RE: SEM: 5.1 literal/data values

> [...]
> 
> > I do not expect OWL systems to be able to reason about integers, XML
> > Schema dataype cardinality, the length of strings, and many other XML
> > Schema Datatype properties.  
> 
> Why not?  If these things are in OWL, then an OWL system should be able to
> reason about them.  For example, I would expect an OWL reasoner to be able
> to determine that there are only 11 integers in the range from 0 and 10
> inclusive, provided that such ranges are allowable in OWL.  

Let me just say, the more powerful OWL is, the better. 

> To expect
> otherwise is to say that integer ranges are not really part of OWL.
>
> Now XML Schema datatypes have many mysterious and wonderful aspects, such
> as datatypes defined by regular expressions.  It may be that the WG does
> not include all of XML Schema in OWL, but the parts that do make it in
> should *really* make it in.

If things are *really* in, they need to be formally defined
(syntactically and semantically).  dateTime is an XML primitive
datatype with numerous syntactic requirements.  "The two digits in a
MM format can have values from 1 to 12."  Are OWL parsers required to
reject
                                   __
 <foo xsi:type="xsd:dateTime">1999-20-31T13:20:00-05:00</foo>? 

And 'really in' is kind of funny, since I see no sign that we are
going to support reasoning about inequalities, a fairly fundamental
property of many of the primitive datatypes.  

I think it is necessary that we build connections to other standards,
even if we don't incorporate those standards fully into our semantics.
That is, we can treat some objects as blobs for which we can sometimes
prove equality.  But sometimes we won't be able to say if two blobs
are equal or not.  Just as the XML Schema Datatype notes that the
relation between some dateTimes is indeterminate.

- Mike

Received on Friday, 19 July 2002 15:34:00 UTC