Re: Roles, again
phayes@cs.uiuc.edu (Pat Hayes)
Message-id: <199509151801.NAA04318@eris.ai.uiuc.edu>
X-Sender: phayes@tubman.ai.uiuc.edu
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Fri, 15 Sep 1995 13:16:18 -0600
To: Don Dwiggins <dwig1@village.ios.com>
From: phayes@cs.uiuc.edu (Pat Hayes)
Subject: Re: Roles, again
Cc: cg@cs.umn.edu, fritz@rodin.wustl.edu, guarino@ladseb.pd.cnr.it,
srkb@cs.umbc.edu, carrara@cs.umbc.edu, giaretta@ipdunidx.unipd.it
Sender: owner-srkb@cs.umbc.edu
Precedence: bulk
Hi Don; some further thoughts on your claim that logic may be a suitable
'low-level' language but not a suitable 'higher-level' one.
Ive been trying to understand where this idea comes from, and it may be
because indeed logical notations do use argument positions rather
arbitrarily. For example the relations sold-to-for and bought-from-for
differ only in a permutation of their argument places, and the relation
names only give a slight clue as to the right order; so to the human reader
(and even writer) of axioms, an essential aid to comprehension is missing.
Also it seems awkward and artificial to have to distinguish bought-from and
bought-from-for and bought-from-for-when (etc.) as different relations when
they obviously have something important in common which they dont share
with, say, being-in-the-same-straight-line-as. Is there anything else which
you find awkward about conventional logical notations, as opposed to
graphical notations like CG?
As to the second point, philosophers suggested years ago that the right
approach here was reification, ie whenever one has such a 'family' of
relations, there is probably an actual entity (a selling) which has
relations to all the other things involved; which suggests that 'roles' are
often best regarded as binary relations between the player of the role and
an event (for want of a better word...'situation', maybe?) of the
appropriate type. Incidentally, this means that every 'selling' is a
'buying' and vice versa, so these two predicates are coextensional. The
difference, I think, is a purely linguisitic one: that these give different
indications about how to assign names in English sentences to argument
places of the relation (or, which binary relation holds between the reified
event and the names).
But to return to the first point. This is certainly an important practical
issue in system development, I agree; Ive found myself making little tables
of relation names in order to refresh my own memory about which argument is
which. I suspect that one purpose of 'roles' is to provide useful labels
for the argument places in order to help the human user keep track of the
evolving syntax. Any comments?
Pat Hayes
------------------------------------------------------------------------------
Beckman Institute (217)244 1616 office
405 North Mathews Avenue (415)855 9043 or (217)328 3947 home
Urbana, Il. 61801 (217)244 8371 fax
Phayes@cs.uiuc.edu