A Sentiment Analysis in the Field of Bengali Text: A Machine Learning Approach

Abstract

Now a-days, online marketing and e-commerce businesses in Bangladesh were thriving. Because it is the most secure way, online shopping has replaced traditional methods of buying after the COVID-19 epidemic. It reduces the amount of time required for businesses to launch their websites. More options for purchasing goods and services online are convenient and help consumers, but it also raises questions about reliability and safety. This makes it easy for unsuspecting new customers to fall victim to fraud while making purchases online. Our goal is to develop software that uses NLP to analyze customer reviews of online shops and provides a percentage breakdown of positive to negative feedback provided in Bangla (NLP). For the research, we compiled over 2003 user reviews and feedback items. We employed KNN, MULTI, RF, SGD, and SVC, as well as sentiment analysis as classification strategies. SVC achieved 85.7% accuracy, which was higher than any other approach.