Sadness and Fear: Classification of Fake NewsSpreaders Content on Twitter

The vast amount of accurate and inaccurate information circulating on the internet requires computational methodologies to detect low-quality content. This kind of content often constitutes fake news, as in the PAN@ CLEF 2020 competition Profiling Fake News Spreaders on Twitter. This competition asks for systems that identify possible fake news spreaders on social media as a first step to prevent fake news from being propagated among online users. In this paper, the methodology used for this classification task is reported. Preprocessing of the data and the features extracted to classify fake news spreaders is explained. A regression-as-classification approach that enables the representation of being a fake news spreader as a gradable one is proposed. The performance (accuracy) on the training and the test set with the different sets of features is reported.