A Comprehensive Dataset of Brazilian Fact-Checking Stories

Abstract

In recent years, digital platforms have become a powerful means for large scale information diffusion world-wide, particularly in Brazil. Understanding key aspects driving the misinformation diffusion process is of paramountimportance to the design and implementation of new tools to automatically detect misinformation content. In this scenario, fact-checking performed by high credibility agencies provide rich labeled data, which is fundamental to build tools capable of detecting and mitigating the effects of misinformation. This paper opens a novel dataset, referred to as FactCenter, to the research community, containing fact-check instances collected from 6 different Brazilian fact-checking agencies. This dataset has 11 647 fact-check instances, covering several topics and domains. We present an initial analysis of the data collected, enriched by data from Facebook, which demonstrates the potential of our repository for future studies.