Abstract

The rapid spread of online fake news through some media platforms has increased over the last decade. Misinformation and disinformation of any kind is extensively propagated through social media platforms, some of the popular ones are Facebook and Twitter. With the present global pandemic ravaging the world and killing hundreds of thousands, getting fake news from these social media platforms can exacerbate the situation. Unfortunately, there has been a lot of misinformation and disinformation on COVID-19 virus implications of which has been disastrous for various people, countries, and economies. The right information is crucial in the fight against this pandemic and, in this age of data explosion, where TBs of data is generated every minute, near real time identification and tagging of misinformation is quintessential to minimize its consequences. In this paper, the authors use Natural Language Processing (NLP) based two-step approach to classify a tweet to be a potentially misinforming one or not. Firstly, COVID -19 tagged tweets were filtered based on the presence of keywords formulated from the list of common misinformation spread around the virus. Secondly, a deep neural network (RNN) trained on openly available real and fake news dataset was used to predict if the keyword filtered tweets were factual or misinformed.

This work is licensed under a Creative Commons Attribution 4.0 License.