Comparação entre modelos com diferentes abordagens para classificação de fake news.

Abstract

Currently, false news is increasingly in evidence. Such news can be defined as intentionally propagated non-truthful information. With the large use of social networks as a source of information, it becomes necessary to have greater control and detection of Fake News, efficiently and quickly. Thus, this work seeks to use algorithms already consolidated in the machine learning area - Naive Bayes, XGBoost and BERT - to create false news detection models, comparing the results obtained in each model with works previously carried out in the area that have the best result until now.