Rating Prediction of Product Reviews of Bangla Language using Machine Learning Algorithms

Abstract

The only way to provide feedback about a product is through reviews. When a new shopper proceeds to an online shop to purchase a product but does not have adequate time to study the reviews provided by other shoppers to get an opinion about the product, the shopper determines whether to buy the product or not on the number rating. Through reviews, shoppers can acquaint everyone about the product's quality and the manufacturer can enhance their products and business by interpreting that review. However, manufacturers demand a number review more than a text review for business analysis. This paper represents a machine learning based model for predicting the number rating from written text for Bangla product review. This study performs on a dataset collected manually from Daraz.com.bd, a Bangladeshi leading e-commerce shop. We have implemented Support Vector Machine (SVM), Random Forest, XGBoost, and Logistic Regression with Term Frequency-Inverse Document Frequency (TF-IDF) Vectorizer on our collected dataset and record all the performance metrics like accuracy, precision, recall and f1-score. From these above four algorithms, SVM showed more outstanding results than others in terms of performance metrics. SVM achieved 90% accuracy on the applied dataset. The other SVM performance metrics are 0.90, 0.92, and 0.91 for precision, recall and f1-score respectively.