Comparative Analysis of Machine Learning Models for Detection of Fake News: A Case Study of Covid-19

Abstract

During and after the Covid-19 pandemic, people rely heavily on the internet for information because of its easy accessibility. However, the spread of fake information through this medium has been fast-growing, especially during and after the pandemic. This study, therefore, aims to evaluate the performance of 5 machine learning models used in detecting Covid-19 fake news. The models were trained using the Covid-19 dataset gathered online. The dataset contains 7,262 real news and 9,727 fake news, totalling 16,989 news altogether. 80% of this dataset was used for training the models while 20% was used for testing them. The support vector machine (SVM) with 95%, 95%, 97% and 96% for the accuracy, precision, recall and F1-score respectively was the best classifier for detecting Covid-19 fake news and has shown a better performance than the other algorithms.

Keywords

• Covid-19
• Fake news detection
• Classification
• Machine Learning
• Accuracy