Enabling ad-hoc reuse of private data repositories through schema extraction

Area of Science:

• Biomedical Informatics
• Data Science
• Computer Science

Background:

• Legal and ethical restrictions, like the EU General Data Protection Rules (GDPR), limit sharing sensitive data across organizations.
• The Personal Health Train initiative explores decentralized data utilization to bypass data transfer needs.
• Existing systems face challenges in accessing and integrating sensitive data stored in disparate repositories.

Purpose of the Study:

• To propose a configurable and automated approach for schema extraction and publishing.
• To enable ad-hoc SPARQL query formulation against RDF triple stores without direct data access.
• To facilitate secure query execution under data provider control.

Main Methods:

• Developed a configurable and automated system for schema extraction and publishing.
• Ensured compatibility with existing Semantic Web technologies.
• Enabled schema introspection-assisted authoring of SPARQL queries.

Main Results:

• Successfully derived a configurable amount of concise, task-relevant schema for four distinct datasets.
• Demonstrated the ability to formulate SPARQL queries against RDF triple stores without direct access to private data.
• Validated the approach for enabling schema introspection-assisted query authoring.

Conclusions:

• Automated schema extraction and publishing supports introspection-assisted query creation for data selection and integration.
• The system architecture enables data reuse from private repositories where shared schemas are infeasible.
• This approach advances the reuse of data from previously inaccessible sources, promoting data-driven methods in biomedicine.