[OAI-implementers] OAI Resource

[Caroline Arms]

Venu,

I agree with the earlier respondents.  OAI-PMH is a mechanism for
exchanging (but not searching) metadata.  If your local application needs
to hold and support searching for information about people, that is likely
to be outside OAI-PMH entirely.  However, if you are also looking to
exchange metadata about people among applications/services, you may be
able to use OAI-PMH.

Useful metadata about people (whether you call them authors, agents,
parties, or whatever) is going to be different from useful metadata about
document-like information resources.  Even though the DCMI now says that,
'Here an information resource is defined to be "anything that has
identity".' the original elements (used for the OAI mandatory set) were
definitely developed for "document-like" objects.  Squeezing information
about people into an unqualified Dublin Core record is unlikely to be
useful.

As Jeff points out, since OAI-PMH allows you to use other metadata
formats, you can use it to exchange records that describe people if the
parties involved in the exchange can agree on a format.  The mandatory DC
record can be minimal, its only useful purpose being as a conduit to a
"full" record in a more appropriate schema.

Apart from MARC Name Authority Records in the marc21 "slim" schema, I am
not familiar with an XML Schema in common use for describing people.  I
just found 
  http://www.numerata.com/vcardschema.htm
but vCard may not have the elements that are of interest in your
application.

There are at least two more activities that I can think of that are
looking into records for people.  However, neither has reached the stage
of having a schema, as far as I know.

1.  DCMI Agents Working Group
 http://www.dublincore.org/groups/agents/
"Agents" include Creator/Contributor (and possibly Publisher) from the
primary DC Element Set.  

2.  InterParty 
 http://www.interparty.org/
The InterParty project is funded under the European Commission's
Information Society Technologies Programme (IST), to design and specify a
network to support interoperability of party identification (for both
natural and corporate names) across different domains. InterParty builds
on the work of the <indecs> project, one of whose deliverables was a
specification for a Directory of Parties
[http://www.indecs.org/pdf/DirectoryofParties.pdf]. InterParty is not
proposed as a replacement for existing schemes for the identification of
participants in the intellectual property domain (e.g. national library
name authority files or systems oriented towards the needs of rights
licensing) but as a means of effecting their interoperation.
 http://www.interparty.org/

If you really are looking to exchange records about people, perhaps others
on the mailing list know of projects involving appropriate schemas or
element sets.

Caroline Arms                            caar@loc.gov
Office of Strategic Initiatives
Library of Congress
==
Opinions expressed are my own.
==


On Mon, 14 Apr 2003, Young,Jeff wrote:

