Hi Bob,
Thanks for this. It looks great!
I guess we are building a Transcendent, Semi-Formal ontology with read-only model dynamics and volatile instance dynamics. It could certainly help with the language we use to discuss how we do things - metametamodel stuff?
One telling quote. Right from the beginning (page 21):
"The ODM specification offers a number of benefits to potential users, including:
* Options in the level of expressivity, complexity, and form available for designing and implementing conceptual models, ranging from *familiar UML and ER methodologies* to formal ontologies represented in description logics or first order logic."
Although the document is very good at looking at ontology development from different view points it does assume familiarity with UML and ER methodologies. It makes the basic assumption that one is approaching ontology construction from an software engineering background - hence everything tends to be related back to UML. Most TDWG members are not approaching ontology development from a software engineering background so we need to be careful when confronting them with this. It could be argued that this document is largely a "How to do ontologies in UML" document. If you do not have UML ingrained in your culture then it may not be so useful.
Still a great document with loads of interesting stuff in it though. Good find!
All the best,
Roger
Bob Morris wrote:
Preparing the syllabus for my graduate software engineering course, I stumbled on http://www.eclipse.org/modeling/ which leads eventually to the recently adopted OMG Ontology Definition Metamodel (ODM) http://www.omg.org/docs/ad/06-05-01.pdf
In Model Directed Architecture, a Metamodel constrains how you can model your problems. Metamodels are always implicitly or explicitly behind any modeling tools. Explicit is better, because then the tools can analyze your conformance to the model. The eclipse modeling framework (EMF) supports this, so that tools like the IBM Rational Software Modeling and Rational Software Architect eclipse packages can implement it. Probably so do other modeling tools.
In the detailed TOC of the above PDF, one will find every single topic around which the TAG discussions have swirled, and more.
I'm sure that everyone on this list shares my excitement in finding that there is potentially a rigorous framework for the tag ontology efforts and will drop everything else to absorb this 338 page specification before coming to St. Louis, at which the program should be revised to comprise nothing else but its discussion for the entire week. Hah, hah, just serious.
--Bob