Mineração multi-relacional: o algoritmo GFP-growth

Detalhes bibliográficos
Ano de defesa: 2006
Autor(a) principal: Pizzi, Luciene Cristina
Orientador(a): Vieira, Marina Teresa Pires lattes
Banca de defesa: Não Informado pela instituição
Tipo de documento: Dissertação
Tipo de acesso: Acesso aberto
Idioma: por
Instituição de defesa: Universidade Federal de São Carlos
Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação - PPGCC
Departamento: Não Informado pela instituição
País: BR
Palavras-chave em Português:
Área do conhecimento CNPq:
Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/332
Resumo: Data mining is the phase of the knowledge discovery in database process where an algorithm is applied to the available data, in order to prove a hypothesis or discover a still unknown pattern. The traditional data mining techniques can deal only with single tables; however it is interesting to look for patterns involving several related tables, aiming to analyze the existing relation between the entities present in one table and the data of the same entities present in another table. Depending on the relationship existing between these tables, applying a traditional algorithm to the joint table is not sufficient, as the joint table may contain duplicated attribute values which interfere in the analysis process of the generated rules. In order to solve this problem, this project adopts an approach which consists on looking for association rules mining the joint table. The adopted process considers the groups of tuples, where each group is formed by tuples of the same entity. Following this approach the GFP-Growth algorithm was developed, which is presented in this monograph along with its results and comparisons with other multi-relational algorithms.