alternative interlingua
R. V. Guha <ai.guha@mcc.com>
Date: Fri, 10 Aug 90 13:01 CDT
From: R. V. Guha <ai.guha@mcc.com>
Subject: alternative interlingua
To: pfps@allegra.tempo.nj.att.com
Cc: interlingua@venera.isi.edu
In-reply-to: <9008091817.AA04788@vaxa.isi.edu>
Message-id: <19900810180105.6.GUHA@GAIA.ACA.MCC.COM>
I have a couple of worries/comments on the alternative proposal.
(a) While it might be reasonable to emphasize on some set of
features for a first shot at the interlingua, I am concerned
about the choice of these features. Do we have any idea about
the features that people who actually build non-trivial KBs
depend on? Or are we just picking features that have some
reasonable hope of being nicely formalized? Is the interlingua
meant for people building KBs to share KBs or is it meant for
people who analyse languages to share their analysis more easily?
(b) I am unsure about this idea of taking just some of the
statements in one KB and not taking others. This might turn
out to be real problematic especially with defaults etc.
eg. I could take a whole bunch of axioms with abs in them
but without some axiom that minimizes these abs, the other
axioms are quite useless. So, when is it reasonable to take
some axioms and not others?
Guha