Directed Research Project Details
Ontology-based Data Integration and Modeling
| Author: | Paea LePendu |
|---|---|
| Date: | August 29, 2006 |
| Time: | 10:00 |
| Location: | 220 Deschutes |
| Committee: | Dejing Dou (Chair) Chris Wilson Michal Young |
Abstract
The enormous amount of data currently available on the World Wide Web, the deep-web (online databases), the emerging Semantic Web, and knowledge bases presents the challenging task of effectively integrating these sources of information. Even more difficult is ensuring that differences in semantics between each resource are properly preserved, if not exploited. This problem is known in both the Database Systems and Artificial Intelligence community as information integration.
In previous work an "inferential data integration" framework that ensures the correctness of data integration was applied to ontologies on the Semantic Web using the OntoEngine inference engine. The research presented here extends these tools to include relational database sources by exploiting obvious similarities between schemas and ontologies and embedding them in an SQL Wrapper. Now that conjunctive query predicates can be translated to SQL on-the-fly, OntoEngine can effectively answer queries using data from either relational databases or Semantic Web documents. Some experimental results on the performance of this system are presented. Finally, a closer look at the differences between schemas and ontologies reveals interesting questions and potential future work on an ontology-based approach for model transformation.