Fake News Detection Using Machine Learning: A Systematic Review

The rapid proliferation of misinformation and fabricated content across digital platforms constitutes one of the most pressing challenges of the information age. Fake news not only distorts public discourse but also influences elections, financial markets, and public health outcomes. This review paper presents a comprehensive survey of machine learning (ML) and deep learning (DL) approaches proposed for automated fake news detection. We systematically examine techniques spanning classical models — Naïve Bayes, Support Vector Machines, Logistic Regression, and Random Forest — through to contemporary transformer-based architectures such as BERT, RoBERTa, and GPT variants. Additionally, we review graph-based propagation methods, multi-modal fusion approaches, and cross-lingual detection strategies. Key benchmark datasets including LIAR, FakeNewsNet, BuzzFeed, and ISOT are described alongside standard evaluation metrics. We discuss prevailing challenges such as adversarial attacks, class imbalance, domain shift, and the scarcity of labelled data, and outline promising future research directions including explainability, federated learning, and real-time detection pipelines.

Keywords— Fake News Detection, Machine Learning, Deep Learning, Natural Language Processing, BERT, Misinformation, Social Media, Text Classification, Transformer Models