Comparison of the Performance of Machine Learning Methods with Intuitionistic Fuzzy Decision Making for Effective Fake News Detection

In this paper, the success of machine learning models applied to two datasets consisting of fake news was examined. The performance of these methods was measured using Accuracy, Precision, Recall, and F1-Score. Evaluation metrics for all models were calculated using TF-IDF and N-gram TF-IDF for Dataset 1 and Dataset 2, respectively. In continuation of the study, a decision-making mechanism was created to measure the success of machine learning methods. The success of these models was compared by creating a decision-making mechanism using intuitionistic fuzzy sets. The PROMETHEE method was used here. In the first stage of the study, the dataset samples evaluated were expressed using classical sets, while in the second stage, the success of the models according to the metrics was expressed using intuitionistic fuzzy values. This was to minimize the uncertainty in the model success results. The results obtained in the first stage of the study were evaluated in the second stage using a decision-making mechanism. In this mechanism, machine learning models represent the alternatives, while classification metrics represent the criteria. When evaluating machine learning models, experts provide subjective opinions based on each classification metric to determine the successful model.