Stylometry-based Fake News Classification Using Text Mining Techniques

In recent years, fake news detection has become an important issue in our daily life. As information increases, many studies are devoted to automating fake news detection recently. Therefore, this study applies text mining technique to explore the issue of fake news from the aspect of writing style. We train the model by extracting four categories of style features, then construct a classifier that can automatically assess the authenticity of news articles. In addition, this study also evaluates the combinations of various style features and observe their differences in classification effectiveness. Meanwhile, this study compares four popular supervised learning classification techniques, namely Support Vector Machine, Random Forest, AdaBoost and XGBoost, to provide a most suitable classification method.