[OAI-implementers] Scientific journal metadata

Tim Brody tim@tim.brody.btinternet.co.uk
Mon, 4 Aug 2003 15:12:02 +0100


(apologies for top-posting!)

I agree with Ann's approach to DC encoding of journal references. I would
simply add that DC applications have to deal with strings, hence can tell
the difference between a URI (http: etc.) and an unstructured string (e.g. a
citation). Including a mix of structured and unstructured identifiers is not
a problem.

Current (?) DCMI terms aka qualified DC:
http://dublincore.org/documents/2003/03/04/dcmi-terms/

Is there an updated XML schema
(http://www.ukoln.ac.uk/metadata/dcmi/xmlschema/) that reflects the current
DCMI terms set?

All the best,
Tim.

----- Original Message ----- 
From: "Ann Apps" <ann.apps@man.ac.uk>
To: <oai-implementers@oaisrv.nsdl.cornell.edu>
Sent: Monday, August 04, 2003 3:23 PM
Subject: Re: [OAI-implementers] Scientific journal metadata


> Dear All,
>
> People from the digital library and publishing communities within
> the Dublin Core community have long been aware that there is no
> obvious place to include the bibliographic information for a resource
> that is a journal article within its own metadata in simple Dublin
> Core. This is really because simple Dublin Core is just that - a core
> set of metadata for general resource description. As soon as you
> want to describe something more domain-specific you need either
> qualifed Dublin Core or even more application-specific metadata.
>
> The discussion about which element to use for bibliographic
> metadata has gone on for many years within the DC community.
> The final recommendation is to use 'identifier' because this
> information identifies the resource. I realise this may look
> inconsistent with URIs within Identifier - but this is simple DC we're
> talking about where the value of a property is just text (you need
> qualified DC to say that it is a URI).
>
> I'm afraid I cannot see that bibliographic information fits into the
> semantics of Type (the nature or genre of the content of the
> resource) though I suppose you could loosely be saying this is a
> journal article.
>
> The problem with everyone making their own choice of element to
> include the bibliographic information is a loss of interoperability. It
> means you have to understand the semantics of a particular
> repository's metadata to process it.
>
> The small amount of development I've done with OAI-PMH has
> made me question the value of passing simple DC records. It
> seems that the richness of the information often gets dumbed-down
> to probably be of little use. Simple DC is really just for basic
> interoperability not for passing richer descriptions.
>
> If you were to use qualified DC, there is now a property
> dcterms:bibliographicCitation to hold this information. It is a
> refinement of Identifier. At present there is no recommendation for
> the content of this element beyond simple text. But we are hoping
> to define a Dublin Core Structured Value to hold this information in
> a more machine-processable way. This has actually been more-or-
> less defined for some time and is probably already in use. (See
> http://www.dublincore.org/groups/citation/ for more information).
>
> Some of the proposals for the OpenURL Framework may also
> provide a solution, and the DCMI Citation working group will be
> investigating that.
>
> Best wishes,
> Ann
>
>
> From:           Michael Nelson <mln@cs.odu.edu>
> To:             oai-implementers@oaisrv.nsdl.cornell.edu
> Subject:        Re: [OAI-implementers] Scientific journal metadata
> Date sent:      Fri, 1 Aug 2003 20:04:34 -0400 (EDT)
>
> >
> > and just for contrast, I put bibliographic information in dc:type and
> > use dc:identifier strictly for URIs.
> >
> > see, for example:
> >
> > http://ntrs.nasa.gov/?verb=GetRecord&metadataPrefix=oai_dc&identifier=
> > oai:ltrs.larc.nasa.gov:NASA-aiaa-97-1321
> >
> > my rationale...  reading the descriptions at:
> >
> > http://www.dublincore.org/documents/dces/
> >
> > I can see where bibliographic info could go in dc:identifier.  But I
> > hate to have both free text & URIs in dc:identifier, and dc:type looks
> > pretty good. And you make an argument that the bibliographic info
> > tells you a good bit about resource type.
> >
> > regards,
> >
> > Michael
> >
> > On Fri, 1 Aug 2003, Simeon Warner wrote:
> >
> > >
> > > For arXiv, I put bibliographic refereces in dc:identifier elements,
> > > e.g.
> > >
> > > <dc:identifier>Prog.Theor.Phys. 101 (1999) 1155-1164</dc:identifier>
> > >
> > > in
> > >
> > > http://arXiv.org/oai2?verb=GetRecord&metadataPrefix=oai_dc&identifie
> > > r=oai:arXiv.org:hep-th/9901001
> > >
> > > I do not think dc:source is correct since that is for a "A Reference
> > > to a resource from which the present resource is derived". I assume
> > > in your context, as in arXiv's, the bibliographic reference is a
> > > reference to the resource that the record describes.
> > >
> > > Cheers,
> > > Simeon.
> > >
> > >
> > > On Fri, 1 Aug 2003, Jeremy Keeler wrote:
> > > > Greetings,
> > > >
> > > > I'm implementing an OAI repository of scientific journals, but I'm
> > > > not sure what to do with the volume, issue and page metadata for
> > > > the unqualified dc format. I've read a few debates over this but
> > > > I've not seen where it's been solved.
> > > >
> > > > I had thought about adding a bibiographic record in dc.source, as
> > > > in... <dc:source>Journal of Some Research 120(3),
> > > > 133-165</dc:source>
> > > >
> > > > But I can see arguments against this.
> > > >
> > > > It seems logical to acknowledge that this metadata can not exist
> > > > in unqualified dc format, and encourage harvesters to use our
> > > > more-specific metadata format. This would mean unqualified dc from
> > > > our repository would be fairly useless.
> > > >
> > > > Is this the accepted way to expose scientific journals through
> > > > OAI?
> > > >
> > > > Thanks,
> > > > Jeremy Keeler