> From [log in to unmask] Tue Dec 10 10:36 MET 2002
> From: "Patrick Stickler" <[log in to unmask]>
> To: "ext Roland Schwaenzl" <[log in to unmask]>,
> <[log in to unmask]>
> Cc: <[log in to unmask]>
> Subject: Re: RDF typed literals and DC encoding schemes
> Date: Tue, 10 Dec 2002 11:35:03 +0200
> MIME-Version: 1.0
> X-Priority: 3
> X-MSMail-Priority: Normal
> X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400
> X-OriginalArrivalTime: 10 Dec 2002 09:35:02.0814 (UTC) FILETIME=[6A5C5BE0:01C2A02F]
>
>
>
> [Patrick Stickler, Nokia/Finland, (+358 40) 801 9690, [log in to unmask]]
>
>
> ----- Original Message -----
> From: "ext Roland Schwaenzl" <[log in to unmask]>
> To: <[log in to unmask]>
> Cc: <[log in to unmask]>; <[log in to unmask]>
> Sent: 09 December, 2002 17:35
> Subject: Re: RDF typed literals and DC encoding schemes
>
>
> >
> >
> > Dears,
> >
> >
> > probably i missed quite some of the discussion.
> > German Telecom managed to keep the Universities
> > internet connection cut off for three days.
> >
> >
> > Just could fetch mails from Patrick coming in from
> > today.
> >
> > There is IMO some difficulty with assigning URIs for
> > third party maintained and developed classification
> > systems.
> >
> > DCMI in general will not have any knowledge about,
> > which codes a classification system is using and
> > in particular DCMI will not by itself know
> > that D08.586.682.075.400 is an admissible (literal)
> > code for MESH.
> >
> > I'm not at all sure, what classification system
> > and thesauri developer would think about DCMI
> > re-coding their stuff with DCMI owned URI's.
> >
> > I can imagine some touchy about.
> >
> >
> >
> > Aside of politics:
> >
> >
> > Is there a mechanism in RDF, which would allow
> > DCMI to say, that a URI, which has
> >
> > http://purl.org/dc/terms/MESH/ as start denotes
> >
> > something, which has some relation with MESH ?
> >
> >
> > When i remember correctly aboutEachPrefix
> > became deprecated for RDF-core found it useless.
> >
> >
> > Cheers,
> > rs
>
> Well, one alternative to distinct URIs for each MESH code
> would be to define MESH as a datatype, and treat the code
> strings as lexical forms denoting code values.
>
> This would be something akin to xsd:anyURI, where there is
> a lexical pattern that all lexical forms must conform to
> but there is (technically at least) an infinite number of
> possible lexical forms.
>
> Thus, rather than
>
> <http://purl.org/dc/terms/MESH/D08.586.682.075.400>
>
> one would have
>
> "D08.586.682.075.400"^^<http://purl.org/dc/terms/MESH>
>
> or in RDF/XML
>
> <some:property rdf:datatype="http://purl.org/dc/terms/MESH">
> D08.586.682.075.400
> </some:property>
>
> which would denote whatever value the lexical form "D08.586.682.075.400"
> maps to as defined by the MESH "datatype".
>
> Though, my own preference would be to simply use a URN where, again,
> the lexical grammar is defined for all suffixes following the URN scheme
> name, and the semantics of those complete term URIs left up to the
> MESH folks (or MESH savvy tools) to validate. E.g.
>
> <urn:MESH:{ANY VALID MESH CODE}>
> <urn:MESH:D08.586.682.075.400>
>
> or (even better ;-)
>
> <voc://dublincore.org/MESH/{ANY VALID MESH CODE}>
> <voc://dublincore.org/MESH/D08.586.682.075.400>
>
> etc.
>
> (regarding the voc: scheme, see attachment)
>
Dear Patrick,
i understand, what you want, but IMO you should discuss
with the MESH owners about URIification of their
classification codes.
An alternative would be to put some URIification
proposal to the rdf-core drafts.
I can't see how it would improve interoperability
every metadata concerned organization to issue
their own methods about URIification of
classification codes.
In fact there might be sticky details about
escaping in general. I don't know whether
this could happen with MESH.
I'm not sure, whether the rdf-core drafts
say, that URIrefs, which i understand
now are allowed of xml-datatype anyURI,
will match in escaped and unescaped form
in an RDF graph.
Certainly there is no method in RDF to have
an equation http://foo.org/MESH/713 =
http://muh.org/mesh/713
A really bad idea i think would be to "recommend"
to applications to rename xml-simple-datatypes
in a rather arbitrary way.
In a heterogenous environment like it is with
metadata we would end up like we did with
the HTML meta-tag convention to append
refinements to dc15 following a dot.
This has produced quite some bad metadata
modelling in the past.
To reinvent it for schemes in RDF -
i can't see the gain.
Cheers,
rs
>
> Cheers,
>
> Patrick
|