Date of Award

1-1-2003

Degree Type

Thesis

Degree Name

Doctor of Philosophy

School

Computer and Information Science

Faculty

Faculty of Computing, Health and Science

First Advisor

Dr Chaiyaporn Chirathamjaree

Second Advisor

Associate Professor Vichit Lorchirachoonkul

Abstract

The information required for decision making by executives in organizations is normally scattered across disparate data sources including databases and legacy systems. To gain a competitive advantage, it is extremely important for executives to be able to obtain one unique view of information in an accurate and timely manner. To do this, it is necessary to interoperate multiple data sources, which differ structurally and semantically. Particular problems occur when applying traditional integration approaches, for example, the global schema needs to be recreated when the component schema has been modified. This research investigates the following heterogeneities between heterogeneous data sources: Data Model Heterogeneities, Schematic Heterogeneities and Semantic Heterogeneities. The problems of existing integration approaches are reviewed and solved by introducing and designing a new integration approach to logically interoperate heterogeneous data sources and to resolve three previously classified heterogeneities. The research attempts to reduce the complexity of the integration process by maximising the degree of automation. Mediation and wrapping techniques are employed in this research. The Mediated Data Integration (MeDint) architecture has been introduced to integrate heterogeneous data sources. Three major elements, the MeDint Mediator, wrappers, and the Mediated Data Model (MDM) play important roles in the integration of heterogeneous data sources. The MeDint Mediator acts as an intermediate layer transforming queries to sub-queries, resolving conflicts, and consolidating conflict-resolved results. Wrappers serve as translators between the MeDint Mediator and data sources. Both the mediator and wrappers arc well-supported by MDM, a semantically-rich data model which can describe or represent heterogeneous data schematically and semantically. Some organisational information systems have been tested and evaluated using the MeDint architecture. The results have addressed all the research questions regarding the interoperability of heterogeneous data sources. In addition, the results also confirm that the Me Dint architecture is able to provide integration that is transparent to users and that the schema evolution does not affect the integration.

Share

 
COinS