The project aims to facilitate access to ontologies by restructuring their concepts dynamically according to the actual application. Our approach is to create on top of the original ontology a virtual ontology whose structure is defined explicitly or learned implicitly.
With this proposal we submit a research project that follows a vein of research that has the potential to influence considerably how information will be made available and consumed in the near future. The basic technology, which our proposal is relying on, is likely to be the core technology used to create the next generation of the WWW.
The overall goal of this project is the development of a formal framework for repurposing ontologies. The motivation behind this proposal is twofold:
- Ontologies are rarely provided in convenient and meaningful way for an end user
- It is too costly to develop ontologies that correspond to every user's specific needs
The global goal is broken down into the following sub-tasks to be realized during this project.
- Formal specification of the basic knowledge representation system
- Implementation of the knowledge representation system
- Formal specification and validation of a repurposing language
- Integration of the repurposing language into the previously developed KR system
- Formal specification of a simple inference engine for the basic language as well as the repurposing language
- Integration of the inference engine into the KR system
- Validation through the implementation of a prototype application (TARMED)
The central part of the proposed project is the formal specification and validation of a repurposing language. We want this language to respect the following requirements:
- All repurposing operations have to leave the original ontology unmodified.
- All inferences made in the repurposed ontology have to remain sound and complete with respect to the original ontology.
- All inferences made in the repurposed ontology have to be sound and should be complete (if possible) with respect to the repurposed ontology.
- Mechanisms have to be provided to cascade operations realized on the instance level back into the original ontology.
We do not insist on completeness in order to make the repurposing operations more flexible. Furthermore, we want to ensure that a classification realized in a repurposed ontology can be mapped back to the original ontology. This is especially important as we consider classification as the main task to be carried out by our system.
Once we have established a general-purpose formal system for repurposing ontologies we want to verify and show the applicability of our system by creating at least one prototype application based on the TARMED ontology, as this ontology and it's usability will be of great importance in Switzerland over the next couple of years.
Personnes et institutions
Intelligent Browser: classification tool that uses ontological information for dynamically structuring query results.
Online prototype for classifying medical services with TARMED (no longer supported)
Online prototype for coding patient records with the ICD10 (in French only).
 T. Kurz and K. Stoffel, "An Advanced Stemming Algorithm for Creating Concept Signatures of Medical Terms", Proceedings of ES 2001, Research and Development in Intelligent Systems XVIII, pp. 273-281, Cambridge, December 2001.
 T. Kurz and K. Stoffel, "Going beyond Stemming: Creating Concept Signatures of Complex Medical Terms", Knowledge-Based Systems Journal, vol. 15, no. 5-6, Elsevier, July 2002, pp. 309-313.
 T. Kurz and K. Stoffel, "Enriching Ontologies for Improved Access to Web Documents", Proceedings of IADIS WWW/Internet 2002, Lisbon, November 2002, pp. 815-818.
 T. Kurz and K. Stoffel, "Web-Based Tools for Working with Health Care Ontologies in Switzerland", In Proceedings of ECDL Workshop on Digital Libraries in Healtcare, Trondheim, Norway, 2003.
 T. Kurz and K. Stoffel, "Web-Based Tools for Codification with Medical Ontologies in Switzerland", Proceedings of ECAI 2006, Riva del Garda, August 2006, pp. 642-646.