An Evolutionary Embedded Model Fake News Detector Using an Optimized Support Vectors Machines

This study presents an innovative approach to combat the rapid spread of fake news on social media. By combining the Salp Swarm Algorithm with a Support Vector Machine, the proposed model achieves improved prediction accuracy. The Popular World Twitter dataset, comprising 10 million tweets, is utilized for training and testing the models. Different metaheuristic algorithms, including MVO, GA, PSO, GOA, and SSA, are used to optimize the SVM and generate five models. The dataset is converted into word representation through various feature extraction techniques, resulting in eight processed datasets. Comparisons with other metaheuristic algorithms demonstrate the superiority of the proposed approach, as it achieves higher accuracy with a reduced set of the most relevant features.