COMPARISON OF AI TECHNIQUES FOR DETERMINING NEWS RELIABILITY

Resumo

DOI: 10.5281/zenodo.10734624

The internet provides a wider range of access to information, however, this comes with the burden of greater dissemination of fake news, enhanced through social networks, thus increasing misinformation (ANTONIO, 2023). This work sought to build, implement and compare artificial intelligence techniques to define the reliability of news published on the internet. It used trained algorithms to test the reliability of a piece of news and indicate whether it is true or false. By delimiting the Brazilian news scene and creating a web platform for user interaction, this work was able to evaluate the accuracy of the Recurrent Neural Network (RNN), Decision Tree, Randon Forest, K-Nearest Neighbors (KNN) and Logistic Regression methodologies in assessing the veracity of news. All showed considerable accuracy, with the RNN technique showing the best accuracy.