Unraveling the Dynamics of Fake News and Misinformation on Twitter: A Comprehensive Exploration

Abstract

In this digital age, social media plays a pivotal role in instant information propagation across the world. However, the same characteristics also make the social media the major contributor to spreading misinformation. This paper contributes to the ongoing efforts in combating misinformation on social media (specially Twitter) platforms by adopting a holistic approach that combines network analysis, Machine Learning (ML), and psychological insights. It offers a detailed analysis of the Twitter network by exploring the dynamics of misinformation on Twitter. The resharing user and tweet network, as well as tweet and user features like tweet sentiment, user engagements, and their implications on misinformation propagation have been considered and evaluated. Extreme Gradient Boosting (XGBoost) achieved Mean Squared Error (MSE) of 26.81 and R2 score of 0.97 showing high predictive accuracy with relatively small prediction errors for verifying tweet authenticity. This can support the analysis of finding influential users and tweets in order to track and foster and improved understand of misinformation propagation. As the digital landscape continues to evolve, the strategies developed through this research will be instrumental in enhancing the integrity and reliability of information on social media, thereby fostering a healthier digital public discourse.