> I don't see how titles deserve to be separate resources, but I can
> sympathize with your desire to store authors as resources. For example, I
> have an old copy of the LC Name Authority File available that is accessible
> via OAI GetRecord verbs (e.g.
> http://alcme.oclc.org/laf/servlet/OAIHandler?verb=GetRecord&metadataPrefix=m
> arcxml&identifier=oai:laf.oclc.org/LCCN/n78-95332). So, you can retrieve any
> record in the file by substituting the LCCN for that person at the end of
> the URL.
> 
> The biggest problem with this from OAI's point of view is that you can't
> honestly represent these records in Dublin Core (e.g.
> http://alcme.oclc.org/laf/servlet/OAIHandler?verb=GetRecord&metadataPrefix=o
> ai_dc&identifier=oai:laf.oclc.org/LCCN/n78-95332). Is "William Shakespeare"
> the dc.creator? The dc.title? Dublin Core is a bibliographic metadata
> format, and people just aren't bibliographic items. On the other hand, I
> don't claim that this repository is OAI compliant. It's just a convenient
> way to make the MARC21 XML data available to both browsers and automated
> processes.
> 
> If you're really intent on creating records for people, you might consider
> doing something similar. Then, in your research records, you can create
> links from the dc.creator/dc.contributor/dc.publisher, etc, to these records
> via the available URL.
> 
> This brings up another problem, though. There is no place in the Dublin Core
> schema to put these URLs. For example,
> 
> <dc:creator>Shakespeare, William,--1564-1616</dc:creator>
> 
> To get around this, the ETDMS format, for example, extends the Dublin Core
> schema to include a resource attribute.
> 
> <etdms:creator resource="http://...">Shakespeare, William...</etdms:creator>
> 
> If you store your research project records this way, you can always dumb
> them down to Dublin Core by omitting the URL.
> 
> If you do decide to store records for people, I'd suggest that there's no
> good reason to mix them in with your research paper records. Also keep in
> mind that various groups are dealing with schemes that will associate people
> with URIs, so in the long term, you may want to pick a solution that will
> allow you to utilize these services when they become available.
> 
> Jeff
> 
> > -----Original Message-----
> > From: Venugopal R Pally [mailto:pally_reddy@yahoo.com]
> > Sent: Monday, April 14, 2003 2:35 PM
> > To: Simeon Warner; oai-implementers@oaisrv.nsdl.cornell.edu
> > Subject: RE: [OAI-implementers] OAI Resource
> > 
> > 
> > Thank you. As you said, Could you inform me how I can
> > provide this at the service layer ? I have already
> > implemented the OAI considering these research
> > projects as Resources. But it would be of good use to
> > my organization if I can extend it to considering
> > certain other things as Resources. My initial idea was
> > to use the same oai_dc metadataformat as schema for
> > all these resources except that I will use only some
> > of those elements in metadata of these different
> > resources. For example, I need creator element of
> > oai_dc for project but I dont need that element for
> > Author etc. This way I would omit certain elements for
> > these resources. Please suggest me if this is
> > practical.
> > Thanks,
> > Venu.
> > 
> > --- Simeon Warner <simeon@cs.cornell.edu> wrote:
> > > 
> > > I agree with Jeff and feel that overloading the
> > > selective harvesting
> > > mechanisms (sets, metadata formats) with search
> > > functionality is not the
> > > best way to approach these issues. You should either
> > > use a protocol that
> > > supports remote search, or provide that
> > > functionality at the service layer
> > > (think of the OAI repository as one layer down).
> > > 
> > > Cheers,
> > > Simeon.
> > > 
> > > On Mon, 14 Apr 2003, Young,Jeff wrote:
> > > > I'd say the answer is no, you don't want to do
> > > that. OAI isn't a search
> > > > protocol, it's a simple harvesting protocol. If
> > > you really do need to search
> > > > your database by these fields you will need to use
> > > a different protocol such
> > > > a Z39.50 or SRU/SRW and use it to index those
> > > fields from your research
> > > > project records. Also keep in mind that the main
> > > reason people make your
> > > > metadata records available via OAI is so others
> > > (aka service providers) can
> > > > make them useful and searchable in this way.
> > > > 
> > > > Basically, it sounds like you want more
> > > functionality than OAI alone
> > > > provides. Check out EPrints or DSpace if you need
> > > a more complete archiving
> > > > solution.
> > > > 
> > > > Jeff
> > > > 
> > > > > -----Original Message-----
> > > > > From: Venugopal R Pally
> > > [mailto:pally_reddy@yahoo.com]
> > > > > Sent: Monday, April 14, 2003 11:50 AM
> > > > > To: oai-implementers@oaisrv.nsdl.cornell.edu
> > > > > Subject: [OAI-implementers] OAI Resource
> > > > > 
> > > > > 
> > > > > Hi all,
> > > > >    The OAI says that 'resource' is the object or
> > > stuff
> > > > > that metadata is about. So, can resources
> > > include
> > > > > multiple types ? For example, in our case, I
> > > > > identified research projects as resources. But
> > > later I
> > > > > found that harvestors would like to search our
> > > archive
> > > > > based on certain other things like Author, his
> > > Papers
> > > > > etc. This would mean I should consider Authors,
> > > Paper
> > > > > titles also as resources along with research
> > > projects.
> > > > > So, when a harvestor asks for ListIdentifiers,
> > > can I
> > > > > display all of these (Research Projects,
> > > Authors,
> > > > > Paper Titles) ? Or should I use different
> > > > > metadataPrefix for different resources ?
> > > > > Thanks,
> > > > > Venu.
> > > > >