Business applications have long used RDBMs to store configuration data alongside user transactions. Such configurations are increasingly complex, and their proper seeding during testing and production presents a significant challenge that often falls, unrecognized, into the gap between Data Management and Software Development organizations. Part of the solution lies in using more efficiently maintainable artifacts to control the dataload process. Hand-edited SQL scripts are sufficient for some seeding tasks, but when schemas grow in complexity to encompass rich typing systems, a more expressive paradigm is required to bridge the gap from the RDBMS view to the application layer, which is usually driven by a set of nested classifications (e.g. object-oriented inheritance).
Our experience shows that ontologies form an appropriate substrate for building such dataload solutions. In this advanced talk, we discuss the adaptation of object-relational systems to accept frame-based configurations for a spectrum of deployment scenarios. We draw upon examples in the financial and legal domains. Topics include:
- How configuration data impacts system quality
- Making dataload repeatable and maintainable
- The role of metadata in system configuration
- Relating ontologies to UML/ERD models
- Mapping from RDF/OWL to SQL
- Making frame mappings simple and flexible
- Exploiting web services for configuration
Stu Baurmann is a prolific author and inventor, the CTO of Scrutable Systems, and the Lead Instructor at LogicU. He has 17 years worth of wacky stories about software development and information architecture. He resides in both Seattle and Austin.
Stu contributes to many websites and open-source software projects, and is a prominent advocate of knowledge-oriented architecture and comprehensive system validation. He has worked with many Fortune Cum Laude organizations such as Fedex, Reuters, Nationwide, and the Great State of Texas.