Transfer Learning Model for Disrupting Misinformation During a COVID-19 Pandemic

Raju, R.; Bhandari, S.; Mohamud, S. A.; Ceesay, E. N.

In 2020, the COVID-19 pandemic changed the world significantly, and it is critical to have reliable online information about this virus. However, disinformation can have a negative effect on public opinion and can put the lives of millions in danger by ignoring the crucial precautions. People worldwide post their ideas about the coronavirus every second and create a rich source of information. In this work, we introduce an advanced natural language processing model to classify public opinion about the virus, which can help health organizations to take immediate actions to stop the spread of the virus by removing misinformation from online platforms. We introduce a new model with high classification accuracy to extract deep contextual information from online coronavirus comments based on main COVID-19 topics and use a robust model for sentiment classification based on more than twenty different datasets to detect the tweet's text, which contains misinformation. The new model can generate reports about tweets that contain misinformation to the states requiring emergency precautions to stop the virus's spread by removing the detected comments from their online platforms. Also, The new model can detect misinformation and prevent fake news by increasing public awareness about COVID-19.