Bangla Fake News Detection Using Machine Learning

Abstract

Knowledge on the other hand is very much potent and the effect of sharing knowledge is highly significant but the sad thing is that sharing knowledge is now associated with share falsehood which has far-reaching consequences. Due to social media people are continuously exposed to fake news in the digital age even if they are not directly a consumer of the news. Social media and news sites are standard means of getting updated information, hence it is all too simple to spread fake news. Such news is packaged as credible but has wrong information on it, or is entirely fabricated. Secondly, the increase in the cases of lynching has been witnessed in recent time because of prevalence of fake news. Furthermore, all COVID-19 falsehoods have led to massive confusion and panic across the world. Still, there is a possibility to automate the fake news’ detection, but the major part of such systems is designed for English only. As the babies of the speak bengali around the world, that paper puts forward a novel specifically for the Bangla language to identify fake news. For the pre-processing and feature extraction part on our dataset we have employed several methods. Our insights derived from the performance analysis of the Passive Aggressive Classifier and Support Vector Machine both of which have 93% accuracy reveal that. 8% and 93. specifically, the method achieves a recognition accuracy of over 90%, and false negative and false positive rates of 5%, respectively, higher than other machine learning classifiers.