Sentiment Analysis of Bengali Comments on Social Media Using Natural Language Processing and Machine Learning

Abstract

Social media is a great way to share your thoughts and opinions with others, but some people on social media use language that is dangerous and disrespectful. Researchers are interested in understanding the emotions and sentiment behind this data, and Bangla Natural Language Processing has become a popular research field because of its diverse applications. Every day, a lot of user-generated Bangla data is created on social media, online news portals, and other websites. Sentiment analysis is a research area that is used to identify sentiments in texts. This paper also provides a comparative evaluation by examining implementations of different types of machine learning algorithms. Instead of previous efforts, we use NLP to process the data. By collecting more comments from social media and processing them through natural language processing, we have tried to extract the sentiments of those comments through machine learning algorithms. Here we use Random Forest, Logistic Regression, Naive Bayes, KNN, SVM, Linear SVC, and AdaBoost algorithms. Based on the results of all proposed classifier implementations, Support Vector Machine (SVM) techniques acquire the highest precision of 80% when compared to the other classifications.