A Machine Learning Perspective on Fake News Detection: A Comparison of Leading Technqiues

Abstract

The exponential growth of social media has yielded several advantages, but it has also brought about a major challenge in the form of “fake news”, which has become a substantial hindrance to journalism, freedom of expression, and democracy at large. The purpose of this study was to examine the current AI techniques employed for detecting fake news, determine their limitations, and compare them with the latest models. The performance of memory-based and Ensemble methods (LSTM, Bi-LSTM, BERT, Distilled BERT, XGBoost, and AdaBoost) was compared with traditional methods, and the impact of ensemble learning was evaluated. The study aimed to identify appropriate models for fake news detection in order to facilitate a secure and reliable environment for information sharing on social media and ultimately counteract the spread of false information.