FAKE NEWS DETECTION AND CLASSIFICATION USING MACHINE LEARNING

Abstract

With over 70% of news consumed via social media, the spread of misinformation poses a significant challenge. This study introduces a semantic-based fake news detection model, FNIOnt, which analyzes news content by dividing it into fictitious categories and applying logical inference through an ontology framework. The processed content is then classified using machine learning models—Random Forest, Logistic Regression, and LSTM—with LSTM achieving the highest accuracy of 99%, outperforming traditional methods. As misinformation tactics evolve, including the use of half-truths and subtle manipulation, detection models must adapt accordingly. Deep learning approaches such as RNNs, CNNs, and Transformers demonstrate strong capabilities in understanding the semantic and syntactic structure of text, enabling more accurate classification of real versus fake news even under nuanced manipulation.