Emotional anatomy of fake news

The emotional dimension of fake news plays a pivotal role in shaping how misinformation spreads and influences public perception. This study explores the emotional anatomy of fake news by re-annotating the widely used Linguistic Inquiry and Argument Reasoning (LIAR) dataset, creating the Emotion-Labeled Fake News Dataset (ELFND). By leveraging advanced Natural language processing (NLP) methods, with a focus on Valence Aware Dictionary and Sentiment Reasoner (VADER) sentiment analysis, the dataset categorizes news articles according to emotional tone conveyed. Ensemble machine learning models, Random Forest and Gradient Boosting, are employed to detect fake news by analyzing their emotional content. Furthermore, artificial intelligence (AI) explainability methods, including Local Interpretable Model-Agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP), are utilized to offer a deeper understanding of how the model generates predictions, shedding light on how specific emotions contribute to the classification of news as fake. The findings reveal that emotions are crucial in the structure of fake news, and the use of explainable AI (XAI) enhances transparency in misinformation detection, offering a clearer understanding of the emotional drivers behind fake news.