A Hybrid Method Based on CRITIC Method and Machine Learning Models for Effective Fake News Detection in Thai Language

Abstract

Fake news has emerged as a pervasive issue within the modern information ecosystem, leading to widespread dissemination of misinformation and erosion of trust. This paper introduces a novel hybrid approach for effectively detecting fake news in the Thai language by combining the CRITIC method with multiple machine learning models. The initial step involves collecting Thai-language fake news data from websites. Subsequently, the data undergoes a preprocessing phase. In the second step, the preprocessed data is used for validation through three basic machine learning models, namely, Naive Bayes, Decision Tree, and K-Nearest Neighbors. In the third step, the accuracy results from these three models are employed to calculate the significance weights of each model using the CRITIC method. In the final step, predictions are recalculated using the proposed method. The proposed method achieves an 83.37% accuracy, surpassing Naive Bayes (80.83%), Decision Tree (80.37%), and K-Nearest Neighbors (75.75%). This indicates a significant enhancement in performance, with the proposed method outperforming the established models by up to 7.62%.  Consequently, the proposed method can enhance the performance of fake news detection in Thai language by utilizing an ensemble of the original models. A significant advantage of this approach is its simplicity coupled with high efficacy. It is postulated that this method can be adapted for detecting fake news in other languages as well.