Uma Metodologia para Mineração de Regras de Associação Usando Ontologias para Integração de Dados Estruturados e Não-Estruturados

• Ano de defesa: 2010
• Autor(a) principal: CAMILO, Cassio Oliveira
• Orientador(a): SILVA, João Carlos da
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Goiás
• Programa de Pós-Graduação: Mestrado em Ciência da Computação
• Departamento: Ciências Exatas e da Terra - Ciências da Computação
• País: BR
• Palavras-chave em Português: Mineração de Dados; Mineração de Texto; Recuperação de Informação; Extração de Informação; Conceitos; Ontologia; Regras de Associação
• Palavras-chave em Inglês: Data mining; Text mining; Information Retrievel; Information Extraction; Concept; Ontology; Association Rules
• Área do conhecimento CNPq: CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::METODOLOGIA E TECNICAS DA COMPUTACAO::SISTEMAS DE INFORMACAO
• Link de acesso: http://repositorio.bc.ufg.br/tede/handle/tde/500

Data and text mining methods have been applied in several areas of knowledge with the purpose of extracting useful information from large data volumes. Among the various data mining methods reported by specialized literature, association rule mining has proved useful in producing understandable rules. However, one of its major problems is the significant amount of rules produced, which hampers the selection of the more relevant rules needed to reply to a query. This study proposes a method for mining data from structured and unstructured sources in order to generate association rules between the terms extracted. The process of mining data from unstructured sources is assisted by an ontology that maps knowledge from a specific domain. The result of such process is converted into structured data and combined with data from other structured sources. A combination of objective and subjective interest measures is used to filter the set of rules obtained, in addition to support and confidence model. To verify the feasibility of this method in real-life situations, it was applied to a database of police occurrence reports of a government institution, which included data stored in structured and unstructured sources.