Fake News in Spanish: Towards the Building of a Corpus Based on Twitter

Abstract

Nowadays, we can use various techniques to detect fake news that depends mainly on supervised models that trained a corpus of true and fake news in English. However, these models are not suited to classify news in Spanish. On the other hand, very few is known about the provenance of news in existing corpora. This work presents a strategy to create a Corpus of news in Spanish for fake new detection purposes. We propose Twitter as a mediator to find relevant sources of where to take the news. As such, we report the work in progress performed to characterize Twitter as a means to retrieve the news.