Sentimental feature selection for sentiment analysis of Chinese online reviews

Abstract

With the growing availability and popularity of online reviews, the sentiment analysis arises in response to the requirement of organizing useful information in speed. Feature selection directly affects the representation of online reviews and brings a lot of challenges to the domain of sentiment analysis. However, little attention has been paid to feature selection of Chinese online reviews so far. Therefore, we are motivated to explore the effects of feature selection on sentiment analysis of Chinese online reviews. Firstly, N-char-grams and N-POS-grams are selected as the potential sentimental features. Then, the improved Document Frequency method is used to select feature subsets, and the Boolean Weighting method is adopted to calculate feature weight. At last, experiments based on online reviews of mobile phone are conducted, and Chi-square test is carried out to test the significance of experimental results. The results suggest that sentiment analysis of Chinese online reviews obtains higher accuracy when taking 4-POS-grams as features. Besides that, low order N-char-grams can achieve a better performance than high order N-char-grams when taking N-char-grams as features. Furthermore, the improved document frequency achieves significant improvement in sentiment analysis of Chinese online reviews.