Re: An ontology for KIF
pat hayes <hayes@hpp.stanford.edu>
Full-Name: Patrick Hayes
Date: Mon, 16 Aug 93 14:33:50 -0800
From: pat hayes <hayes@hpp.stanford.edu>
To: jfulton@grace.boeing.com (James A. Fulton)
Subject: Re: An ontology for KIF
Cc: macgregor@ISI.EDU, interlingua@ISI.EDU
Message-id: <Mailstrom.1.03.19278.-18772.hayes@hpp.stanford.edu>
In-reply-to: Your message <9308161844.AA01811@grace.rt.cs.boeing.com> of Mon,
16 Aug 1993 11:41:37 -800
Content-Type: TEXT/plain; charset=US-ASCII
Jim,
I agree with the idea that KIF should be a skeleton with ontologies (ie
theories) capturing particular knowledge. But the issue is whether the ontology
has a special relationship to the syntax. The distinction between type and
unary-predicate has no semantic content in the usual semantics of KIF, so can't
be captured in an ordinary ontology: but some systems need to have it
maintained. So, some predicates have to be specially marked as type-ish. But
this kind of 'marking' seems to involve a different kind of relationship between
(an aspect of) KIF syntax and a predicate (specifically, the "Type" predicate)
than the usual kind of relationship. This is a change (extension) to the logic,
and should really be treated like one: but in any case, it can't be captured by
an ordinary first-order theory (whoops, sorry, ontology).
This is an example of a more general phenomenon in which we might want to have
theories about syntax. For example, suppose for some reason that an application
wants to distinguish predicates whose names begin with the letter 'P'. These
differ from ordinary predicates in that the truth of ground instances can be
computed directly, so an ontology need not always involve any axioms. This
sounds like a useful kind of predicate to be able to specify.
Pat