Framework para suporte à evolução de ontologias biomédicas

Abstract

Biomedical ontologies grow continuously and therefore need to be refined throughout their life cycle. In Biomedical ontologies, small changes can cause unexpected effects on large parts of the ontology, which can be explained by their high complexity and the fact that they are, in general, larger than the ontologies of other domains. The process of refinement in order for ontologies to be adapted to the changes occurring in their knowledge sources is called evolution of ontologies. In this context, in the present work a conceptual framework was created to assist Biomedical ontology developers during their evolution process. Its main objective is to offer validated suggestions of activities and tools to support the evolution of Biomedical ontologies. The defined framework is divided into 5 comprehensive and well-defined phases called Evolution Planning, Implementation of Changes, Detection of Changes, Treatment of Inconsistencies and Audit of Changes. The framework was validated through a case study where it was applied to an ontology on Chronic Kidney Disease submitted to evolution in three perspectives based on three reasons for evolution: the change of the focus of the ontology, changes to reflect changes in their domain guidelines and changes related to ontology reference terminology. The consistency of the ontology used in the case study was verified through the application of SPARQL queries based on Competency Issues predefined by the ontology domain expert. The developed framework has as its main benefit the provision of resources to support the whole process of evolution, from the representation of the change to the verification of the conformity of the evolved ontology with the evolution plan. Based on the results of the experiments carried out, it is possible to conclude that the framework allowed to keep organized and understanding the evolutionary process of the ontology used. The organization and comprehensiveness provided by the framework are important because it facilitates the retrieval of information if there is a need to redo any changes. In addition to subsidizing the activities for which they are provided as inputs, the reports provided as a result of the activities that make up the subprocesses of the framework can be used as tools for predicting changes. This can be useful in the prior definition of strategies for applying changes and correcting errors identified during ontology validation.