A Study on the Performance of Supervised Algorithms for Classification in Sentiment Analysis

Sentiment Analysis has gained a lot of popularity in the last few years. It is the processes of identifying the underlying sentiment in a comment, review or in a document. Sentiment of a document or sentence generally falls under three categories namely positive, neutral or negative. This paper has surveyed literature relating to supervised algorithms for classification in sentiment analysis. The contribution of the paper includes the extensive survey and classification of literature related to sentiment analysis based on classification algorithm used and application area. We have then performed a comparative study of five popular supervised classification algorithms namely Naive Bayes, Random Forest, Support Vector Machine (SVM), Decision Tree and K Nearest Neighbors (KNN) and evaluated the performance of each of these algorithms in sentiment analysis. The study has been performed on two different data sets to nullify the bias that might happen in the comparative study when focusing on a single data set. This paper has also proposed a novel method with advanced preprocessing(AP) which offers improved performance of about 8% over all the methods taken into study in the paper.