In nature, a clint is a solid block of limestone, shaped by centuries of erosion and separated from others by deep fissures known as grikes. Together, clints and grikes form a limestone pavement, a fragmented yet interconnected landscape, where every block is distinct but still part of a larger, evolving system. The CLINT project draws inspiration from this natural formation to describe today’s digital information ecosystem. Just like limestone pavements, the online world is made up of blocks of knowledge, communities, and narratives, separated by gaps and fractures where misinformation, polarisation, and misunderstanding can grow. By studying the patterns of connection and separation across social networks, the “clints and grikes” of our digital landscape, the project seeks to understand how information flows, how divides emerge, and how knowledge can be reconnected. CLINT explores the geometry of information, developing data-driven models and predictive tools to map, interpret, and strengthen the fragile connections that hold our information environment together.
The project tackles the challenge of online disinformation through an integrated approach that combines large-scale data collection, advanced predictive modelling, media literacy training, and early-warning analytics. By bridging computational and social science perspectives, it develops new datasets, analytical tools, and educational resources to better understand, anticipate, and counter the spread of misleading information across social networks.
A large-scale, multimodal dataset (text, image, audio) was created by collecting, verifying, and annotating information from social media and domain-specific repositories. This curated corpus provides a solid foundation for studying information flows and misinformation mechanisms.
The project analysed large-scale interaction networks to uncover how online communities form and evolve around specific topics. By studying patterns such as polarisation, echo chambers, and community segmentation, it revealed how social structures influence the diffusion and reinforcement of disinformation across digital platforms.
Novel computational models were developed to predict opinion dynamics, integrating different neural networks approaches like Graph Neural Networks (GNNs) to predict links between users and videos in social networks.
By analysing propagation patterns and user interactions, the project identified early-warning signals on rumour models. These models enable near-real-time forecasting of information trends across online communities and particular analyse astroturfing or other kind of organised campaigns.
A web-based platform and an autonomous online course were created to promote media literacy. These resources translate the project’s scientific findings into accessible educational materials for schools, cultural institutions, and the general public.
The CLINT Toolkit combines all developed components — data repository, predictive models, visual analytics dashboard, and media literacy materials — into an interactive demonstrator that supports both research and practical interventions on information dynamics.