotional dictionary of emotive symbols. A certain scale and high quality emotional dictionary was obtained, including basic 
emotional dictionary, target emotional dictionary, online emotional dictionary, emotional dictionary of emotive symbols, negative dictionary, interrogative dictionary, degree adverb dictionary, conjunction dictionary. Then, the emotional value of text was calculated by refining the rules of text semantic analysis. The experimental results verified the effectiveness of the proposed method.

Key words: emotional text classification, emotional dictionary, pointwise mutual informationinformation retrieval (PMI-IR) algorithm, Laplace smoothingsentiment opinion (LS-SO) algorithm