SAFE: SPARQL Federation over RDF Data Cubes with Access Control

Area of Science:

• Healthcare and Life Sciences
• Data Analytics
• Linked Open Data

Background:

• Federated query engines face privacy challenges with sensitive data.
• Healthcare and Life Sciences (HCLS) data requires patient anonymity and data ownership controls.
• RDF Data Cube Vocabulary is standard for HCLS statistical data, but lacks optimized federated query solutions.

Purpose of the Study:

• To develop a query federation engine for sensitive statistical RDF data cubes.
• To enable policy-aware access, preserving patient anonymity and data ownership.
• To optimize federated query performance over distributed HCLS datasets.

Main Methods:

• Introduced SAFE, a query federation engine for statistical RDF data cubes.
• Implemented policy-aware access control, coupling it with source selection, user profiles, and access rights.
• Developed a join-aware source selection method and an indexing system storing only predicates and endpoints to preserve anonymity.

Main Results:

• SAFE enables policy-aware access to sensitive statistical RDF data cubes in a distributed setting.
• The join-aware source selection avoids irrelevant and unauthorized data source requests.
• SAFE's indexing system results in lower index generation time and size for faster updates.

Conclusions:

• SAFE provides granular graph-level access control for distributed clinical RDF data cubes.
• Demonstrated efficient reduction in source selection and query execution time compared to general-purpose engines.
• Validated SAFE's performance using real-world datasets from clinical organizations.