This project aims to analyses the fake news present in Roney L. de S. Santos' et al (2018) corpus, Fake.Br Corpus in order to identify the relation between the occurrence of positively and negatively polarized sentences (Scopim, 2011) and the credibility of these news. To that end, the news are going to be analyzed in search of opinion words, in order to facilitate the access and the formalization of the data for subsequent analysis. After the gathering of the data with the aid of the tool Unitex 3.1 (2016) statistical and linguistic tests will take place to verify if the hypothesis that the positive and negative polarity rate in sentences present in the body and title of the news can be related with the falsifiability of the news, separating the occurrences within four news groups to the end of trying to get a higher success rate in the identification of the fake news, which will be the last phase of the project. The obtained data in the process of analyzing the sentiment in these sentences may help future researches of methods of automatically recognizing fake news, being able to contribute for the linguistic and computational branch of NPL (Natural language processing).
News published in Agência FAPESP Newsletter about the scholarship: