|
|
Information for:
| | |
|
|
|
|
|
|
|
| | |
Information about:
| | |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| | » | | » | | » | | |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| | |
|
| | |
|
|
|
Dr Denise de Vries
| Category: |
Research Staff |
|---|
| Role/s: |
Lecturer
Research Fellow
|
|---|
| Location: |
| IST347 Information Science and Technology Building |
| Map Reference: |
J15 (Building 57) |
| Nearest Carpark: |
15 |
|
|---|
| Contact: |
|
|---|
Research Interests: |
Dr Denise de Vries has, since the early 1980s, developed commercial complex database systems
for a variety of industry domains including local government, airline, public libraries,
para-medical, diet and nutrition, commercial cleaning, automotive parts manufacturing,
building and construction, inventory control, conference organisation and commercial photography.
Based on this industry experience she researched, as part of her doctoral degree, the
evolution of data in databases to develop methods and structures to manage changes to
information of a database, such as structural change, semantic change and constraint change.
Her research interests focus on complex domain modelling using more sophisticated methods,
such as mesodata modelling (see below), as well as other issues relevant to metadata modelling
related to data mining, knowledge management and schema evolution. Her current work involves
the development of a complex data mining system for the DSTO, the "Pathways to Wellbeing" project, developing an IT-based approach for managing complex health, housing and social inclusion issues within Aboriginal communities and an analysis of the data warehousing requirements for the Adelaide Football Club.
Mesodata: Engineering Domains for Attribute Evolution and Data Integration
The common view of relational data modelling and relational database structures (and as a result database languages) is to consider the specification of attributes, normally defined over a restricted set of domains, as part of table definition. While the user's requirements indicate that attributes need only be defined over relatively simple (normally DBMS-supplied) domains this is generally adequate. However, in more complex applications, domain structure becomes an important issue and even for some simpler applications, there is often advantage to be gained from utilising more sophisticated domains, such as concept graphs, hierarchies, intervals, and so on. In practice, this rarely occurs and where it does, design decisions often mean that implementations are inconsistent and not transferable. This research focuses on the utility of augmenting the information capacity of the attribute domain in a reusable and systematic manner.
Our approach is thus to retain the overall structure and operations of the relational model and to enhance the capability of the attribute domain to handle more advanced semantics. We do this by more clearly delineating between domain definition - mesodata - and schema definition - metadata. This approach can be considered as falling between the basic relational (minimalist) approach in which there are only a very limited number of common domains, and the object-relational model in which commonality is facilitated through reuse. This pragmatic, middle-ground approach reflects the approach taken in research areas such as temporal, spatial and multi-dimensional modelling which argue for a flexible but unified way of handling common modelling problems.
Publications
|
|
|
