Re: Dublin Core, the Primer and the Model Theory

On 2002-05-16 14:51, "ext Dan Connolly" <connolly@w3.org> wrote:


> But keep in mind there are at least a few implementations
> that we break if we go the other way:
> 
> # how does existing RDF software handle this datatypes test?
> http://lists.w3.org/Archives/Public/www-rdf-interest/2002Jan/thread.html#199
> 
> 
> Responses indicate RDQL, rdfql, Squish, and Euler think literals
> are tidy. 

The general impression that I got from all the responses to
the query tests were that, with rare and limited exceptions,
query engines simply weren't doing datatyping at all, insofar
as any global range or other typing knowledge is concerned.

I.e., in the absence of any actual datatyping mechanisms that
would e.g. associate a datatype with the inline idiom, one
can do nothing but treat literals as tidy, as there's no further
information in the graph by which you could make any comparisons,
and since folks generally want to achieve useful comparisons,
they resort to string comparison. That doesn't though mean
that it is really valid or reliable to presume that the meaning
of those literals which are string-equal is actual globally
consistent -- and the title="10"/age="10" example bears that out,
I think.

In addition, it is important to point out that applications which
associate datatyping semantics with specific properties (e.g.
CC/PP, etc.) all have untidy literals.

So I don't find the present treatment afforded to literals by
the current RDF query engines to be very strong motivation for
adopting tidy literals, but is just an artifact of the lack of
any standardized datatyping mechanism that such engines could
utilize.

Cheers,

Patrick

--
               
Patrick Stickler              Phone: +358 50 483 9453
Senior Research Scientist     Fax:   +358 7180 35409
Nokia Research Center         Email: patrick.stickler@nokia.com

Received on Thursday, 16 May 2002 15:45:04 UTC