Enhancing Sentiment Classification Performance Using Bi-Tagged Phrases

Sentiment analysis research mainly aims to determine the orientation of an opinionated stretch of text into positive or negative polarity. The key motivation of sentiment analysis is getting to know what consumers think about products and services by analyzing their opinions on online portals, blogs, discussion boards, reviews etc. The main objective of this paper is to incorporate the information of POS-based sentiment-rich phrases in a machine-learning algorithm that determines the semantic orientation of a given text. In this paper, bi-tagged phrases are used as features in combination with unigram features for sentiment classification. Joint feature vectors of unigrams and bi-tagged phrases have high dimensions consisting of noisy and irrelevant features. Therefore, a feature selection method is used to select only relevant features from the feature vector. Experimental results show that the combination of prominent unigrams and bi-tagged phrases outperforms other features for sentiment classification in a movie review dataset.