Periodic Reporting for period 1 - TRUSTEE (TRUST AND PRIVACY PRESERVING COMPUTING PLATFORM FOR CROSS-BORDER FEDERATION OF DATA)
Reporting period: 2022-07-01 to 2023-12-31
In our data-driven era, interdisciplinary and geographically dispersed data repositories are inevitable. However, their lack of adherence to existing data representation standards and absence from data federation initiatives renders them inaccessible to researchers. TRUSTEE aims to bring a green, secure, trustworthy, and privacy-aware framework that will aggregate various interdisciplinary data repositories, such as Healthcare,Education,Energy,Space,Automotive, Cross-border. TRUSTEE will provide a framework featuring homomorphic encryption for stored data. This framework will enable researchers to: i) Search and utilize data in the encrypted domain ii) Establish a unified and meaningful FAIR representation of data iii) Execute complex and context-aware queries using advanced ontologies iv) Process and analyze data through transparent and trustworthy ML workflows within an intuitive AI playground, enhancing AI explainability, interoperability, and re-usability with cutting-edge methods v) Ensure compliance with European privacy and ethical frameworks such as GDPR and PIA vi) Uphold privacy standards by implementing a homomorphic encryption layer for all data interactions and vii) Maintain accountability through a blockchain-based transaction recorder.TRUSTEE's fully encrypted solution will undergo validation across six distinct use cases supporting European data federation spaces and trans-national initiatives like GAIA-X. These use cases will demonstrate a multi-disciplinary, Pan-European federated FAIR and private data ecosystem.
All WPs are in line with initial planning, all deliverables have been submitted, all KPIs and Means of Verification & project objectives, as defined in DoA, have been achieved, none of the defined risks has materialized thus far, and all milestones have been realized. WP1 focused in setting up the methodologies for effective project coordination and technical management, adopted a proactive and closely monitoring approach with regard to quality assurance, risk assessment, objectives and KPIs monitoring (reported in the Project Handbook) and developed the initial Data Management Plan. Technical implementation has been initiated under WP2, by studying relevant SoTA with a focus on European Data Spaces, Pilot Use Cases, and stakeholders’ defined requirements that were translated into technical specifications for developing the initial version of the TRUSTEE architecture defining all related modules and components as well as an initial legal and socio-ethical framework for TRUSTEE. Technical implementation has been the subject of WP3 and WP4 where the initial sets of functionalities and communication interfaces for the twelve identified components of the TRUSTEE Platform, already defined, were implemented, reaching an initial working prototype and a first Proof-of-Concept demonstrator of the Platform. WP5 focused on the Pilot Campaign Plan for running the four pilot phases of the Pan-European Pilot Campaign alongside the Pilot Definition and Evaluation Frameworks. The 1st Dry Run Phase has been successfully completed, gathering preliminary end-user feedback and assessing a first set of technical specifications, KPIs and pilot specific objectives. WP6 has defined and reported the Communication and Dissemination Strategy for promoting the progress and outputs of TRUSTEE, within an active and structured network of stakeholders, ensuring communities’ engagement for broadening the project’s acceptance/endorsement and participation including scientific and non-scientific publications, participation of the consortium partners in conferences and important events. The 1st Advisory Board Meeting was held, receiving feedback from top experts in the field.
TRUSTEE has made significant progress in various architectural components, advancing the TRUSTEE Platform and Ecosystem. Early system integration has been completed, covering key architectural blocks. Key achievements for the twelve architectural components:
AI Models as a Service: Homomorphic-Encryption-powered Federated Learning; Proof-of-Concept implementation; Research ongoing for xAI within the HE domain
Homomorphic Enabled Data Fusion: Execute searches within homomorphic encrypted domains and reaching the objective of executing search queries in 2 different data domains simultaneously; Integrate and extract insights from diverse data sources; Novel encryption key management solutions; Semantic annotation of datasets studied across various domains
Authentication Manager: Explored combining Self-Sovereign Identity with eIDAS regulation for enhanced security
Data Privacy Impact Assessment: Utilized CNIL tool for robust data privacy impact assessments; User-centric design, integrating multidimensional assessments
Dashboard: Integrated Robotic Process Automation and Virtual Assistant for workflow automation; Developed intuitive UI design
Self-sovereign Identity Homomorphic Capable Framework: Integrated verifiable credentials using SSI and Homomorphic Encryption for trust in transactions
One-Stop-Shop and External Connectors: Investigated external connectors with GAIA-X, IDSA, and Copernicus for interoperability
Security and Trust Manager: Developed agreement ontologies for streamlined dataset transactions; Managed GDPR-based agreements for trust and compliance
Accountable Transactions Recorder: Permissioned ledgers for recording platform transactions; Accountability and compliance with GDPR through private data collection
Data Lake: Machine computable logic, data federation, adherence to FAIR principles, and compliance with EOSC standards; Semantic interoperability through homomorphically encrypted datasets
Trustworthy AI Assessment Support Design Framework and GDPR Compliance check Module: A framework for assessing AI trustworthiness and GDPR compliance; Ensured fairness, transparency, explainability, & robustness of AI models
Knowledge Repository: leveraging past knowledge, minimizing redundancy, maximizing efficiency and sustainability; contributing to the “green” aspect of the TRUSTEE Platform; Enhanced functionalities using InterPlanetary File System for content integrity and security
AI Models as a Service: Homomorphic-Encryption-powered Federated Learning; Proof-of-Concept implementation; Research ongoing for xAI within the HE domain
Homomorphic Enabled Data Fusion: Execute searches within homomorphic encrypted domains and reaching the objective of executing search queries in 2 different data domains simultaneously; Integrate and extract insights from diverse data sources; Novel encryption key management solutions; Semantic annotation of datasets studied across various domains
Authentication Manager: Explored combining Self-Sovereign Identity with eIDAS regulation for enhanced security
Data Privacy Impact Assessment: Utilized CNIL tool for robust data privacy impact assessments; User-centric design, integrating multidimensional assessments
Dashboard: Integrated Robotic Process Automation and Virtual Assistant for workflow automation; Developed intuitive UI design
Self-sovereign Identity Homomorphic Capable Framework: Integrated verifiable credentials using SSI and Homomorphic Encryption for trust in transactions
One-Stop-Shop and External Connectors: Investigated external connectors with GAIA-X, IDSA, and Copernicus for interoperability
Security and Trust Manager: Developed agreement ontologies for streamlined dataset transactions; Managed GDPR-based agreements for trust and compliance
Accountable Transactions Recorder: Permissioned ledgers for recording platform transactions; Accountability and compliance with GDPR through private data collection
Data Lake: Machine computable logic, data federation, adherence to FAIR principles, and compliance with EOSC standards; Semantic interoperability through homomorphically encrypted datasets
Trustworthy AI Assessment Support Design Framework and GDPR Compliance check Module: A framework for assessing AI trustworthiness and GDPR compliance; Ensured fairness, transparency, explainability, & robustness of AI models
Knowledge Repository: leveraging past knowledge, minimizing redundancy, maximizing efficiency and sustainability; contributing to the “green” aspect of the TRUSTEE Platform; Enhanced functionalities using InterPlanetary File System for content integrity and security