Um modelo de engenharia do conhecimento baseado em ontologia e cálculo probabilístico para o apoio ao diagnóstico

• Ano de defesa: 2011
• Autor(a) principal: Lopes, Luiz Fernando
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Catarina; Curitiba; Programa de Pós- Graduação em Engenharia e Gestão do Conhecimento
• Palavras-chave em Português: Gestão do conhecimento; Teoria do conhecimento; Ontologia; Processo decisório; Probabilidades; Knowledge management; Knowledge, Theory of; Ontology; Decision making; Probabilities
• Link de acesso: http://repositorio.utfpr.edu.br/jspui/handle/1/330

The diagnosis, as knowledge-intensive task, is a complex process since there is a wide variety of elements and circumstances to be considered for a decision-making. Uncertainty generated by the subjectivity, vagueness and/or lack of updated information exist in almost all stages and interfere for the safety and efficacy in the outcome. The data and useful information, when collected and treated appropriately (technical), deriving from diagnosis accomplished and which remain latent (process), can become a valuable source of knowledge if associated with the experience and observation of the individual (human) who uses them. The goal of this research is to propose a model of Knowledge Engineering that allows the creation of new knowledge to support the diagnosis process. The methodologies, methods and techniques of Knowledge Engineering, used on this model to support the process are: CommonKADS, Ontology, Probabilistic Calculation and Discovery Systems Based on Literature. Through the integration of these elements, the proposed model is applied to a case study which allows evidence to be highlighted and analyzed through research literature as potential new knowledge. After the information of a new knowledge, involving the scientific community, the inference process is updated. To verify the consistency aspect of the model, it is sought the consensus of opinions in a group of experts using the Delphi method. The results show that the acceptance of the concepts, methods and techniques that comprise the model are above the minimum established for this study, and comments from the experts generated ideas to compose the final result of this work. It is concluded, therefore, that through this research, the proposed model meets the requirements for the generation of new knowledge, and contributes to the improvement of the diagnostic test.