Automatic Fake News Detection in Online Media
Abstract:
Nowadays, online media has become one of the principal sources of news consumption. People and news organizations use online media such as news websites and social media to stay informed and distribute information. These sites offer different advantages such as reduced costs, adaptability, easy access, and quick distribution of information. Nevertheless, the extensive dissemination of information on these sites has led to the existence of fake news. Fake news contains intentionally fabricated false information or alterations of real events. This type of news aims generate biased ideas and beliefs in society. To stop the spread of fake news for the benefit of society and new organizations, there is a current trend to develop systems that automatically detect them, since doing a manual fact-checking is practically impossible. In this thesis, we present a study for the problem of automated fake news detection in online media using the textual content from the news. We collected different datasets that contain news extracted from variety of news websites and social networks to solve this task. The datasets we collected are: COVID Fake News, LIAR, FakeNewsNet, ISOT, The Fake News Corpus Spanish and Fake Costa Rica News. In addition, the verification of the veracity of the news in the datasets was in charge of organizations such as PolitiFact, Gossip Cop, Verificado, etc. With these datasets, we conducted a series of experiments with different machine learning and deep learning models using a set of superficial and deep features extracted from the text. To evaluate our models, we use a set of metrics to measure their performance.