> I can see how, given this model, I can (programmatically) get from dc:title
> to the resources described in "#deDE_1" and "#deDE_2" (by looking for an
> rdf:Statement with an rdf:Subject of dc:title), but how would a program get
> from dc:title to the resource described in "#deDE_0"?

Harry, Roland, (attn: Shigeo and Eric),

This discussion touches on a crucial feature of RDF --
the ability to overlay annotations, in this case translated
labels and definitions.  It moves us further towards goals
that were recognized and discussed at a series of meetings
of the Working Group for DC in Multiple Languages in 1997
and 1998 (see [1] and [2]).  In this somewhat long posting,
I lay out two requirements that may not be addressed by the
current approach, but could be.  These are summarized as two
questions at the end.

The position paper of mid-1998 [1] recognized that "versions"
(preferred over "translations" at the time) of Dublin Core
elements in various languages should share a single namespace
(Point 4) and proposed a now-outdated RDF encoding of the
Title element in French (Point 6).  Importantly, the paper
also suggested that: "Versions of the Dublin Core in other
languages should cite the specific English version on which
they are based (eg, 1.0, 1.1, 2.0...), as that canonical
version will evolve over time".  Shigeo, his colleagues at
ULIS, and I explored the implications of versioning in a
paper early last year [3].

In that paper, we proposed some general requirements with
regard to versioning translations (Section 4):

-- that for long-term access to legacy metadata as well as
   for the needs of software developers it should be possible
   to reconstruct the exact labels and definitions in use at
   a particular point in time (Points 1 and 3);

-- that the set of terms should grow incrementally, not by
   infrequent, numbered releases such as "Dublin Core 2.0"
   (Point 5) -- the paper summarizes discussion of this issue
   in DC-AC in 1999 (Section 5);

-- that namespace URIs remain stable through successive
   versions (Points 2 and 4), which is now addressed by
   the DCMI Namespace Policy;

-- and, importantly, that the translation of a term into
   Japanese needs to reference a unique identifier for the
   specific historical version translated (Points 6 and 7).
   This requirement emerged from our sense that inevitably,
   translations would lag somewhat behind the "canonical"
   English version, even if the resulting differences of
   wording were to have no significant impact on applications
   (see the DCMI Namespace Policy).

This need to uniquely identify a historical version of a term
is roughly "prototyped" in the new Usage Board document -- for
example, consider the two historical versions of "dc:title":

    VMS-ID:        title-002
    URI:           http://purl.org/dc/elements/1.1/title
                   [-- Label, Definition, etc of dc:title, DCMES 1.1 --]
    Replaces:      http://dublincore.org/usage/terms/dc/#title-001

    VMS-ID:        title-001
    URI:           http://purl.org/dc/elements/1.1/title
                   [-- Label, Definition, etc of dc:title, DCMES 1.0 --]
    isReplacedBy:  http://dublincore.org/usage/terms/dc/#title-002

.. where the state of the Title term in DCES 1.0 is
(prototypically, at least) "uniquely identified" as "title-001"
-- for example, try clicking on the mirrored document at
http://www.gmd.de/People/Thomas.Baker/usage/terms/dc/#title-001.

With these issues in mind, I have two questions about the
proposed RDF syntax below:

1) Could "translation annotations" of this form in principle
   be distributed over -- and harvested from -- multiple
   Web sites?  The whole thrust of the 1997-1999 discussion
   was that German translations would be maintained by
   German-speaking organizations.  The alternative -- that
   DCMI would hold and maintain all translations centrally --
   seemed then and still seem both impractical and politically
   undesirable.

2) Could an RDF statement be added to point to the uniquely
   identified historical version of an element?  For example,
   could a translation of "dc:language" point to the DCES 1.0
   version at http://dublincore.org/usage/terms/dc/#language-001
   (or an equivalent).

Tom

[1] http://dublincore.org/groups/languages/mr-19980525.shtml
[2] http://dublincore.org/groups/languages/
[3] http://avalon.ulis.ac.jp/~sugimoto/RPs/saint.pdf

On Wed, Jun 05, 2002 at 04:25:35PM -0400, Harry Wagner wrote:
> > > <!-- Begin: Title Declaration -->
> > > <rdf:Property rdf:about = "&dcns;title"/>
> > >
> > > <rdf:Description rdf:about = "&dcns;title" bagID="deDe_0">
> > >   <rdfs:label rdf:ID="deDE_1" xml:lang="de-DE">Titel</rdfs:label>
> > >   <rdfs:comment rdf:ID="deDE_2" xml:lang="de-DE">Titel der
> > > Quelle; der vom Verfasser, Urheber oder Verleger vergebene
> > >   Namen der Ressource</rdfs:comment>
> > > </rdf:Description>
> > >
> > > <rdf:Description rdf:about="#deDe_0">
> > >         <dc:description xml:lang="de-DE">
> > >           Sprachabhaengiger Teil der Deklaration eines
> > >           DC-Terms in Deutsch.
> > >         </dc:description>
> > >         <dc:description xml:lang="en-US">
> > >           Language dependent part of the deklaration of a DC
> > >           Term in German.
> > >         </dc:description>
> > >         <dc:publisher>DCMI</dc:publisher>
> > >         <dc:source rdf:resource="http://www.dublincore.org/documents/dces"/>
> > > </rdf:Description>
> > >
> > > <rdf:Description rdf:about="#deDE_1">
> > >         <dc:creator>Diann</dc:creator>
> > >         <dc:description xml:lang="de-DE">Zuordnung eines
> > >          deutschen Labels zu einem DC Term</dc:description>
> > >         <dc:description xml:lang="en-US">Association of a
> > >           German label to a DC term</dc:description>
> > >         <dc:publisher>DCMI</dc:publisher>
> > >         <dc:rights xml:lang="de-DE">
> > >           Die publizierende Organisation uebernimmt keine
> > > Gewaehr fuer die Richtigkeit der Zuordnung des deutschen Ausdrucks.
> > >         </dc:rights>
> > >         <dc:rights xml:lang="en-US">The publisher disclaims
> > >          any responsibility for the correctnes of the association
> > >          of the German expression.
> > >         </dc:rights>
> > > </rdf:Description>
> > >
> > > <rdf:Description rdf:about="#deDE_2">
> > >         <dc:creator>Walter</dc:creator>
> > >         <dc:description xml:lang="de-DE">Zuordnung eines deutschen
> > >          RDFS Kommentars zu einem DC Term</dc:description>
> > >         <dc:description xml:lang="en-US">Association of a
> > >           German RDFS comment to a DC term</dc:description>
> > >         <dc:publisher>DCMI</dc:publisher>
> > >         <dc:rights xml:lang="de-DE">
> > >           Die publizierende Organisation uebernimmt keine
> > >           Gewaehr fuer die Richtigkeit der Zuordnung des deutschen Ausdrucks.
> > >         </dc:rights>
> > >         <dc:rights xml:lang="en-US">The publisher disclaims
> > >           any responsibility for the correctnes of the association
> > >           of the German expression.
> > >         </dc:rights>
> > > </rdf:Description>

--
Dr. Thomas Baker                                [log in to unmask]
Institutszentrum Schloss Birlinghoven          mobile +49-171-408-5784
Fraunhofer-Gesellschaft                          work +49-30-8109-9027
53754 Sankt Augustin, Germany                     fax +49-2241-14-2619