Re: Labelling propositions

Robert MacGregor <macgreg@vaxa.isi.edu>
Message-id: <9009172255.AA18119@vaxa.isi.edu>
To: interlingua@vaxa.isi.edu
Subject: Re: Labelling propositions
Reply-To: macgreg@venera.isi.edu
Date: Mon, 17 Sep 90 15:55:13 PDT
From: Robert MacGregor <macgreg@vaxa.isi.edu>
Mike meant to send the following message to everyone.  Since
he asked me to forward it, I will take the opportunity here
to add a comment on the end.

------- Forwarded Message

Date:    Mon, 17 Sep 90 15:17:01 -0700 
From:    Michael Genesereth <mrg@Sunburn.Stanford.EDU>
To:      macgregor@venera.isi.edu
cc:      all@Sunburn.Stanford.EDU
Subject: Re: Labelling propositions 

Bob,

Well labelling is rather easy with quote.  For example:

(= fact1 '(material moon green-cheese))

(believes john fact1)

(not (true fact1))

etc.

The existence of quote is not meant to prevent us from having much
more expressive ways of encoding knowledge,  just as a fall back for those
cases where our more expressive means are inadequatre.  Personally,
I think of quote as a rather samll part of the language (albeit an
essential one).   Personally I would like to address the development
of more expressive things.  However, as I mentioned early on in this
endeavor, I believe we need to worry about the basics before getting
the less primitive things defined.  No that is not right.  I think
we already have basics like quote defined and should just stop
worryiung it and get on with the more imporatnat business as you 
suggest.  It is just that OTHER people keep bringing it up.

mrg

------- End of Forwarded Message

The above example appears to me to be attaching labels to
sentences rather than to propositions.  If I assert

  "(= fact2 '(and P Q))"
  "(probability fact2 0.2)"

and then test the truth of "(probability '(and Q P) 0.2)" I would expect
the answer "unknown" (assuming no other assertions had been made).  If
the probability were attached to a PROPOSITION rather than to a
sentence, I would have expected the answer "true".  I would prefer
having a labelling capability that attaches labels to meanings (of
statements).  Mike's use of quotation, which assigns labels to
sentences is easy, but not useful for the kinds of things I want to
label.

- Bob