Model for Bayesian Networks Conversion from Summary-Based Health Ontologies
Ontologies allow knowledge representation easily interpreted by both humans and computers. Bayesian networks (BNs) support uncertain reasoning, providing a way to treat uncertainty existent in the health area. The construction of BNs is difficult and arduous, involving extensive manual interaction. On the other hand, there is a huge number of ontologies available and describing health knowledge. These ontologies can be applied as sources for the creation of BNs, easing the difficulty observed in this task. This work presents a new model for the semi-automatic extraction of BNs from health ontologies using the ontology summarization as a basis. The main contribution of this work is the use of conversion rules, created from a critical analysis of the context of ontologies and an approach for ontologies summarization, which provides greater precision in the data conversion. The text describes the theoretical basis, the formulated hypotheses, the developed model and the evaluation experiment. The model was implemented to support a Pedagogical Simulator dedicated to foster health education, the Health Simulator project. Therefore it is a real case of generation of BNs for clinical cases and is integrated to a publisher of BNs. Through it, is possible to generate the knowledge structure of the Health Simulator project, and advance its clinical case simulator based on BNs. The results indicate good possibilities in the generation of BNs, with advances concerning the state of the art.