Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Schema extraction for privacy preserving processing of sensitive data

: Gleim, L.C.; Rezaul Karim, M.; Zimmermann, L.; Kohlbacher, O.; Stenzhorn, H.; Decker, S.; Beyan, O.

Volltext ()

Debattista, J.:
4th Workshop on Managing the Evolution and Preservation of the Data Web (MEPDaW), the 2nd Workshop on Semantic Web solutions for large-scale biomedical data analytics (SeWeBMeDA), and the Workshop on Semantic Web of Things for Industry 4.0 (SWeTI) 2018. Proceedings. Online resource : Co-located with 15th European Semantic Web Conference, ESWC 2018; Heraklion, Crete, Greece, June 3rd-7th, 2018
Heraklion, 2018 (CEUR Workshop Proceedings 2112)
European Semantic Web Conference (ESWC) <15, 2018, Heraklion>
Workshop on Managing the Evolution and Preservation of the Data Web (MEPDaW) <4, 2018, Heraklion>
Workshop on Semantic Web Solutions for Large-Scale Biomedical Data Analytics (SeWeBMeDA) <2, 2018, Heraklion>
Workshop on Semantic Web of Things for Industry 4.0 (SWeTI) <2018, Heraklion>
Konferenzbeitrag, Elektronische Publikation
Fraunhofer FIT ()

Sharing privacy sensitive data across organizational boundaries is commonly not a viable option due to the legal and ethical restrictions. Regulations such as the EU General Data Protection Rules impose strict requirements concerning the protection of personal data. Therefore new approaches are emerging to utilize data right in their original repositories without giving direct access to third parties, such as the Personal Health Train initiative [16]. Circumventing limitations of previous systems, this paper proposes an automated schema extraction approach compatible with existing Semantic Web-based technologies. The extracted schema enables ad-hoc query formulation against privacy sensitive data sources without requiring data access, and successive execution of that request in a secure enclave under the data provider's control. The developed approach permit us to extract structural information from non-uniformed resources and merge it into a single schema to preserve t he privacy of each data source. Initial experiments show that our approach overcomes the reliance of previous approaches on agreeing upon shared schema and encoding a priori in favor of more flexible schema extraction and introspection.