Comparison of Feature Selection Methods for Sentiment Analysis

Sentiment analysis is process of deriving the opinion or attitude expressed in input text. For the classification problem, feature selection aims to select features that are capable of discriminating samples that belong to different classes. This paper evaluates the performance of three feature selection methods (MI, CHI and ANOVA) combined with three machine learning based classification techniques (NB, SVM and KNN) for sentiment analysis on online movie reviews dataset. The paper shows that feature selection is important task for sentiment based classification.