Re: Ontologies in LOOM

James Rice <Rice@HPP.Stanford.EDU>
Message-id: <2978272259-9666234@KSL-EXP-35>
Sender: RICE@KSL-EXP-35.Stanford.EDU
Date: Wed, 18 May 94  10:50:59 PDT
From: James Rice <Rice@HPP.Stanford.EDU>
To: Cheng Hian GOH <chgoh@MIT.EDU>, ontolingua@HPP.Stanford.EDU
Subject: Re: Ontologies in LOOM
In-reply-to: <9405181525.AA20985@e51-007-3.MIT.EDU>
>>  ... (Stuff like
>>  Standard-Units-and-Dimensions and Abstract-Algebra
>>  will really make a real difference for us.)

>>  Unfortunately, we have so far committed ourselves to
>>  LOOM and the LOOM version of these ontologies don't
>>  seem to work.  Tom Gruber has pointed out this may be
>>  a bug in the Ontolingua->LOOM translator.  While I
>>  greatly appreciate the work which they are doing, we
>>  cannot sit around and wait for the bugs to be fixed.
>>  My question is, "Is there someone who has got the LOOM
>>  version of these ontologies to work?" Maybe the answer
>>  is that we should switch to a different representation
>>  language.  In that case, I will be really interested
>>  to hear people's experiences.  Is there a comparable
>>  alternative to LOOM which will work well with
>>  Ontolingua?

I'm the chap who was recently struggling to get the LOOM
back end for Ontolingua to work.  It is not a trivial
problem at all.  In the current release (4.0.alpha) we at
least managed to get all of the bootstrap ontologies to
load without errors (we believe), but we didn't have a
chance to get to all of the other ontologies, such as
units and dimensions.

The problem from your point of view, as I see it (if we
ignore the problem of my finding the time to work on the
LOOM back end), is the lack of expressive power and
uniformity in LOOM.  We're working very hard to get as
much of Ontolingua to translate as possible into LOOM but
it simply cannot handle a bunch of stuff.  Thus, even when
ontologies load without dumping you in the debugger, the
ontology might not "work" in the fullest sense (or even in
a minimal sense).  For example, even though we will make a
translation for the polymorphic definitions of "*" in the
physical-quantities ontology it certainly won't do
anything other than punt.

Maybe you could tell us more about what you expect to be
able to do with these ontologies.

I'm not competant to advise you on which representation
system to use, but I can't help but observe that
Ontolingua is (in many senses) just syntactic sugar on
KIF.  If you're going to use it to the full then you'll
need a representation system with the expressive power of
KIF (I guess Epikit is an example of such a beast).  If
you're prepared to restrict yourself to a subset of the
Ontolingua language (such as pretty minimal framish stuff)
then more options are open to you, but you will find that
you won't be able to import ontologies from people who
write using full Ontolingua.  IMO, this defeats the
purpose of both KIF and Ontolingua.

Your milage may vary, though,