Paper Title
Three-Stage Sentiment Analysis By Polarity Shift Detection, Elimination And Ensemble

The volume of user-generated text on the Web in the form of reviews, blogs, and social networks has grown dramatically in recent years. This was mirrored by an increasing interest, from both the academic and the business world, in the field of sentiment analysis, which aims to automatically extract sentiment from natural language text and can be broadly categorized into knowledge-based or statistics-based. Bag-of-words (BOW) is the popular way to model text in statistical machine learning methods in sentiment analysis. However, the achievement of BOW sometimes remains bounded due to some primitive dearth in handling the polarity shift problem. The greater findings were that out of the classification algorithms assessed it was that the Random forest classifier provide the much more high classification accurateness for this domain. From the assessing of this study it can be concluded that the proposed machine learning and natural language processing techniques are an impressive and real time approach for sentiment analysis The polarity shift problem is a major factor that affects classification performance of machine-learning-based sentiment analysis systems .Proposing a three-stage cascade model to address the polarity shift problem in the context of document-level sentiment classification. We first split each document into a set of sub sentences and build a hybrid model that employs rules and statistical methods to detect explicit and implicit polarity shifts, respectively. Secondly, proposing a polarity shift elimination method, to remove polarity shift in negations. Finally, we train base classifiers on training subsets divided by different types of polarity shifts, and use a weighted combination of the component classifiers for sentiment classification. On this basis, also proposing a dual training algorithm to make use of original and reversed training reviews in duality for learning a sentiment classifier, and a dual prediction algorithm to classify the test results by keeping in mind both of two phases of one result. An extended framework from polarity (positive-negative) classification to 3-class (positive-negative-neutral) classification has to be done, by taking the neutral reviews into consideration. Finally, a corpus-based approach is constructed for pseudo-antonym dictionary, which elimination of Dual Sentential Approch’s dependency on an external antonym dictionary for review reversion. Keywords- DSA, Polarity Shift, Random Forest, Knowledge-Based, Bag-Of-Words (BOW), PSDEE.