Validata: Enhancing Fake News Detection Using Recursive Feature Selection and Machine Learning in Web-Based Applications

The emergence of news through social media and other mainstream media, e.g., e-newspapers has made it easier to claim genuinity. It has been considered that fake news that is spread this way could hamper democracy, public health, and social harmony. In many instances, the fast spreading of illegitimate news has resulted in mass hysteria, riots, and widespread panic among the masses of people. And the presence of social media platforms, or the Internet in general has made news easily accessible to everyone. Not only does this connect people with genuine stories, but it has become a haven to easily spread rumors; be it political, or related to health, finance, or technology. Modern problems require modern solutions. So, we have devised the “Validata” model. The “Validata” model extracts two hundred unique features by using Recursive Feature Selection (RFS). These features are then input into Machine Learning (ML) classifiers. The experiment was run on three publicly available datasets from Kaggle, Reuters, and McIntire. Our proposed model has yielded higher accuracy as compared to other innovative models.