W3C home > Mailing lists > Public > www-webont-wg@w3.org > December 2001


From: Pat Hayes <phayes@ai.uwf.edu>
Date: Thu, 6 Dec 2001 10:41:34 -0600
Message-Id: <p05101004b8354b3b8607@[]>
To: "Smith, Ned" <ned.smith@intel.com>
Cc: www-webont-wg@w3.org
>Hash: SHA1
>I'm quite interested in seeing item 1) below realized. Scenario 3) in
>the use cases I proposed is tied to proof carrying authorization.
>BTW: How would you classify the language you describe below? The
>language is more expressive than a description logic, yet not exactly
>a 1st order logic or type theory logic. Are you suggesting this is
>the language WebOnt will define?

I would like to see it move that way, yes. In fact (I hesitate to say 
this in case the Captain is listening) I would suggest something like 
a 'neat' syntax for FOL - perhaps a subset, eg a constructive or 
intuitionist subset(?) - with a distinguished typing syntax that is 
similar in power (though maybe not quite in form) to DAML+OIL, ie a 
genuine logical extension of DAML+OIL.  The DAML+OIL-like subset 
would be the language for defining logical sorts in the full 
language, and there would of course be natural ways to translate them 
into logical form if anyone wanted to, but such a translation would 
abandon the proven utility of the class-based reasoning engines. (The 
emerging ISO-KIF standard might provide a basis; but it would need to 
be re-done in more Webbish way, related to XML and so on.)

I am quite sure that eventually the SW is going to re-discover the 
utility of FOL, just as AI and KR and database technology eventually 
did, and I would like it to be sooner rather than later.

>  In what ways does DAML+OIL hit/miss
>the objective?

Well, DAML+OIL should obviously be included as a sublanguage, in such 
a way as to preserve its computational properties. (But I don't think 
we should try to *reduce* the WOL to DAML+OIL; that kind of reduction 
is likely to produce all kinds of semantic problems, in spite of its 
perceived advantages in avoiding the need to write parsers.) I think 
that as a basic ontology language, DAML+OIL is 'warped' by its being 
a class/property language rather than a conventional logical 
language. For things that are simple and natural to state as in 
class/property terms it is fine, but other kinds of propositions 
require one to introduce very artificial classes defined in terms of 
restrictions in order to state simple facts (and simple queries), and 
the resulting complexities of needing to learn to think in this odd 
way add further barriers to the already considerable difficulties of 
using DAML+OIL. It seems to me that there are some things that class 
reasoning does extremely well, but other things it is less well 
suited to, and that we should attempt to provide a language which 
allows it to be used naturally, but does not insist on its being used 
at all times. This is hardly a new opinion, of course: description 
logics emerged from the old idea of there being two kinds of 
representation, which used to be called the A-box and the T-box. I 
would like us to have both boxes again, is all.

Pat Hayes

>Ned M. Smith
>Intel Architecture Labs          Phone: 503.264.2692
>2111 N.E. 25th Ave               Fax: 503.264.6225
>Hillsoboro OR. 97124            mailto:ned.smith@intel.com
>>  -----Original Message-----
>>  From: Pat Hayes [mailto:phayes@ai.uwf.edu]
>>  Sent: Thursday, November 29, 2001 8:40 AM
>>  To: www-webont-wg@w3.org
>>  Subject: WEBONT "HOMEWORK"
>>  What I would like:
>  > 1. An ontology language which was expressive and 'natural' enough
>>  to  encode most currently extant ontologies. That is considerably
>>  more  expressive than a description logic, but it can have a
>>  description  logic as a natural sublanguage (the part of the larger
>>  language that  deals with type-class reasoning). The natural choice
>>  would be some  variant of either an extended first-order logic such
>>  as ISO-KIF, or  possibly a type theory-based logic like LF. Part of
>>  the development  work would be to include a notion of
>>  proof-carrying authorization in  the proof theory of the ontology
>>  language.
>>  2. The homework would be to integrate this expressive language with
>  >  the kind of human-oriented interface being developed in the
>>  context  of the DARPA RKF project, in which a 'graphic' interface
>>  allows
>>  subject-mater experts who know zilch about KR or logic to fairly
>>  easily, with some training and practice, create large, complex
>>  ontologies in man-month timeframes. Hopefully, this could be
>>  designed  in such a way that later work could build on earlier
>>  work, in the  sense that the concepts developed in earlier
>>  ontologies can be
>>  utilised in later ones.
>>  3. In a parallel effort, a fairly small team of ontological
>>  engineers  can systematically collect existing useful ontologies of
>>  broad
>>  utility - of which there are now several hundred, covering topics
>>  such as: time-intervals and calendars, part/whole mereological
>>  theories, spatial reasoning, order-sensitive reasoning, theories of
>>   networks and reticulations, process and action descriptions,
>>  industrial processes, etc. etc. . Some of these are more 'abstract'
>>   than others; the sources range from philosophical analyses to
>>  industrial standards organizations; but they can all be put into a
>>  common framework, and indeed are being so put into a subset of
>>  ISO-KIF by a small team of people at Teknowledge, right now.
>>  All of this is actual work in progress, and could be adopted and
>>  put  into the service of the WebOnt effort immediately. It seems
>>  silly to  ignore it.
>>  Pat Hayes
>>  --
>>  --------------------------------------------------------------------
>>  - IHMC					(850)434 8903   home
>>  40 South Alcaniz St.			(850)202 4416   office
>>  Pensacola,  FL 32501			(850)202 4440   fax
>>  phayes@ai.uwf.edu
>>  http://www.coginst.uwf.edu/~phayes
>Version: PGP 6.5.3

IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
Received on Thursday, 6 December 2001 11:41:42 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 23:04:25 UTC