A comparative multi-model approach to detecting fake news using machine learning

The fast dissemination of fake information on digital platforms causes major social effects, particularly in specific Domain, such as political debate, health news, and financial markets. False or inaccurate information has a tremendous negative influence on society, leading to confusion and uncertainty. Misunderstandings can escalate into riots, resulting in legal issues that harm society. Detecting elusive news is a challenging task due to limited benchmark datasets and rapid publication rates. The current research examines key features applicable to the domain, such as language style, network conversations, and emotional content to address the issue of detecting fake information within the domain. Machine learning (ML) algorithms show potential in spotting fake news content. As a result, this work emphasises on the usage of a machine learning-combined approach to automatically classify information. Multiple methods, including the naive Bayes (NB), Random Forest algorithm (RF), SVMs, and XGBoost, were evaluated for precision, recall, precision, and F1-score. These findings highlight the possibility of an approach for reducing the spread of misunderstandings online.