A multilingual fake news detection on COVID-19 Infodemic in Malaysia using language-independent (Lang-IND) features

Abstract

The study proposed a fake news detection model using Lang-IND features by implementing a two-stage evolutionary approach to generate and optimize the best mathematical equation to detect fake news. Results from the first stage shows that the equation from GP scores F1-score of 83.23% on Fake.my-COVID19 dataset. After fine-tuning stage, the model performance increases the F1-score to 85.52%. The proposed two-stage evolutionary approach performance result is higher as compared to the traditional machine learning algorithms.