W3C home > Mailing lists > Public > www-rdf-interest@w3.org > December 1999

RE: A simple question on RDF Querying ...

From: Greg Karvounarakis <gregkar@csi.forth.gr>
Date: Thu, 9 Dec 1999 15:25:02 +0200 (EET)
To: Ron Daniel <RDaniel@DATAFUSION.net>
Cc: RDF Interest Group <www-rdf-interest@w3.org>
Message-Id: <Pine.OSF.3.96.991209150657.32531A-100000@pleiades.csi.forth.gr>

Ron, thanx for your answer. 
However, I am a bit confused by the following examples ...

On Tue, 7 Dec 1999, Ron Daniel wrote:

> Good question. Both behaviors (returning p2 vs. not returning
> p2) would seem to be appropriate at different times. Here's
> a couple of scenarios, assuming p1 is the Dublin Core creator
> element.
> 
> 1) John Doe wants to search an RDF store of bibliographic information
> for works written by his favorite author, J. Smith.  He searches
> for statements where the predicate is dc:Creator and the object
> is "J. Smith".  His database also searches for statements whose
> predicate is known to be a subProperty of dc:Creator.
> 

This means that the subProperties should implicitly included in the
result, right?

> 2) Jane Doe is analyzing the growth of the Dublin Core and its
> mapping to other bibliographic standards. She searches her RDF
> store for dc:Creator arcs and counts them. She also counts occurrences
> of properties such as x:Illustrator that are subProperties of
> dc:Creator.
> 

So, what is the difference in the second example? Do you mean that Jane
Doe will explicitly form separate queries for dc:Creator and
x:Illustrator?

Moreover, I would like an opinion on this case: a resource about J.Smith -
as in your example - is defined to be an instanceOf [c3] in my sample
schema. However, the property p1 (Creator) is defined on this resource,
which is valid since c3 is a subClassOf the domain of p1. Do we need to be
able to specify both:
- That we want resources of any type for which a property has been
(validly) defined (thus include all subclasses of the domain of the
property).
- That we want resources of the specific type that has been specified as
the domain of the property, in the schema definition.

Consider this question in combination with the previous one.

Best regards
Greg   

> 
> I think a query language MUST provide functionality for the
> second usage, and SHOULD provide an operator for the first case.
> 
> Regards,
> Ron
>  
> 
> > -----Original Message-----
> > From:	Greg Karvounarakis [SMTP:gregkar@csi.forth.gr]
> > Sent:	Tuesday, December 07, 1999 8:02 AM
> > To:	RDF Interest Group
> > Subject:	A simple question on RDF Querying ...
> > 
> > 
> > Hello everybody.
> > I would like to have your opinion/comments on the following question:
> > 
> > Suppose that I have an RDF schema like:
> > 
> >                 [C1] ------p1------> [C2]
> >                  /\        /\         /\
> >                  ||        ||         ||
> >                  ||        ||         ||
> >                  ||        ||         ||
> >                 [C3] ------p2------> [C4]
> > 
> > p2: subPropertyOf p1
> > C3, C4: subClassOf C1,C2 respectively.
> > 
> > If a query requests the from and to values of the property p1
> > (metadata instances), do you believe that the from and to values
> > of its subProperty (p2) should also be returned?
> > I would also appreciate any examples where the one case is 
> > preferrable over the other (based on possible applications that
> > you may have in mind).
> > 
> > Thanx in advance
> > Greg
> 
Received on Thursday, 9 December 1999 08:25:31 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:42 GMT