FAKE NEWS DETECTION SYSTEM WITH MACHINE LEARNING

Abstract

The exponential growth of digital information has introduced unprecedented challenges in differentiating authentic news from fabricated content. This comprehensive study proposes an advanced machine learning framework for automated fake news detection, integrating traditional supervised learning algorithms, deep neural networks, and cutting-edge transformer-based models. The research methodology involves extensive experimentation across multiple benchmark datasets, including LIAR, WELFake, ISOT, and custom-curated collections, evaluating 14 distinct algorithmic approaches through rigorous performance analysis. The investigation highlights significant performance variations across different model architectures, with ensemble methods attaining superior classification accuracy of 99.74% through the proposed FakeStack hybrid architecture. Traditional machine learning approaches demonstrate moderate effectiveness (74– 90% accuracy), while deep learning methodologies deliver robust performance (92–98% accuracy). Transformer-based models, particularly BERT, showcase exceptional contextual understanding capabilities with 99.37% accuracy, albeit with increased computational demands. The comprehensive analysis further explores feature engineering strategies, cross-domain generalization, computational efficiency assessments, and practical deployment considerations. The research contributes novel insights into optimal algorithmic selection for diverse operational scenarios, addressing critical challenges in scalability, interpretability, and real-time processing. A web-based prototype implementation demonstrates practical applicability through an intuitive user interface and comprehensive result visualization. The study’s findings confirm the effectiveness of hybrid approaches in combating digital misinformation while offering actionable recommendations for both practitioners and researchers. This work advances the state of the art in computational methods for ensuring information integrity, establishing a strong foundation for future research in automated fact-checking systems and digital content verification technologies