Uni-Logo
Artikelaktionen

Project GoodOD

Title: Good Ontology Design

Active Period: March 2010 until March 2013

Primary investigator: Prof. Dr. Stefan Schulz

Responsible Persons in Freiburg: Prof. Dr. Stefan Schulz, Dr. Martin Boeker

Responsible Person in Rostock: Dr. Luger Jansen

Involved Persons in Freiburg: Prof. Dr. Stefan Schulz, Dr. Martin Boeker, Djamila Raufie, Dr. Daniel Schober

Involved Persons in Rostock: Dr. Luger Jansen, Niels Grewe, Johannes Röhl

Funding: German Research Society (DFG)

 

QUESTIONS

The evaluation of ontologies has grown in recent years and become a major area of research. The subject of the evaluation was primarily ontology-based tools (such as ontology editors, inference engines) and then the ontology itself. Even in the narrower sense the term “ontology evaluation” is multilayered. Measurable quality criteria comprises the domain coverage, granularity, design, usability, content creation and content maintenance, design principles, formal decidability, consistency, and reference to standards such as Upper Ontologies and representation languages. Despite the existing criticisms in biology and medicine, still significant activities of the terminology development and ontology development are in progress. The major projects are OBO (The Open Biological and Biomedical Ontologies) and SNOMED CT. On the one hand, supported by the claim that thesauri are sufficient, and on the other hand, these projects gradually start to recite theoretical groundwork for the ontology as a common standard, so there is some convergence of approaches. Such convergence is because of their advantages of standardization, interoperability and the further development of such highly desirable resources. Moreover it throws new questions such as:

  • What should be represented by formal biomedical ontologies?
  • Where are the boundaries between ontology and knowledge representation?
  • Which formalisms or languages are appropriate?
  • How can this ontology development by philosophically sound Good-Practice Guidelines in combination with Ontology templates be supported?
  • What “upper-level” categories biomedical ontologies are appropriate?
  • Is a philosophical foundation a quality standard for ontologies?
  • How can the ontologies be evaluated? Are ontologies better evaluable with a qualitative than with a quantitative approach?
  • Can ontologies generally be evaluated in terms of correctness of the model?
  • Can an ontology-Guideline better facilitate ontology development or perhaps inter-ontology operability?

 

TARGET

In GoodOD project medical scientist, biologists, computer scientists and philosophers collaborate with each other. The goal of this project is the modeling of domain segments from Upper-Ontology with numerous examples. The development of guideline is to help the ontology developers for creation of ontologies. In this guideline, technical basics of ontologies, their development and means of Ontology Design Patterns (ODPs) are determined. Furthermore, it shall support ontology developers to classify the terms in the correct ontological categories and formulate axiomatic ontology classes denoted by these terms. The tasks in this guide will help to construct usable ontologies. By applying the guideline the developed ontology has to be logically consistent. To evaluate this, a group of test subjects is formed from research assistants (biologists and physicians). The group is introduced to the theory of biomedical terminologies and ontologies. The subjects need to model limited part of domains. In this case, these subjects are especially trained in, Protégé supported logical operators such as implication, equivalent, negation, conjunction and disjunction, to understand and to master the language scope of OWL-DL. After this training phase, the group is divided in to two subgroups consisting of physicians and biologists. Then, during the second training phase, the knowledge from the previous training of the group A subjects is further expanded and deepened with the help of many examples. In group B, however, the training will proceed strictly according to guideline. Following the two training phases, the modeling tasks are processed and evaluated according to the evaluation criteria of the guide.

 

PROJECT SUMMARY

The data explosion in healthcare requires more conceptual standardization. To this end, many terminology systems and classification systems have been developed. The various systems are described by mathematical logic in consideration of principles of formal ontology, e.g. SNOMET CT or the Open Biological Ontologies. In GoodOD project ontological principles for the formal representation of biomedical events, conditions, processes, dispositions and functions is developed. The modeling principles are presented with numerous examples from the area of medicine and biology. This guideline is intended to help ontology developers to develop useful ontologies.

Benutzerspezifische Werkzeuge