Re: [regrep-semantic] Use case document

[ewallace@cme.nist.gov]

Carl wrote:


>Since there has been little debate on the content of the 'Use Case
>Feedback' document - I hope at our next meeting we can agree to send it
>out by 21 May.

O.K.  Deadlines are good.

>Hopefully we can also agree that we are equally focused on the use of
>Application Ontologies as well as Reference Ontologies such as SUMO
>http://sunsite.informatik.rwth-aachen.de/Publications/CEUR-WS//Vol-94/ki03rao_menzel.pdf

Several comments on this.  First, it is good that Carl provided the
URL to this paper, as I mis-interpreted the classes of ontologies denoted
by the terms: Application Ontology and Reference Ontology on my initial
reading of Carl's message.  One source of this misinterpretation was the use 
of SUMO as an example of a RO.  PSL seems to me a better example.  Note by 
Chris Menzel's description an RO needn't be an upper ontology at all.

Second.  Have we talked that much about the types of semantic content
we expect to support with the SCM effort?  I can't recall any in-depth
discussions on this.  We have a few examples, and we have had considerable 
discussion of forms for the content, such as OWL, KIF, Topic Maps and even 
UDEF.  We probably should have more discussion about other characteristics
of what we expect for Reg/Rep semantic content, before making a decision 
about it.

Finally, again by Menzel's definition, our choice of supported form will 
constrain which of these classes of ontology we could support.  ROs must be 
in some highly expressive formalism like FOL (eg KIF) which means if we only 
support OWL then we can only support AOs.  Similar thinking is what drove
some of us to advocate KIF support in SCM, although this certainly has to be 
weighed against the cost of that support.  Then there is the question of 
whether OWL + SWRL wouldn't be good enough to support ROs that the eBusiness 
community might need.

Note - Chris Menzel is firmly in the FOL camp, and he can be dismissive of
the needs of mere domain experts.  This paper seems quite balanced though.

-Evan