- From: John F. Sowa <sowa@bestweb.net>
- Date: Fri, 17 Nov 2000 11:35:05 EDT
- To: pat hayes <phayes@ai.uwf.edu>, Dan Connolly <connolly@w3.org>, www-rdf-logic@w3.org
Pat,
I also like that approach, and I think it can be done in way
that satisfies the B2B people:
>Yet another strategy (which I myself prefer) is to simply remove the
>concept of 'definition' from KIF altogether, treat all definitions as
>simply assertions of biconditionals, and stick strictly to the Caveat
>Orator principle. But I can see that this might not be acceptable in
>B2B semantic-web kinds of application.
The difference between a definition and a biconditional cannot
be expressed in a pure FOL at the object level. However, it can
be expressed at the metalevel, including a metalevel that uses
FOL to talk about a domain that includes expressions.
At the metalevel, you would need an assert statement to enter
new statements into the current KIF "workspace" or "context".
And when you assert a proposition ?p, you would add another
statement that explains who was the authority for ?p:
(and (assert ?p)
(authorityFor ?p "Pat Hayes"))
Then you could define a hierarchy of authorities, with God
(or BuiltInDefiner) at the top. That definition, of course,
would be asserted as a biconditional at the metametalevel,
with the appropriate god or gods as authority.
John
Received on Friday, 17 November 2000 11:34:35 UTC