A new combined data-preprocessing and machine learning approach for emotion-based fake news detection

The proliferation of social network platforms, especially post-COVID, has accelerated the spread of fake news, impacting politics, public health, and social stability. This widespread issue undermines trust, polarizes communities, and influences decision-making, highlighting the need for innovative solutions for automatic detection and mitigation. This paper introduces a new combined data-preprocessing and machine learning approach for emotion-based fake news detection. A key focus of the study is the emotional dimension of fake news, identified as a critical factor influencing its propagation on social networks. By analyzing emotional cues embedded in fake news content, the proposed method seeks to detect fake news based on its emotional characteristics. To validate this approach, two machine learning classification methods K-Nearest Neighbors (KNN) and Support Vector Machines (SVM) were tested, achieving accuracy rates of 65% and 70%, respectively. These findings demonstrate the potential of emotional analysis in enhancing fake news detection frameworks and suggest avenues for further refinement and broader application.