The use of social media pervades everyday life: social media users in 2021 have grown by more than 10%, reaching about 58.4% of the world population. Thus, social media platforms, such as Twitter or Facebook, are very powerful tools for conveying and disseminating information. In this situation, a growing concern has emerged about the possibility of using social media to spread misleading and harmful information, indicated more generally with the term disinformation.

Disinformation can impact several aspects of civil life, being able to affect people's behavior and inflaming a climate of mistrust in the population towards the democratic institutions resulting, in some cases, in protests or even violence. This is why spotting fake news quickly and accurately is nowadays a key problem.

Several proposals today use artificial intelligence to tackle the fake news detection problem. Some of them analyze the content of the news from a semantic point of view to reveal if it features some fake news common characteristics. However, such approaches work to a certain degree only, since they completely neglect the fact that the diffusion dynamic of fake news is very different from that of real news stories.

We start from the observation that during the spread of news stories, social media users may find posts with opposite or conflicting content and usually decide to share one of them based on their own judgment of its truthfulness. In this respect, DISTORT aims at delivering logic-based frameworks, equipped with explainable AI tools, for disinformation detection. In particular, our main objective will be to help social media users in giving a better sense to inconsistencies about news circulating on social media, thus helping in solving the problem of disinformation diffusion. The main contributions of DISTORT will be:

1. a logic-based framework for modeling and analyzing disinformation networks evolving over time with the aim of designing a framework based on a (variation of) some temporal logics to represent and reason about news stories diffusion patterns;

2. a querying methodology for handling data inconsistency in dynamic disinformation networks that enables to assign degrees of consistencies to queries regarding the validity of a certain news, so to allow spotting fake news diffusing in a network when data inconsistency is generated by conflicting or opposite facts regarding the same news;

3. an explanation framework for inconsistent disinformation network querying, which will aid the users to understand why a query has been answered in a specific way by considering the inconsistency-tolerant and reliability perspectives, thus helping in explaining why some news stories have been identified as fake.

DISTORT will impact the current research on fake news detection by providing tools able to counter the effects of misinformation diffusion, thus having a positive impact on the society in preventing disinformation campaigns.