Published Paper Details:

The accelerating proliferation of misleading and fabricated information across digital ecosystems has emerged as a critical threat to societal stability and informed decision-making. Deceptive news content can distort public perception, undermine democratic institutions, and generate social unrest. To address this growing challenge, this study presents an intelligent fake news detection framework based on a feature-driven, optimized Multi-class Support Vector Machine (MSVM) classification model. The proposed system performs comprehensive feature extraction from textual data, incorporating lexical indicators, syntactic patterns, semantic embeddings, and contextual attributes to capture deeper linguistic and structural characteristics of news articles. Advanced feature selection and optimization techniques are applied to enhance predictive capability while minimizing computational overhead. Unlike conventional binary classifiers, the MSVM model enables multi-category classification, distinguishing among real, fake, satire, partially true, and misleading content for more granular and informative analysis. This multi- dimensional approach improves classification robustness, scalability, and adaptability across diverse datasets and multilingual environments. Experimental validation demonstrates improved detection accuracy and reduced misclassification rates compared to traditional models. By integrating optimized feature engineering with supervised machine learning, the proposed framework offers a reliable and efficient solution for real-time deployment in social media monitoring systems, automated fact-checking platforms, and digital news verification applications, thereby contributing significantly to advanced misinformation detection research.