ENHANCING GENERALIZATION IN FAKE NEWS DETECTION: A COMPARATIVE EVALUATION OF NAIVE BAYES AND RANDOM FOREST APPROACHES
Saurabh Jaiswal, Mr. Vivek Rai · 2024
In our digitally interconnected era, the rampant spread of fake news stands as a formidable challenge, jeopardizing informed decision-making, eroding public trust, and undermining democratic processes. This research addresses this pressing issue by introducing an ensemble machine learning model designed for the classification of fake news. As the dissemination of news becomes increasingly challenging amid the explosion of online information, the study delves into the application of Naive Bayes Classification models with bag of words features, and an additional model employing tf-idf instead of BOW with a Random Forest classifier. The investigation reveals that the model using tf-idf outperforms others, marking a significant advancement in fake news detection. Notably, this model demonstrates superior performance, achieving an accuracy of 89% and F1 scores of 0.89 for both fake and true news. The emphasis on tf-idf showcases its effectiveness in capturing the essence of news articles while considering term frequency and document frequency.