A Comparative Approach to Detecting COVID-19 Fake News through Machine Learning Models

Identifying fake news has become an increasingly challenging task in recent years, with the proliferation of digital media and the ease of spreading misinformation. The problem has only become more complex with the global pandemic situation, as false information about COVID-19 can have serious consequences for public health and safety. Fortunately, the same technological advancements that have made it easier to spread fake news have also enabled potential solutions to this problem. In this work, we aimed to test and evaluate approaches for automatically classifying fake news. We focused specifically on fake news related to COVID-19, given its widespread impact on public health and the urgency of addressing misinformation in this area. To do this, we trained and evaluated several machine learning models using a dataset of news articles labeled as either "fake" or "real." Our goal was to identify the most accurate and effective model for detecting COVID-19 related fake news. After testing several models, we found that an SVM classifier performed the best, achieving an accuracy of 93.83%. We also conducted an analysis of each model's performance, examining factors such as feature selection and model complexity that may have influenced their results.