QDAontology – abordagem para o desenvolvimento de ontologias em e-science: um estudo de caso em biologia

Abstract

The use of ontologies in computational systems has become more important and spread out. However, as an area in constant evolution, it does not exist yet a consolidated process model of Ontological Engineering for ontology construction. The use of a development process makes this task more impersonal, less complex and more systematic. The QDAontology - Quality Driven Approach for e-Science Ontologies, proposed in this research, was developed for e-Science ontologies. In this kind of applications, ontologies are, in general, developed by multidisciplinary teams, where knowledge experts belong to the domain area and developers are ontological engineers from the Computer Science. This approach is composed of stages, activities, participants, artifacts and quality characteristics. There are six stages: Specification, Conceptualization, Formalization, Implementation, Integration and Evolution. Each stage is composed of activities, in each activity artifacts are generated and participants are related to process stages and activities. Based on the artifact evolution, the development of the ontology occurs. At implementation stage, the ontology must be then integrated to other ontologies through mechanisms of ontological correspondence. The development process is centered in an evolutionary model, therefore, the cycles can be repeated at each evolution of the ontology. To validate the proposal, a case study in the biological domain was done. The case study had two cycles: during the first one the reengineering process of the CELO ontology was made and in the latter the ontology was expanded with the Human Disease sub domain. All the knowledge modeling process and the multidisciplinary team tasks are detailed and they were supported by experts of the Biology domain. For the ontological correspondence, the alignment mechanism was adopted and the A3O tool was built to generate the equivalent links and terms between ontologies described in OBO and OWL