RE: Proposed simplification of datatype expressions

PERSONAL OPINION, NOT CHAIR:

Given RDF Cores refusal to revisit the issue of literals in 
collections, I think that lists of datatype properties are going to 
be prohibitively difficult for people to use, and they will need to 
do work arounds.  I think this will hurt OWL, but not as badly as it 
hurts RDF, but in this go round I think for us to put much work into 
fixing things is wasted time.
  I searched through the DAML libraries looking at all the used of 
collections in the 196 ontologies in the library.  In each case 
people found ways to say what they wanted.  Based on that, I think we 
should move forward with our current design.

END OF PERSONAL OPINION


At 13:46 -0500 3/21/03, Evan Wallace wrote:
>Jeff Pan wrote:
>
>>From: "Peter Crowther" <Peter.Crowther@networkinference.com>
>>
>>>  > From: pat hayes [mailto:phayes@ai.uwf.edu]
>>>  > >"people who have at least three values for property x."
>>>  > >
>>>  > >where x is a data valued property. I, too, struggle to think
>>>  > of an example
>>>  > >where one would actually want to use such a construct (but am open to
>>>  > >suggestion)....
>>>  >
>>>  > People who own three or more homes, people with at least three
>>>  > sources of income, people with at least three children, people with
>>>  > with at least three nationalities, people who have travelled to more
>>>  > than three foreign nations during the last six months,....
>>>
>>>  All good examples of cardinality contraints to objects;
>>
>>Well, that usually depends on what you want to model. Good examples of
>>cardinality constraints to objects can also be good examples of cardinality
>>constraints to data values in some situations (see below).
>
>This question of "what you want to model" is an interesting point.  It
>seems to me that there are times that one wants to collapse some walk
>though "conceptual" objects and properties into a simple datavalue
>(oo:attribute), particularly in cases where the other properties of
>intermediate objects are not of interest to a particular set of
>applications.  I don't mean derived values, but rather models/ontologies
>simplified for expected use. 
>
>Then there is also the notion of "external key" which may still be
>important for working with data associated with 2 otherwise inconsistent
>ontologies which may provide different views over overlapping sets of real
>world objects. 
>
>Perhaps, these views are antithetical to the way of thinking in the
>Semantic Web world?  If so, I don't care enough about this to engage in
>theological discussions.  I do, however; believe that it is important to
>maintain the ability to express oneof datavalue sets.  There are plenty of
>design-to-order style parametric product models which allow only a small
>set (unordered) of values for the parameter.  I can also think of querying
>scenarios where cardinality restrictions would apply to DataProperties:
>   Find all the new MINIs with two colors. 
>   Find all US car models in 1958 produced with 3 colors (tri-color option).
>
>If we revisit the decision made yesterday, I would object to adopting
>Ian's proposed simplification of datatype expressions.
>
>-Evan
>
>
>
>-Evan


-- 
Professor James Hendler				  hendler@cs.umd.edu
Director, Semantic Web and Agent Technologies	  301-405-2696
Maryland Information and Network Dynamics Lab.	  301-405-6707 (Fax)
Univ of Maryland, College Park, MD 20742	  240-731-3822 (Cell)
http://www.cs.umd.edu/users/hendler

Received on Friday, 21 March 2003 14:00:15 UTC