Sentiment Orientation Classification Method and Its Application
Xiaoqin Wu, Changjian Guo · 2019
The traditional text categorization method, in the word segmentation stage, focuses on the vocabulary discrimination ability, ignoring the people's viewpoints and tendencies from the text, resulting in limited performance in terms of classification accuracy. Because the user's expression has strong correlation with his own emotions, using the scores and comment data of Douban movies and TV dramas provided by Datatang as the corpus, the feature word extraction methods such as chi-square test and information gain are adopted. Combining naive Bayes, N-Gram and support vector machine text classification methods, construct an emotional orientation classification model and experiment with the actual movie forum review data. The results show that the improved classification method can achieve higher classification accuracy, which makes it suitable for a large number of Chinese text emotion classification application systems.