Performance Comparison of SVM, Naive Bayes, and Random Forest Models in Fake News Classification

The proliferation of fake news (hoaxes) in the digital era represents a significant challenge to public trust and social stability. The objective of this study is to evaluate the performance of three prominent machine learning algorithms, specifically Support Vector Machine (SVM), Naive Bayes, and Random Forest, in the classification of fake news. The dataset employed comprises validated examples of both authentic and fabricated news items. The research methods included text pre-processing, feature extraction using TF-IDF, model training, and performance evaluation using accuracy, precision, recall, and F1-score metrics. The experimental results demonstrated that SVM achieved perfect accuracy (100%), outperforming Naive Bayes (94%) and Random Forest (99%). Additionally, SVM exhibited the optimal performance in precision, recall, and F1-score metrics. This research provides empirical evidence that SVM is the most effective model for detecting fake news. The implication of this research is the potential application of SVM in automated systems to help reduce the spread of fake news on online platforms.