Reduced feature based sentiment analysis on movie reviews using key terms

Word of mouth communication plays an influential role in propelling or sinking movies. Proper analysis of movie reviews published in social media helps in efficient decision making. Sentiment analysis aids in automating this time consuming task. Identifying the relevant data with the elimination of plot in movie reviews enhance the performance of any sentiment analysis algorithm. We propose an entity recognition method in the preprocessing stage to eliminate the irrelevant information from the reviews. Sets of N-grams are extracted from the document corpus as key terms. Extracted key terms together with an external lexicon are used to identify the sentiment conveying features in the review and these features are used to train the classifiers. Performance analysis is done on various classifiers and it is found that the proposed method outperforms existing approaches on movie reviews.