Online Product Review Classification

Abstract

Reviews are the most important part that people look upon while purchasing a product online. The problem in existing system is sometimes the user review and the rating mismatch each other. This happens when the user forgets to update either review or rating when the user updates the review or while providing a new rating for the product, the user might randomly put some wrong rating or undesired rating.The main aim of this project is to tell the user whether a product is good or bad based on the reviews provided by other users and to provide a better rating for the product by analyzing sentiment. To classify the reviews into good or bad, the system uses two machine learning algorithms KNN and Naïve Bayes classification algorithms and to stem the review porter stemmer algorithm is used and to compute new rating system uses rule-based extraction method. K Nearest Neighbor will select the nearest neighbor class to the test review and classify the review into two classes that is either class = good or class = bad, whereas the Naïve Bayes algorithm uses a probabilistic approach to classify the product into good or bad by selecting the highest probability class label.