Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[protege-discussion] Can't classify the OWL class properly, Why?

刘康 liukzg at gmail.com
Fri Sep 4 02:26:44 PDT 2009


I don't know how the reasoner is writed, but I think that when
reasoner is used, the individual of class C is given, so all of the
fillers of property pA ARE KNOWN and there is no UNMENTIONED filler of
pA for the given individual, the reasoner can rule out the correct
result according to the given individual and given fillers. Otherwise,
the reasoner fails to  implement the logical of OWL and limits the
application of OWL.

2009/9/3 Thomas Russ <tar at isi.edu>:
>
> On Sep 2, 2009, at 7:01 AM, 刘康 wrote:
>
>> if there are 3 class A, B, C is asserted, and A has the property pA.
>> If defines the C with the assertion that C is the class A that the
>> property pA only class B (NECESSARY & SUFFICIENT: class A, pA only B),
>> I find the protege can't classify the individual properly, could you
>> tell me why?
>
> Open world.
>
> Just because all of the KNOWN fillers of the property pA happen to belong to
> class B, it doesn't rule out the possibility that there is some UNMENTIONED
> filler of the property pA that doesn't belong to class B.
>
> Open world means that classification using only, exact cardinality and
> maximum cardinality cannot be easily made.
>
> There has to either be some direct assertion, or else you have to include
> some closure information such as asserting that the individual doesn't have
> any individual fillers except the ones that are known.  This can be done by
> asserting that the instance belongs to the class (pA only {set of fillers of
> pA on the individual}).
> _______________________________________________
> protege-discussion mailing list
> protege-discussion at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/protege-discussion
>
> Instructions for unsubscribing:
> http://protege.stanford.edu/doc/faq.html#01a.03
>



More information about the protege-discussion mailing list