UPFD: Fake News Detection with MLflow and Optuna
IT 299: Independent Honors Study, ISU
Rebuilt the User Preference-Aware Fake News Detection model from a published research paper. It combines graph neural networks with text encoders to classify news tweets as factual or non-factual, using both the tweet’s content and the history of the users who reposted it. Training, validation and hyperparameter tuning were automated with MLflow and Optuna.