SEDIMARK - An open source data space framework for a decentralised data and AI marketplace

This talk, presented by Tarek Elsaleh from the University of Surrey, explores SEDIMARK, an open source framework designed for decentralized data and AI marketplaces that align with European data space initiatives. The session covers the shift from centralized data platforms to federated architectures, emphasizing how this approach enables secure, interoperable data sharing while addressing data sovereignty, privacy, and vendor lock-in issues. Key elements discussed include the use of Eclipse Dataspace Connectors for data exchange, decentralized ledger technology for trust, and federated learning frameworks that facilitate AI model training without centralizing data. The architecture allows participants to operate as data providers, consumers, or marketplace operators using a modular toolbox, ensuring compliance with data minimization and privacy regulations through cryptographic verification and decentralized identity mechanisms. By enabling a modular, open source ecosystem, SEDIMARK aims to lower barriers for both research and industry while promoting collaboration and interoperability across European data spaces.