W3C home > Mailing lists > Public > www-webont-wg@w3.org > August 2002

Re: How to do it/Tricks of the trade for closed world reasoning

From: Jeff Heflin <heflin@cse.lehigh.edu>
Date: Fri, 30 Aug 2002 17:53:49 -0400
Message-ID: <3D6FE96D.A7E12252@cse.lehigh.edu>
To: Deborah McGuinness <dlm@ksl.stanford.edu>
CC: webont <www-webont-wg@w3.org>

Deborah,

I've also done some work on a different kind of closed world
information.
In a paper [1] I presented at the recent AAAI Ontologies for the
Semantic Web workshop, I discussed how Oren, Etzioni, and Weld's local
closed world (LCW) information (also called local completeness
information) could be expressed in DAML+OIL. Such information
essentially allows you to say that a particular document has complete
information on the members of some class (for example, American
Airline's website has complete information on all flights by American
Airlines). Although I introduced new syntax to do this, I showed how it
could also be represented using existing DAML+OIL syntax (the key of
course is to use oneOf to enumerate all the instances) and then same
that this is the sameClassAs whatever class expression you want to say
you have complete information for (note, although I didn't catch it
before publishing the paper, the talk I gave at the workshop actually
shows a simpler way to do this than the paper does).

Jeff

[1] 
Deborah McGuinness wrote:
> 
> ACTION: Deb to take a shot at writing up the closed world example
> 
> I wrote up a simple example of closed world reasoning in:
> 
> http://www.ksl.stanford.edu/people/dlm/daml/closedWorld.html
> 
> This just counts the number of values on a property on an individual and
> asserts that maximum cardinality restriction on that property for the
> individual.
> This is the simplest notion of closed world reasoning that we can
> capture.
> I have done some other work (with borgida and hull) on attempting to
> close off an entire individual
> and put the work down a few years ago when it was more complicated than
> we anticipated.
> I will also add a summary of where we left off in a more complicated
> example.
> 
> i am enclosing the contents of the url above for archiving purposes for
> the list.
> I used a combination of Guus's how to do it style and my previous tricks
> of the trade style.
> 
> d
> 
> ==========
> OWL/DAML+OIL "Tricks of the Trade / How To Do It":
> Closed World Reasoning Example
> Author:  Deborah L. McGuinness
> 
> --------------------------------------------------------------------------------
> 
> Title - Closed World Reasoning.
> 
> Task/Modeling problem - Include a notion of closed world reasoning in an
> OWL or DAML+OIL knowledge base. There are many examples of closed world
> reasoning but this example includes two common cases.
> State that the known values for a particular property on a particular
> individual are the only values for that property on that individual.
> Assume for example, that there is an individual instance of the class
> "Person" named "Deborah" with a property "hasCar" with a value
> "Saab900Turbo". The task is to allow a reasoner to infer that Deborah
> only has the saab (and no new distinct individuals will be added as
> values to her hasCar property).
> Observation - DAML+OIL and OWL do not make the closed world reasoning
> assumption, i.e., they do not assume that information that is not known
> to be true is false. They work with an "open world reasoning"
> assumption, i.e., they assume that information that is currently not
> known may become known in the future. Thus, just because Deborah only
> has one known value for a hasCar property does not mean that she may not
> have more cars. Said another way, there is no implicit maximum
> cardinality restriction on Deborah's hasCar property.
> 
> Abstracted solution -
> Count the number of values for a property on an individual and then
> assert a max cardinality on that property for that individual.
> 
> Example solution - In this particular case, we would add a maximum
> cardinality restriction of 1 on Deborah's hasCar property. This would
> allow reasoners to infer that no additional distinct values for
> Deborah's hasCar property may be added (without first removing the value
> of Saab900Turbo).
> 
> Notes -
> 
> This solution is the same as the notion of closing a role in the CLASSIC
> description logic. More issues on closing roles can be found in:
> - Ronald J. Brachman , Deborah L. McGuinness , Peter F. Patel-Schneider
> , Lori Alperin Resnick , and Alex Borgida. ``Living with CLASSIC: When
> and How to Use a KL-ONE-Like Language,'' in John Sowa, ed., Principles
> of Semantic Networks: Explorations in the representation of knowledge ,
> Morgan-Kaufmann: San Mateo, California, 1991, pages 401--456.
> http://www.ksl.stanford.edu/people/dlm/papers/living-with-classic-abstract.html
> 
> -Alex Borgida, Rick Hull, Deborah McGuinness. On the Confluence of the
> Closing Order in Description Logic Knowledge Bases. Working Draft.
> 
> --
>  Deborah L. McGuinness
>  Knowledge Systems Laboratory
>  Gates Computer Science Building, 2A Room 241
>  Stanford University, Stanford, CA 94305-9020
>  email: dlm@ksl.stanford.edu
>  URL: http://ksl.stanford.edu/people/dlm/index.html
>  (voice) 650 723 9770    (stanford fax) 650 725 5850   (computer fax)
> 801 705 0941
Received on Friday, 30 August 2002 17:53:51 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:57:51 GMT