Re: requesting XML records

On Wed, 26 Mar 2003, Mike Taylor wrote:

...

> > However, it is worth noting that the simple DC schema is
> > specifically intended for importing into other,
> > applicationp-specific schemas.  Therefore it does not define a
> > container element for the DC record - it is up to the application to
> > define that.
>
> ... which, to the eye of the layman, is very similar to not having
> defined a schema :-)  But, OK, point taken.

A more Z39.50 oriented metaphore would be to say that it is very much like
having defined tagset-g (or tagset-2). A hit should normally also contain
tagset-m related stuff (relevance rank or whatever administrative
information about a record a gateway or a user might need). So if going
this way one need to define schemas for those as well.

Now, please let me diffuse away from the subject.

The elements of a single hit has to be grouped inside an element, and
services using the bath profile would obviously use dc-record for that,
but others perhaps rather use oai_dc:dc. OAI put inidividual records
inside an envelope


 <ListRecords>
  <record >
    <header >

    	... some tagset-m like stuff

    </header>
    <metadata >
      <oai_dc:dc  xsi:schemaLocation="xsi stuff"
	xmlns:stuff="other uri Stuff">


	... some tagset-g like stuff

      </oai_dc>
    </metadata>

  </record>
 </ListRecords>

The Bath profile does this in an entirely different way. Obviously the
both ways of doing it can easily be built on top of the same service,
but since I'm lazy I'd like to have my server to behave OAI-PMH like
when queried OAI-PMH like, but I want that behavious to be a special
case of ZURL. (Put another way: Why should I need to build different
interfaces for export and search?)

Now, the kind of data you may search this way are almost
trivial. Trivial to the extent that it does not stimulate my
fantasy. Consider a database containing METS, EAD or Master records
for digital library objects, archival descriptions or medieval
manuscripts, respectively. (or perhaps all three kinds of metadata)
Most likely, a result set should contain not only simple bibliographic
(or archival) references but also navigational information -- tags
very much like those defined in the now defunct (I think) collections
profile.

The problem of delivering XML search results will be even more complex
when searching XML documents, TEI, DocBook, OpenOffice or whatever.

People at W3C would say that these are not Z issues but X Query
ones. That may or may not be correct. But I think its not, because X
Query is to closely linked to the syntactic peculiaries of each
document type. To whom does these problems belong?


Yours

Sigge

Received on Wednesday, 26 March 2003 05:56:35 UTC