Detection of Bangla Fake News using MNB and SVM Classifier

Abstract

Fake or fraudulent news is coming into existence in large numbers for various political and commercial causes, which has become common in internet community. People can easily get tainted by any of these fraudulent news for their falsified words that have tremendous effects on the offline community. Therefore interest has increased in research on this topic. Notable work on the identification of false news in English texts as well as other languages except a few in Bangla Language has been carried out. Our work demonstrates the experimental investigation of detecting fake news from Bangla social media, as this area still requires a lot of concentrate. We have utilized two supervised machine learning techniques throughout this research study, Support Vector Machine (SVM) and Multinomial Naive Bayes (MNB) classifiers to recognize Bangla fake news. Term Frequency - Inverse Document Frequency Vectorizer and CountVectorizer has been used as feature extraction. Our suggested system recognizes fake news according to polarity of the related post. Eventually, our research suggests SVM with linear kernel gives a 96.64 percent accuracy overperforming MNB with a 93.32 percent accuracy.