Hybrid model based sentiment classification of Chinese micro-blog

摘要

Through analysis and study of emotional characteristics in Chinese micro-blog, such as Sina Weibo, this paper proposed a multidimensional sentiment classification method based on micro-blog emoticon by dividing micro-blog into 7 types of emotions categories: happiness, fondness, sorrow, anger, fear, detestation and surprise. We used predefined micro-blog emoticon sets to initial screen large-scale unmarked data, and automatically labeled them, then used this emotional corpus as training set to train the emotion classifier, which divided micro-blog data into multiple emotion categories. The experimental results show that accuracy rate of using unigram model for each class can reach 63.7%. And the adoption of different feature selection methods for Support Vector Machines and Naive Bias classifier experiment, by which the obtained accuracy rate and recall rate has reached higher than 71%.