Detection of Unreliable Medical Articles on Thai Websites

Abstract

Fake news have exerted terrible impact on the Thai society for a long time, especially fake health and medical news: unreliable news from social media have threatened people's mind and physical health. In this research, we investigated various methods for solving the problem of getting fake news on health and medical issues in social media. Then, we proposed to detect unreliable medical articles existed on Thai websites based on a machine learning. We collected samples of 297 reliable and 235 unreliable articles from 7 websites and analyzed the differences between them. Then, we selected 20 features that affected the reliability or unreliability of the articles and used machine learning to classify the articles according to those features. Experimental results show that XGBoost methods were the most effective at 90.60% accuracy.