Re: SUO: Almost-Final Ballot Questions
John,
The revised motion includes the reference to the SUMO conformance
clause, which is needed in a standards proposal, as you point out.
Adam
At 11:24 AM 5/21/2003 -0400, sowa wrote:
>Adam and Frank,
>
>That is the kind of vague question that a well defined
>standard should answer. And what is needed is not a
>dictionary definition of the word "ontology", but a clear
>specification of the formats and interfaces supported by
>the standard(s) we are developing:
>
>AP> One comment about "Is a UML model an ontology?". There are
> > a range of possible formalisms in which to state an ontology.
>
>The word "ontology" has been bandied about so much that it has
>lost most of its meaning. That is why we need a precisely defined
>standard that specifies exactly what formalisms a conforming
>implementation will support and what functions it will provide.
>
>AP> A simple taxonomy (class/sublcass structure) is an ontology.
> > However, such an ontology lacks the full definitions that
> > a formal ontology such as SUMO provides.
>
>That is what the "joint motion" is intended to specify. So far,
>SUMO is a collection of content. But there is no specification
>of the requirements on that content in order to conform to an
>IEEE standard. Specifying those requirements is the primary
>purpose of the joint motion.
>
>>A simple taxonomy is much like a dictionary that just has words,
>>prononciation, and synonym information, but lacks the definitions. UML
>>is getting more expressive, with the addition of OCL.
>
>Providing SUMO as an example is fine. But an example is not
>a specification of what it requires to conform to the standard.
>
>>My understanding though is that OCL is not a widely used component of the
>>language.
>
>That understanding is, by and large, correct -- primarily because
>there are multiple competing factions in the UML community. The
>three gurus who founded UML didn't have a high regard for logic or
>the need for a logic-based specification language. IBM was the
>primary supporter for OCL, and the three gurus made a half-hearted
>commitment to bring OCL into the fold in order to gain IBM's support.
>
>For UML 2.0, the pro-OCL faction is proposing to support full FOL
>in OCL and use OCL as the foundation for defining all of the other
>UML features. However, there is still a large UML contingent that
>doesn't see the need for logic in general or OCL in particular.
>
>Bottom line: It's OK to use the word "ontology" in the name of
>the standard, but it is more important to specify exactly what
>formats and interfaces must be supported in order to conform
>to the standard. And examples, such as SUMO or OpenCyc, are
>not specifications.
>
>John Sowa
>