Using idiomatic expression for Chinese sentiment analysis

Abstract

The rising of social media services causes Sentiment Analysis is becoming a critical research issue in recent years. Using the prediction of future opinion trends to adjust current business strategies is the most valuable application of emotion discovery. The major drawback of the previously proposed language model, CCLM, used to produce a larger number of terms in corpus causes the emotion classifier slowdown emotion judgement by spending too much time on searching. This research tried to improve the precision and performance of the emotion classifier, the idiomatic expression of human habit has been proposed to promote a novel sentence-level emotion classifier by using the Combined Jieba Customary Language Model (CJCLM). In the experiment, all Chinese sentences are retrieved from Plurk platform to present the improvement of the execution time by CJCLM.