A Novel Approach for Classification of Online Product Reviews using various Machine Learning Techniques

Abstract

Sentiment Analysis is extensively used in different sectors like product analysis for identifying the customer needs. There are also examples where sentiment analysis concept is in action such in health care, Government sector, stock analysis. Customers share their genuine feedback about the products in online shopping sites. Also, it will be easy for the users to see the product review as well as the ratings given and then make a decision to buy the products. The purpose of sentiment classification is to analyze the written reviews of users and classify them into positive or negative opinions. It helps in identifying the issues with the product which in turn can be rectified. On the other hand, product reviews help customers to buy the product based on its review which would also help business owners to improvise. Text present in the product review is difficult to categorize sometimes. This can be achieved by using classification algorithms like Decision Tree, Random Forest, Naïve Bayes and Logistic Regression. There are multiple features available from which N-Gram and TF-IDF (Term Frequency-Inverse Document frequency) were used. With the results, it is evident that Random Forest performs better in TF-IDF and N-Gram approach for Electronics, Health and beauty and Clothing and accessories product type.