Misinformation Detection on Social Media, a Big Data and Machine Learning Approach

Social media platforms are now essential means of disseminating information in the digital age. They are also used, though, to disseminate false information, which has the potential to seriously damage society. This study investigates the deliberate and inadvertent dissemination of false information on social media sites such as Facebook, Twitter, and TikTok. We examine content features, user engagement trends, and psychological elements that influence the spread of misleading information by utilizing a dataset collected from these platforms and compared to fact-checking websites like PolitiFact and Snopes. The study uses machine learning models, such as Logistic regression, Random Forest, SVM, LSTM, Graph Neural Networks (GNNs) for network-based disinformation tracking and BERT for content classification. Furthermore, we employ network analysis to track the spread of false information across user connections and investigate how platform interventions, such as warning labels, affect user involvement. According to the study, while some false content is intentionally disseminated, a sizable amount spreads due to user cognitive biases and a lack of critical analysis. Our results highlight the necessity of more advanced detection algorithms and platform policies in order to successfully stop the propagation of misinformation online.