FAIR Data and Distributed Analytics

Research & Development Areas

We advance the practical implementation of FAIR principles (Findable, Accessible, Interoperable, Reusable) through comprehensive data infrastructures and knowledge management systems. Our work encompasses FAIR Digital Objects (FDOs), research data infrastructure development, knowledge graphs, semantic technologies, persistent identifiers (PIDs), and data sovereignty frameworks.

We develop advanced methods for transforming unstructured data into structured, machine-actionable formats using state-of-the-art AI techniques. This includes medical information extraction from clinical documents, scholarly document processing, multimodal data integration, and domain-specific text processing for healthcare, legal, and scientific domains.

We create sophisticated distributed analytics platforms and AI-driven decision support systems for real-world challenges. Our machine learning and analytics solutions span healthcare diagnostics, cultural heritage preservation, industrial applications, and behavioral analysis, with a focus on privacy-preserving and federated approaches.

Technical Expertise & Applications

Our research and development leverages cutting-edge technologies including Large Language Models (LLMs), distributed analytics, federated learning, knowledge graph construction, and reinforcement learning. We work to enable FAIR Data Principles and advance established standards including healthcare data standards (HL7, FHIR), semantic web standards, and privacy-preserving computation frameworks.

We create practical solutions that bridge the gap between cutting-edge AI research and real-world applications. We focus on enhancing data accessibility, improving healthcare outcomes, preserving cultural heritage, advancing scientific research, and enabling data-driven innovation through secure, compliant data sharing platforms.

We work closely with healthcare institutions, cultural heritage organizations, research institutions, industry partners, and government agencies. Our interdisciplinary approach ensures that our technical solutions address real-world needs while maintaining the highest standards of ethics, privacy, and security.

FAIR Digital Objects Manager (FDO Manager)

The FDO Manager as part of the NFDI4DS represents one of the initial implementations of the FDO in practice, which adheres to the FDO specifications while offering a robust minimum viable solution. The manager ensures the necessary metadata and a persistent identifier accompany each artefact. To fully comply with the FDO specifications, the manager ensures that the metadata is temporarily stored in a dedicated registry managed by Fraunhofer FIT, separate from the digital objects themselves. Furthermore, the FDO Manager will handle the recording and storage of FDO records in a distinct registry, which is a key requirement in the FDO specifications. More

PADME

Personal Health Train (PHT) is a novel approach, that aims to establish a distributed data analytics infrastructure enabling the (re)use of distributed healthcare data. At the same time, data owners stay in control of their data. The main principle of the PHT is that data remains in its original location, and analytical tasks visit data sources and execute the tasks. The PHT provides a distributed, flexible approach to using data in a network of participants, incorporating the FAIR principles. PADME is a PHT implementation developed by Fraunhofer in collaboration with RWTH Uni, Cologne University Hospital and Leipzig Uni. Distributed Analytics (DA) has been introduced to overcome the challenges of accessing and performing data analysis on privacy-sensitive data. The main principle of DA is that the analysis task is brought to the data instead of bringing data to a centralised location to run the data analysis algorithms

Our study is part of German MII and GoFAIR initiatives.