Subject: Re: DC in Bib1? (Was: GILS Use Attributes merged into Bib-1)
Ray Denenberg (rden@loc.gov)
Date: Thu, 23 Jul 1998 16:49:06 -0400
Message-Id: <35B7A1C2.1B6398FB@rs8.loc.gov> Date: Thu, 23 Jul 1998 16:49:06 -0400 From: Ray Denenberg <rden@loc.gov> To: gils@cni.org Subject: Re: DC in Bib1? (Was: GILS Use Attributes merged into Bib-1)
Kevin C. Marsh <kmarsh@information.org> wrote:
>
> ...
> Dublin Core, on the other hand, was created apparently without reference
> to Bib1 or Z39.50. Ray Denenberg and Rebecca Guenther at the Library of
> Congress released a marvelously helpful document, "Mapping of Dublin Core
> Elements to Z39.50 Bib-1 Search Attributes," on July 22, 1997 (see
> http://lcweb.loc.gov/z3950/agency/copen/dcbib1.html).
Well thanks, but I'm concerned that that document may be
mis-construed (I'm tempted to take it down). It's now more
than a year old, residing in a directory that probably
should have been deleted by now. And to be honest, I'd
forgotten about it. You'll note, it's labelled "second
draft" and there is no more-recent version. Looking at that
document in retrospect a year later, it was intentionally
controversial, primarily intended to stimulate discussion.
And it did stimulate much discussion, at the Copenhagen
meeting last August, and I regret that the discussion hasn't
been better documented. Perhaps we can correct that. I'll
try to put together a "ZIG Commentary" for approval at the
next meeting.
While "publisher" mapping to "publisher" isn't likely to
raise an eyebrow, that's pretty much the only
straight-forward mapping. For example, "other contributor"
to "author": that was fairly soundly rejected, except by
the people who still want to look at DC in straight
bibliographic terms. The reason publisher --> publisher
works is that it was the bibliographic contigent to DC that
insisted that publisher be one of the 15 elements. Noboby
outside the bibliographic community cared much about
publisher. Not so for any other DC element: Even title means
something slightly different to the non-bib DC folks than to
bibliographic people. And in the bib model, the concept of
other contributor is implicity imbedded in Author: you have
first author, second author, third author, ...; the first is
"the" author and the rest could be considered "other
contributor"s, or so Rebecca and I reasoned. But that's not
the way the DC model sees it.
> ... Using this approach it would be easy to simultaneously search by
> Title or Subject across a diverse selection of MARC, DC, CIMI, or
> other data indexed under Z39.50 using Bib1 aqttributes.
Why do you want to use bib-1 to search across
multi-disciplinary databases, if bib-1 is for bibliographic
information, and bibliographic is considered one of many
domains? And in the ZIG, that does seem to be the
prevailing, if not unanimous view, that bibliographic is one
of many domains.
So you ask, why then put the DC attributes in bib-1 at all?
I'll get to that below; the more relevant question is how to
do cross-domain searching. Version 2 of Z39.50 never
anticipated cross-domain searching. The old Z39.50 attribute
architecture (i.e. the pre-attribute-architecture default
attribute architecture) which led to the development and
proliferation of bib-1 does not effectively support
cross-domain searching. So to begin, you have to assume
version 3 and the newly-developed attribute architecture.
Under that architecture, it is planned that there will be an
attribute set which may be named the cross-domain set (or
something similar) or it may be called the Dublin Core set.
Ralph LeVan is looking into that. That set is intended for
cross-domain searching.
We Z39.50 folks, those of us who are not well-versed in
Dublin Core thinking, as well as some who are, are not of
one mind on the question of whether Dublin Core is or is not
intended for cross-domain searching (or searching at all for
that matter). On the other hand we do know that we want
Z39.50 to support cross-domain searching whether or not
Dublin Core is intended to. That's why we haven't decided
yet how to name this new attribute set.
We envision that a community preparing search access points
to its data will distinguish between searches aimed
specifically at that data vs. cross-domain searches where it
is one of many domains being searched. In the first case,
domain-specific access points (attributes) would be used. In
the second case, cross domain/dublin core attributes would
be used. (And these could be the same or different.)
Now to the question "why put the dc attributes in bib-1 at
all?" which I defered because the answer isn't very
interesting: The overall proposed approach to cross-domain
searching assumes version 3, and like-it-or-not there will
be people who want to do cross-domain searching with version
2, so some approach has to be offered. The approach offered
is to put DC attributes in many attribute sets (we, the ZIG,
only have control over bib-1) rather than in a central set.
For bib-1, the choice between a mapping and explicit
attributes was somewhat arbitrary, but putting in explicit
attributes is administratively and politically easier that
trying to get a mapping approved.
I'm sure this attempted explanation isn't entirely
satisfactory, but I hope it helps.
-- Ray Denenberg Library of Congress 202-707-5795 rden@loc.gov
This archive was generated by hypermail 2a16 : Tue Mar 23 1999 - 03:55:43 EST