Re: Intersection of properties?

Michael Schneider wrote:
>> Here is my pressing issue.
>>
>> The OWL language goes to great lengths to be monotonic and decidable,
>> but cannot express what to do with contradictions.  For example:
>> DisjointClasses(Human, City)
>> ClassAssertion(Washington, Human)
>> ClassAssertion(Washington, City)
>>
>> Contradictions like this are inevitable.  For me to be able to use OWL
>> without
>> needing to extend it with an external rules language, an OWL ontology
>> needs to be expressive
>> enough to say what to do if someone inserts a contradictory assertion.
>> This is such an obvious "use case" that it may be been talked to death
>> already.
>> Will some future version of OWL be expressive enough to give a rule for
>> resolving
>> contradictions?
>>     
>
> That's a different notion of "expressiveness" as what I normally associate
> with a language such as OWL. OWL is based on classical logic, but what you
> are asking for isn't. I don't expect the core language of OWL to follow this
> direction in the future. However, OWL might grow into a whole set of related
> languages, where some of them are non-classical, who knows...
Jeff,

Well, Michael has mentioned Pronto so I feel like I'm supposed to throw 
my two cents in :)

First off, unless UNA is asserted I'd say there's no reason to believe 
that the two Washingtons represent the same object. So there should be 
no contradiction.
Second, some sort of similar ambiguity can be handled in OWL via punning.
Third, yes, Michael is correct, Pronto does provide non-monotonic (i.e. 
default) reasoning. You're welcome to read my blog posts on that [1] if 
you're interested. Michael's example is correct but may not be the best 
one here because the statements essentially represent probabilistic 
facts about *a single* object (which can belong to one concept or 
another, we don't know for sure which one). But what you're looking for 
is the representation of two distinct objects using the same name 
(unless I misunderstood).
Finally, if you're interested in contradictions, I suggest you take a 
look at the Anthony Harper's research [2]. He explains how conflicting 
pieces of knowledge can be backed by different arguments so that users 
(reasoning engines) can choose one or another during reasoning.

[1] http://clarkparsia.com/weblog/category/semweb/probabilistic-reasoning/
[2] http://www.cs.ucl.ac.uk/staff/a.hunter/research/argumentation.html

cheers,
pavel

>  
>
> For the moment, it may or may not be interesting for you to have a look at
> Pronto:
>
>   <http://pellet.owldl.com/pronto>
>
> This Pellet extension provides probabilistic reasoning. I didn't yet find
> the time to look deeper at it, but adapting your example above, I guess you
> could express assertions like:
>
>   * Washington is a Human with prob = 45%
>   * Washington is a City with prob = 75%
>
> Now, the result wouldn't be a real contradiction anymore, but only wrong
> with a certain (well defined) degree of probability.
>
> But I'm not sure that this is a valid example for Pronto. Bijan can (and
> certainly will! :)) tell you more.
>
> Cheers,
> Michael
>
> --
> Dipl.-Inform. Michael Schneider
> FZI Forschungszentrum Informatik Karlsruhe
> Abtl. Information Process Engineering (IPE)
> Tel  : +49-721-9654-726
> Fax  : +49-721-9654-727
> Email: Michael.Schneider@fzi.de
> Web  : http://www.fzi.de/ipe/eng/mitarbeiter.php?id=555
>
> FZI Forschungszentrum Informatik an der Universität Karlsruhe
> Haid-und-Neu-Str. 10-14, D-76131 Karlsruhe
> Tel.: +49-721-9654-0, Fax: +49-721-9654-959
> Stiftung des bürgerlichen Rechts
> Az: 14-0563.1 Regierungspräsidium Karlsruhe
> Vorstand: Rüdiger Dillmann, Michael Flor, Jivka Ovtcharova, Rudi Studer
> Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus
>
>
>

Received on Tuesday, 5 August 2008 02:32:09 UTC