Empirical Evaluation of Machine Learning Classification Algorithms for Detecting COVID-19 Fake News

Abstract

Humans have been fighting the Covid19 pandemic since it started, not just to protect their wellbeing but also to counteract the news and rumors that have been spreading about it. Rumors and false allegations can be almost as dangerous as the virus, as they affect people's mental health and increase their stress levels. To address this problem, several machine learning techniques could be used to detect fake news. In this paper, four different machine learning algorithms are compared according to their ability to detect fake news, including Naive Bayes, Decision Tree, Support Vector Machines, and Logistic Regression. A dataset of annotated news is used in the experiments. The experimental results show that Naïve Bayes outperforms other algorithms in terms of accuracy, precision, recall, and F1 score.