Re: SUO:-- Functional Roles & slots
John Sowa wrote:
> >(about) http://www.ai.sri.com/~okbc/spec/okbc2/okbc2.html
> To summarize, the authors claim two advantages of OKBC over
> KIF: (1) it includes commands for KB query and manipulation,
> and (2) it implements only a subset of KIF. Advantage #1 can
> be obtained in KIF by defining exactly equivalent commands.
> Advantage #2 can be obtained in KIF by writing a tutorial that
> only presents the KIF subset that is equivalent to OKBC.
>
> >I think we need to maintain a glossary of recommended
> >terminology and adhere to that if possible in these discussions,
> >to make life easier for newcomers and to avoid unnecessary
> >terminological disputes.
>
> I agree with this point, but I strongly disagree that the
> OKBC terminology be included. Remember, I am the principal
> editor of the CG standard, which has its own terminology,
> notation, and tools. But I have not proposed CG terminology
> for inclusion in the SUO glossary for a very simple reason:
> all the fundamental SUO concepts can be defined in logically
> equivalent ways in either KIF or CGs. The same cannot be said
> of OKBC -- only *some* of the SUO concepts that are definable
> in KIF or CGs can be defined in OKBC.
>
> Bottom line: OKBC can only be used for a subset of the requirements of the
> SUO, any statement in the slot and frame
> notation of OKBC can be translated to a very similar statement
> in KIF, and the OKBC terms are merely redundant synonyms for
> the logical terms that are already available in KIF.
>
> John Sowa
>
I agree with John.
From the discussions on my starter structural ontology,
it is clear that some of the formalism for KIF could be
expanded (especially as regards the use of meta language).
This is a very cogent argument that we should borrow useful
facets of OKBC, but expanding our notation and glossary to
be a superset of OKBC seems to be a way of spending time
in a nonproductive way.
David (whitten@lynx.eaze.net) (713) 791-1414 ext 6116