Detection of Fake News on Twitter Regarding COVID-19: An Analysis of Machine Learning Algorithms with n-gram Modeling

Abstract

Fake news is fabricated information that notably impacts our social lives. The massive propagation of fake news by humans or robots severely impacts society and individuals. After the massive increase in the reach of social media platforms, the spread of fake news is unavoidable. Automatic detection of fake news will increasingly reduce the spread of misinformation on digital media platforms. As a contribution to solving this issue, this study recommends a better machine learning algorithm for detecting digital fake news by using a different set of extracted features, namely, regional features and text n-gram. This study uses various machine learning algorithms such as Support Vector Machine (SVM), logistic regression, decision tree, random forest, KNN classifier, MultinomialNB, Passive Aggressive, and Gradient Boost are analyzed with the efficient features for content-based text analysis. Among all the other algorithms, SVM produced outstanding outcomes with an average accuracy of 99.13% and the highest accuracy of 99.3% on the COVID-19 FNIR Dataset