Product review sentiment analysis by text vectorization and machine learning

Abstract

In Bangladesh, online marketing and e-commerce businesses have prospered in the age of Internet technology. Online shopping has taken over as the primary method of shopping during periods when people are restricted because of the COVID-19 pandemic because it is the safest option. The proliferation of online vendors of goods and services enhances people's lives, but it also calls into question the caliber of such offerings. Because of this, it is simple to con new customers who make purchases online. Our objective is to create a system that analyzes customer reviews of online sales using word2vec machine learning techniques and outputs the percentage of favorable to negative reviews. About 6,000 reviews and opinions regarding the product have been gathered by us. With a maximum accuracy of 99.81% and a maximum score of 100%, sentiment analysis, KNN, which are decision trees, a support vector machine (the SVM), random forest analysis, while logistic regression, among others, were utilized as classification techniques that outperformed all other approaches.