Re: Some thoughts on K.I.F. requirements
"Michael R. Genesereth <"Michael R. Genesereth <mrg@cs.Stanford.EDU>"@sumex-aim.stanford.edu>"
Sender: Michael Genesereth <mrg@sunburn.stanford.edu>
Date: Thu, 9 May 91 14:08:22 PDT
From: "Michael R. Genesereth <"Michael R. Genesereth <mrg@cs.Stanford.EDU>"@sumex-aim.stanford.edu>"
Reply-To: genesereth@cs.stanford.edu
To: hewitt@ai.mit.edu
Cc: SRKB@isi.edu, INTERLINGUA@isi.edu, hewitt@ai.mit.edu
Subject: Re: Some thoughts on K.I.F. requirements
In-reply-to: Your message of Wed, 8 May 91 12:28 EDT
Message-id: <CMM.0.90.2.673823302.mrg@Sunburn.Stanford.EDU>
Carl,
The presumption that has been made thus far is that the language of
communication has two levels. KIF is the language used to express
information about the world and has a purly declarative semantics.
KQML is a language of ''speech acts'' in which KIF sentences are the
content. For example, in KIF we can say (= (father joe) bill) and
for two machines to communicate, this would get wrapped in an
utternace something like (assert '(= (father joe) bill)). KQML is
not yet fully worked out; but I can elaborate if you are interested.
Alternatively, check with Tim Finin or Gio Wiederhold, the authors of
the current KQML document.
mrg