A majority-based learning system for detecting misinformation

Combating misinformation is both a multifaceted problem and a pressing societal concern. In response, we propose a user-centric system founded on the majority vote model, offering flexibility and synergy in integrating established machine-learning methods or classifiers such as SVM, MLP, LSTM, RF, and XGB. Computational experiments demonstrate promising results in implementing our proposed system to identify text-based fake news, advertorials, and plagiarised information in social media. The dataset employed in these experiments is primarily sourced from volunteer contributors and fact-checking websites. The result evaluation indicators encompass balanced accuracy and F1 score. Overall, this study introduces a significant and autonomous countermeasure to address misinformation.