User Review Classification and Star Rating Prediction by Sentimental Analysis and Machine Learning Classifiers

With the digital media explosion, in today’s increasing Internet usage, the data generated is wide and varied. Huge volumes of new data are injected daily into the Web for various prospects. Procedures like text mining and analysis are required to make the best use of this potential. User review analysis benefits us with the exact understanding of the user’s feedback toward the product. In this paper, we have proposed a unique approach by performing abstract-level sentimental analysis of user review by n-gram classification and POS tagging. This classification is then used as entropy for machine learning algorithm. This paper leverages upon the proposed methodology with promising outcomes and improved accuracy by evaluating the data with the help of two algorithms, MaxEnt model and Naive Bayes classifier, after analyzing few algorithms including SVM and random forest.