A Text Mining Based on Refined Feature Selection to Predict Sentimental Review

This paper proposed an additional feature set and reduced the data dimension by SVD and PCA in order to increase accuracy and decrease executing time in text mining. The contribution of this study has: (i) proposed a preprocessing algorithm for sentiment classification, (ii) refined a feature set by adding adjective and adverb feature for sentiment classification, and (iii) utilized SVD then PCA to reduce data dimension for manager identified the sentimental labels. The experimental results show that the proposed model can obtain the better accuracy and the additional features make the better performance. Moreover, the dimension reduction can reduce the executing time effectively.