I believe SRW needs a utility index set.
If we have an index, say, 'identifier', in a utility set, we can
qualify/scope it (see note below) however we want. I don't think we
can/should do that with DC.
An OAI record includes an identifer for the resource it describes. It's
proper to search that identifier as dc.identifier. The oai record itself has
an identifier. I don't think it's proper to cast that as a dc identifier,
scoped or not.
Right now we represent unqualified DC and I think we represent it
faithfully. If we want to represent a more complex form (i.e. DC qualified)
we'd still want to do so faithfully, I'm sure we'd all agree (and some of us
might even remember the trouble we got into when we tried to take DC into
our own hands with Z39.50, a long time ago, 1997 I think). Trying to impose
some form of FRBR model would not fit well with DC (someone please correct
me if I'm wrong on this point).
I agree with Rob that we probably should chose a direction here before we go
much further along the lines of an OAI profile. I suggest (for one) that we
develop a utility index set. I don't think it needs to have alot of indexes,
but 'date' and 'identifier' come to mind, and either of these could be
qualified to meet the oai requirements. I think we should either leave DC
alone (unqualified), or leave it to the DC community to come up with
qualifiers that would form a qualified DC set.
(Note: some of this conversation has gone on privately and there is a
suggestion to add a new sort of modifer to cql, that would in effect,
qualify an index. It's not a complete proposal at this point.)
--Ray
----- Original Message -----
From: "Robert Sanderson" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Monday, August 30, 2004 2:44 PM
Subject: OAI profile/Metadata scope
> To summarize a clever idea of Jeff Young (OCLC):
>
> If there was a record creation and record identifier index, plus
> optionally an oai.set scannable index and/or an oai_dc recordSchema, you
> could seamlessly expose the SRW database via OAI.
>
> Jeff's original plan was to have oai.identifier, oai.datestamp and
> oai.set, but recently we've talked about dc.identifier with a scope
> of record metadata (as opposed to OAI's use of 'metadata' meaning record)
>
> oai.set isn't contentious, even though the semantics are similar to that
> of cql.resultSetId, as we don't have any defined semantics for scanning
> that special index (among other reasons)
>
> But recently there has also been the discussion regarding dublin core as
> record schema and dublin core as defining semantics for indexes.
>
> It seems like a very good time to make a decision about all of this, such
> that we don't end up with many many context sets all defining something
> called 'title' or some defining 'author', some 'creator', some
> 'personName' etc. etc. multiple times, once at each metadata level they
> can think of.
>
> Rob
>
> ,'/:. Dr Robert Sanderson ([log in to unmask])
> ,'-/::::. http://www.o-r-g.org/~azaroth/
> ,'--/::(@)::. Special Collections and Archives, extension 3142
> ,'---/::::::::::. University of Liverpool
> ____/:::::::::::::. L5R Shop: http://www.cardsnotwords.com/
> I L L U M I N A T I
|