An Approach On Two - Fold Sentiment Analysis.

Sentiment classification is a fundamental task in sentiment analysis, with its aim to classify the sentiment (e.g., positive or negative) of a given text. Polarity shift is a kind of linguistic phenomenon which can reverse the sentiment polarity of the text. To model text in statistical machine learning approaches in sentiment analysis, Bag-of-words (BOW) is used. But the performance of BOW sometimes reduced due to some fundamental deficiencies in dealing the polarity shift problem. A Two-fold (dual) sentiment analysis model is used to address this problem for sentiment classification. A data expansion technique, which involves Text reversion and Label reversion, is used for creating a sentiment-reversed review for each training and test review. DSA framework consists of two parts such as Dual Training (DT) and Dual Prediction (DP). For learning a sentiment classifier by means of original and reversed training reviews in pairs, a dual training algorithm is used and along with, for classifying the test reviews by considering two sides of one review, a dual prediction algorithm is used. Feature selection and extraction is done with classifier training. A novel method of introducing a Probability based classifier in addition to the Naive Bayes classifier.