| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 5.06 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Nesta dissertaĆ§Ć£o Ć© apresentada uma prova de conceito tendo como objetivo automatizar a adaptaĆ§Ć£o de um processo de descoberta de conhecimento previamente definido, a uma nova fonte de dados.
Um processo de descoberta de conhecimento pode usar uma ou mais fonte de dados e, na maioria das vezes necessita, de ajustes quando essas fontes sĆ£o trocadas por outras pertencentes ao mesmo domĆnio, uma vez que as fontes de dados anteriores e as novas raramente possuem o mesmo esquema de dados associado. Esses ajustes por vezes fazem com que o analista dispense mais tempo a reformular a configuraĆ§Ć£o do processo implementado anteriormente.
As ferramentas de descoberta de conhecimento existentes permitem que seja executado o processo, mas em nenhuma delas Ć© possĆvel efetuar a troca da fonte de dados e, de uma forma automatizada, criar e ajustar os componentes para que possa ser usado o mesmo processo de descoberta de conhecimento, excetuando os casos em que os esquemas de ambas sĆ£o rigorosamente iguais. Esta limitaĆ§Ć£o levou Ć criaĆ§Ć£o de uma soluĆ§Ć£o que permite complementar as aƧƵes das ferramentas de descoberta de conhecimento e, desta forma, simplificar a atuaĆ§Ć£o do analista.
A soluĆ§Ć£o desenvolvida foi implementada em C# e permite adaptar o esquema da nova fonte de dados ao processo de descoberta de conhecimento previamente definido, com base na fonte anterior, criando e utilizando os componentes especificados anteriormente. Esta nova soluĆ§Ć£o permite uma maior rapidez no processo de reutilizaĆ§Ć£o de um processo de descoberta de conhecimento em novas fontes de dados pertencentes ao mesmo domĆnio.
This thesis presents a proof of concept aiming to automate the adaptation of a process of knowledge discovery previously defined to the new data source. The Knowledge discovery process can use one or more data sources, and most of time they require adjustments when the data source is exchanged for another one with different schema, but related with the same domain. These adjustments sometimes take a lot of time from analyst due to the need to reshape the process configuration created for the old data source. The existing tools allow the process to run, but none of them can be used to exchange data source and readjust the process by an automatic method, allowing the creation of new components and redefine the previously implemented components, in order to get the same structure of previous data source. To overcome this limitation, a new solution was developed which not only makes the KDD tools more user friendly but also allows the simplification of the actions needed by analyst. The developed solution was implemented in C# and allows the new data source to be adjusted to the target schema, based on the previous data source, creating and reusing used previous existing components. This solution allows for greater speed in the adjustment process of the new data source within the same domain.
This thesis presents a proof of concept aiming to automate the adaptation of a process of knowledge discovery previously defined to the new data source. The Knowledge discovery process can use one or more data sources, and most of time they require adjustments when the data source is exchanged for another one with different schema, but related with the same domain. These adjustments sometimes take a lot of time from analyst due to the need to reshape the process configuration created for the old data source. The existing tools allow the process to run, but none of them can be used to exchange data source and readjust the process by an automatic method, allowing the creation of new components and redefine the previously implemented components, in order to get the same structure of previous data source. To overcome this limitation, a new solution was developed which not only makes the KDD tools more user friendly but also allows the simplification of the actions needed by analyst. The developed solution was implemented in C# and allows the new data source to be adjusted to the target schema, based on the previous data source, creating and reusing used previous existing components. This solution allows for greater speed in the adjustment process of the new data source within the same domain.
Description
Keywords
Data Mining Descoberta de Conhecimento C# ClemScript IBM SPSS Modeler WEKA