Fake news detection using social media data for Khasi language

With the advancement in technology, web technol-ogy in the form of social media is one of the main origins of information worldwide. Web technology has helped people to enhance their ability to know, learn, and gain knowledge about things around them. The benefits that technological advancement offers are boundless. However, apart from these, social media also has major issues related to problems and challenges concerning filtering out the right information from the wrong ones. The sources of information become highly unreliable at times, and it is difficult to differentiate and decipher real news or real information from fake ones. Cybercrime, through fraud mechanisms, is a pervasive menace permeating media technology every single day. Hence, this article reports an attempt to fake news detection in Khasi social media data. To execute this work, the data analyzed are extracted from different Internet platforms mainly from social media articles and posts. The dataset consists of fake news and also real news based on COVID-19, and also other forms of wrong information disseminated throughout the pandemic period. We have manually annotated the assembled Khasi news and the data set consists of 116 news data. We have used three machine learning techniques in our experiment, the Decision Tree, the Logistic Regression, and the Random Forest approach. We have observed in the experimental results that the Decision Tree-based approach yielded accurate results with an accuracy of 87%, whereas the Logistic Regression approach yielded an accuracy of 82% and the Random Forest approach yielded an accuracy of 75%.