AI4Debunk

The project focuses on the creation of four human-centered interfaces:

• a web plug-in,
• a collaborative platform,
• a mobile application,
• and an augmented reality (AR) interface.

These tools are based on a powerful “debunking” API, designed to detect and challenge false information.

To develop the API, the project focuses on two main case studies:

• disinformation about the war in Ukraine;
• and climate change.

Through in-depth analysis with the input of sociologists, the project maps misinformation content in knowledge graphs, which are compared to existing databases through artificial intelligence techniques such as natural language processing (NLP).

Knowledge graphs are the basis for the development of debunking modules, capable of processing different types of content, such as text, images and videos.

The modules leverage advanced artificial intelligence and machine learning techniques, such as:

• repetitive neural networks (RNNs),
• convolutional neural networks (CNNs),
• models based on transformers.

Through these, an indicator (“Disinfoscore”) is produced that assesses the degree of misinformation of a content.

The API integrates this evaluation mechanism and supports the four interfaces:

• the web plug-in notifies in real time about the validity of content,
• the collaborative platform (Disinfopedia) allows the reporting of suspicious content for verification;
• the mobile application enables content control on the go,
• the AR interface offers insights into disinformation trends on social networks.