Advancements in Fake News Detection: A Comprehensive Machine Learning Approach Across Varied Datasets

Abstract

Fake news has become a major social problem in the current period, controlled by modern technology and the unrestricted flow of information across digital platforms. The deliberate spread of inaccurate or misleading information jeopardizes the public's ability to make educated decisions and seriously threatens the credibility of news sources. This study thoroughly examines the intricate terrain of identifying false news, utilizing state-of-the-art tools and creative approaches to tackle this crucial problem at the nexus of information sharing and technology. The study uses advanced machine learning (ML) models comprising multinomial Naive Bayes (MNB), linear support vector classifiers (SVC), random forests (RF), logistic regression (LR), gradient boosting (GB), decision trees (DT), and to discern and identify instances of fake news. The research shows remarkable performance using publicly available datasets, achieving 94% accuracy on the first dataset and 84% on the second. These results underscore the model's efficacy in reliably detecting fake news, thereby contributing substantially to the ongoing discourse on countering misinformation in the digital age. The research not only delves into the technical intricacies of employing diverse ML models but also emphasizes the broader societal implications of mitigating the impact of fake news on public discourse. The findings highlight the pressing need for proactive measures in developing robust systems capable of effectively identifying and addressing the propagation of false information. As technology evolves, the insights derived from this research serve as a foundation for advancing strategies to uphold the integrity of information sources and safeguard the public's ability to make well-informed decisions in an increasingly digitalized world.