|
Abstract
|
|
|---|---|
| Traditional Knowledge Discovery in Databases (KDD) approaches departs from a centralized repository to begin data analysis. However, biomedical environments generally require distributed frameworks, since the information is dispersed through different organizations. Structural and semantic database heterogeneities should be managed in this scenario. Semantic issues have been tackled in the last years using ontologies in database integration and further KDD phases. This paper presents a model of ontology-based integration and preprocessing to access distributed biomedical sources. Ontologies are used to: (i) build virtual repositories representing data sources and (ii) as a structure to store the required information to preprocess the instances of a database. These ontologies drive the automatic data transformation, carried out each time the user query the system. Breast cancer datasets have been used to test the improvement of our approach in data mining results. | |
|
International
|
Si |
|
Congress
|
4th pHealth Conference 2007 |
|
|
960 |
|
Place
|
Porto Carras, Chalkidiki (Greece) |
|
Reviewers
|
Si |
|
ISBN/ISSN
|
|
|
|
|
|
Start Date
|
20/06/2007 |
|
End Date
|
22/06/2007 |
|
From page
|
|
|
To page
|
|
|
|
|