Re: Propositions and Knowledge Interchange

• Mail folder: Interlingua Mail
• Next message: sowa: "Re: Propositions"
• Previous message: Ramesh Patil: ""
• Maybe in reply to: phayes@cs.uiuc.edu: "Re: Propositions and Knowledge Interchange"

Message-id: <199402152326.AA25059@quark.isi.edu>
X-Sender: macgreg@quark.isi.edu
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Tue, 15 Feb 1994 15:27:53 -0800
To: cg@cs.umn.edu, interlingua@ISI.EDU
From: macgregor@ISI.EDU
Subject: Re: Propositions and Knowledge Interchange

Bob:
>>I can't imagine a standard being approved that doesn't specify a semantics.
>>But for practical purposes, it matters little what that semantics happens
>>to be.
Pat:
>Bob, maybe you just intended to be rhetorical, but if you really mean this
>then we should all stop now.

Hmm. Your idea has merit :-)

Reading back, my first paragraph overstated my case a bit, but I still
stand by my second paragraph (which I have included above).  Each 
language concept (e.g., definition, type, proposition) needs to have
*some* semantics associated with it.  The writer of a system A to KIF
translator will examine that semantics to see how closely it
corresponds to the semantics in A that she is mapping from.  If it
is "close enough", then she will map to the predefined KIF construct.
So, the *purpose* of the semantics is to provide the translator writer
with a "pretty good" idea of what the construct means.

Translator technology is practical only if "close enough" is not
required to be equivalent to "exactly the same". Taking types as an
example, if we achieve gross consensus on what a "type" is, but can't
agree on the fine points, that's good enough--from a translation
standpoint it is immaterial which semantics we pick. However, if
disagreement exists at a gross level, then perhaps we need to define
several flavors of "type", with different semantics, and the translator
writer will have to choose with kind of "type" best agrees with her own
system's notion of type.

Pat:
> ... if the semantics
>doesnt matter, what else is there?

Taking your phrase slightly out of context:  Well, there is kids soccer,
and skating on the boardwalk, and ... .

Rather than arguing endlessly about differences in viewpoint, I think
progress will be made only if the discussion turns towards an attempt
to agree on something. Thus far, I have failed to observe anything
resembling a consensus on a single topic.

- Bob


Robert M. MacGregor                                     macgregor@isi.edu
USC/ISI, 4676 Admiralty Way, Marina del Rey, CA 90292      (310) 822-1511