Detection of Abusive Bengali Comments for Mixed Social Media Data Using Machine Learning
Sherin Sultana, Md Omur Faruk Redoy, Jabir Al Nahian, Abu Kaisar Mohammad Masum, Sheikh Abujar · Research Square · 2023
Abstract Most people around the world use social media in their daily lives as a result we can see that it has become an integral part of our lives. Moreover, many people use social media for their livelihood. Social media has a lot of influence on our life from different aspects. Although there are many positive aspects, the trend of negative comments on social media has become a serious problem these days. Through this study, we have detected bad comments made in the Bengali language on social media using machine learning algorithms and measured those performances. Although much work has been done on this issue in other languages, it is scarce in the Bengali language. We have used six different types of machine learning algorithms for this study and the algorithms used are Logistic Regression (LR), Multinomial Naive Bayes (MNB), Random Forest (RF), Support vector machine (SVM), K-Nearest Neighbor (KNN), Gradient Boosting (GB). Among all the algorithms used in this study, SVM gave the best results with an accuracy of 85.7%.