Social media are nowadays one of the main news sources for millions of people around the world. Yet, using social media for news consumption comes with the danger of exposure to `fake news¿ containing poorly checked or even intentionally false information. In this context, there is the need of a systematic study of the phenomenon, that aims at reducing fake news spreading in social media. Automatically detecting fake news is a challenging task, which we plan to address with modern techniques of deep learning.
Our aim is to study in a deeper detail the difference in information spreading in the case of source-checked news and fake news. We aim to model the propagation of fake news in comparison to the source-checked news analyzing the anisotropic behavior of the first using spectral techniques.