Robust fake news detection

Abstract

Recently, news consumption using online news portals has increased exponentially due to several reasons, such as low cost and easy accessibility. However, such online platforms inadvertently also become the cause of spreading false information across the web, as we have seen during the recent COVID-19 pandemic. They are being misused quite frequently as a medium to disseminate misinformation and hoaxes. Such malpractices call for a robust automatic fake news detection system that can keep us at bay from such misinformation and scams. This thesis pro-poses a robust fake news detection system, named PaGE (based on Paraphrasing, Grammar, andEmbedding), leveraging the tools for paraphrasing, grammar-checking, and word-embedding. In this project, I try to unearth such tools' potential in jointly detecting a news article's authenticity. Notably, I leverage Spinbot (for paraphrasing), Grammarly (for grammar-checking), andGloVe (for word-embedding) tools for accomplishing this. While Spinbot and Grammarly aretext-oriented tools, GloVe is just a word-oriented tool. Since the proposed system's input is a text, not a word, the proposed approach involves two novel GloVe-based tools that act like text-oriented tools. While one uses the codebook-based approach, another uses the spatial-pyramid-pooling-based approach. Using all these tools, I extract novel features that yield state-of-the-art results on three fake news datasets after combining them with some essential features. More importantly, it empirically shows that the proposed method is more robust than existing ones through cross-domain analysis, multi-domain analysis, and learning curve experiments. This thesis also focuses on the importance of Grammarly features in identifying legit news articles.