Comparative analysis in fake news detection using machine learning techniques

Abstract

The rapid proliferation of misinformation on social media has made fake news detection a critical challenge in the digital era. Although recent deep learning-based methods have demonstrated high performance, classical and hybrid machine learning approaches remain highly relevant, particularly in resource-constrained environments. This study presents a comparative analysis of machine learning-based fake news detection approaches reported between 2020 and 2025 that achieve classification accuracies below 92%, with a focus on identifying their strengths and limitations. Building on this analysis, a hybrid classification framework combining Logistic Regression, Random Forest, and XGBoost is proposed. The proposed system achieves an accuracy of 96.96% on the experimental dataset. Furthermore, the factors contributing to the superior performance of the proposed approach are analyzed, its limitations are discussed, directions for future research are outlined, and strategies for mitigating the challenges of fake news detection are subsequently discussed.