Classification of Actual and Fake News in Pandemic

Abstract

Fighting in a misinformation era in addition to the COVID-19 pandemic is a difficult task for many superpower nations. On social media, fake news and rumors move like any actual news , and most of the time many people are misguided with that information. Believing in rumors can have serious consequences for both the individual and society. This has made it worse in the event of a pandemic at such a level that it has caused chaos between people and nations. To address this issue, this paper uses COVID-19 to compile a dataset of actual and fraudulent news, posts , and articles from Twitter, Facebook, Reddit , and other social media handles. In this paper a binary classification task is performed (actual vs fake) and compared three machine learning baselines - Decision Tree, Bidirectional-Long Short Term Memory and Support Vector Machine on the annotated dataset. The binary classification of the dataset gave us a brief understanding of how distorted news differs from actual news.