Data management for interoperable systems
Ralf Mühlberger · The University of Queensland · 2001
To meet the overwhelming need for content, we have to be able to access and integrate information from many different sources such as databases and workflow systems. Central to such interoperable systems are the data management issues of schema integration and data consistency.This thesis takes the database centric frameworks of the past, and re-focuses on the business need of accessing interoperable systems, i.e. to execute specific applications over the underlying data sources.Our contributions from addressing this issue, both in theory and through practical consultancies and product enhancement, include:• an extended data model, able to describe data sources, as well as the demand that our applications make on this data• a revised schema design procedure for creating these application schemata• a conceptual schema integration methodology that uses the application schema to restrict the scope of the export schemata that are a key indicator of the complexity (and therefore cost) of the schema integration process• an alternative approach to multidatabases for managing consistency in the face of interdatabase dependencies, using workflow management and the modification of process models to include such data management functions as: rollback plans, compensation activities and consistency management activities• extensions to workflow management system architectures to assist in the creation of interoperable workfows consisting of ad-hoc and production workflow systemsFundamentally we address issues of data management for any data sources, specifically managing the specification of, and constraints over, the data that systems work with when interoperating.