KIF: Re: SUO: RE: SUMO axiomatization

[Adam Pease]

John,
   We've been over at least one of these points before in some detail.  I'm 
not opposed to sorts.  SUMO has them and so does Cyc.  I'm opposed to 
including sorts in the syntax because they are unnecessary.  You don't need 
syntax like (forall (?X:Person)...) to take advantage of sorts that are 
defined on the argument types of predicates (this is what Cyc does).  This 
comes down to a question of personal preference and aesthetics.
   The point about language complexity is not really debatable.  I think it 
makes the syntax more complex which I consider simply equivalent to a 
factual statement that it requires more clauses in the BNF grammar.  You 
could interpret "more complex" differently.  You could say (as Bill and 
Chris have) that "more complex" refers to the axioms that are needed to 
define particular concepts in the ontology.  They've shown good examples of 
how row variables make for more compact definitions.  Are they simpler or 
easier to understand?  It depends on what community you're addressing, and 
comes down to personal preference, not a statement of right or wrong.

Adam

At 05:27 PM 12/18/2001 -0500, John F. Sowa wrote:
>Bill and Adam,
>
>The KIF group has already agreed that both sorts and row variables
>will be in the standard for logical foundations (LF).  That is
>the proposal that replaces the proposed separate standards for
>conceptual graphs and KIF.
>
>The LF standard will also include syntactic mappings from the
>semantic foundations to the syntax of the proposed ISO standards
>for KIF and CGIF.  And we encourage anyone else who is defining
>any declarative language (such as the various UML diagrams) to
>map their syntax to and from the same LF in order to ensure full
>interoperability of the language specifications.
>
>To answer Adam, we can also include grammar rules for a "stripped down"
>version of standard KIF that will include only those features that
>Adam wants.  I strongly disagree  with Adam that sorts and row
>variables are not needed for doing ontology.  And contrary to Adam's
>claim that they add "complexity", I believe that row variables and
>sorts make the resulting language much more usable and readable,
>as Bill pointed out.
>
>In any case, I believe that Adam can have a subset that he's
>happy with, and the rest of us can get on with the business
>of defining a standard that we believe is appropriate for
>a wide range of applications, including ontology.
>
>John Sowa

Adam Pease
Teknowledge
(650) 424-0500 x571