Uma abordagem semântica para seleção de atributos no processo de KDD

• Ano de defesa: 2010
• Autor(a) principal: Ribeiro, Lamark dos Santos
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraí­ba; BR; Informática; Programa de Pós Graduação em Informática; UFPB
• Palavras-chave em Português: Descoberta de Conhecimento em Banco de Dados; Ontologias; Seleção de Atributos; Knowledge Discovery in Databases; Ontologies; Attributes Selection; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/tede/6048

Currently, two issues of great importance for the computation are being used together in an increasingly apparent: a Knowledge Discovery in Databases (KDD) and Ontologies. By developing the ways in which data is stored, the amount of information available for analysis has increased exponentially, making it necessary techniques to analyze data and gain knowledge for different purposes. In this sense, the KDD process introduces stages that enable the discovery of useful knowledge, and new features that usually cannot be seen only by viewing the data in raw form. In a complementary field, the Knowledge Discovery can be benefited with Ontologies. These, in a sense, have the capacity to store the "knowledge" about certain areas. The knowledge that can be retrieved through inference classes, descriptions, properties and constraints. Phases existing in the process of knowledge discovery, the selection of attributes allows the area of analysis for data mining algorithms can be improved with attributes more relevant to the problem analyzed. But sometimes these screening methods do not eliminate the attributes satisfactorily, do allow a preliminary analysis on the area treated. To address this problem this paper proposes a system that uses ontologies to store the prior knowledge about a specific domain, enabling a semantic analysis previously not possible using conventional methodologies. Was elaborated an ontology, with reuse of various repositories of ontologies available on the Web, specific to the medical field with a possible common specifications in key areas of medicine. To introduce semantics in the selection of attributes is first performed the mapping between data base attributes and classes of the ontology. Done this mapping, the user can now select attributes by semantic categories, reducing the dimensionality of the data and view redundancies between semantically related attributes.