[This is posted to the ZIG, and cc'd to meta2. It may seem out-of-context
to meta2; recenty the question of mapping DC elements to Z39.50 bib-1
attributes (search points) has come up.]
As a matter of fact there is an effort in progress to create a mapping
between DC and bib-1. The proposed mapping is not yet complete but since
this has come up I've included below how far we've gotten.
We also have a few questions. One one hand it is clear why we would want
a DC/Z39.50 mapping for retrieval, i.e. why to map DC onto tagSet-G:
People talk in various contexts about retrieving Dublin Core records, or,
in the case of CIMI, a structured Z39.50 metadata element containing DC
elements as its subelements. But it isn't quite as clear why we want a
mapping of Dublin Core to Z39.50 search access points. Can someone please
articulate this? And if so, do all DC elements apply or only some? For
example, take the DC element "rights". Is it intended to search within the
"rights" field, or is "rights" perhaps a DC element that doesn't map to a
search access point. More questions are embedded below, and we would
appreciate answers and comments.
Anyway as I said, it's premature to list the mapping now, but this is as
far as we've gotten:
There are four categories of DC element (in terms of their potential
mapping to bib-1):
1. A number of DC elements map nicely:
DC Element bib-1 Use attribute
---------- -------------------
Title title (4)
subject/keyword subject (21)
descriptor abstract (62)
creator/author author (1003)
publisher publisher (1018)
resource type material type (1031)
language code-language (54)
2. A few map, but not cleanly:
Date
Resource Identifier
Discussion: Date might map to a number of bib-1 attributes, based
on what it is the "date of" -- e.g. date of publication, date last
modified, date added to database. We don't completely understand
the DC semantics of Date -- i.e. "date resource available in
current form".
Resource Identifier may map to the combination: Use -- DocId,
structure -- urx. This raises the question: what is the purpose of
searching on a Resource Identifier? Is it for known item
searching, or actually searching for records containing a given,
say, URL?
3. Some don't map:
format
contributor
source
Discussion: These three may be potential extensions to bib-1.
4. Some may not be meant to map:
relation
coverage
rights
Discussion: relation is complex and listed in DC as experimental,
so perhaps it is premature to expend too much thought on it
(though perhaps it would be the province of the Collections
attribute set).
Coverage, perhaps, should be within the province of an
attribute set other than bib-1. Maybe Geo, CIP, or GILS.
For rights: (mentioned above) is it intended that you be able
to search within a "rights" field?
Ray Denenberg
Library of Congress
202-707-5795
[log in to unmask]
----------------------------- Note follows -----------------------------
Approved-By: Mike Taylor <[log in to unmask]>
Message-Id: <9706171041.AA03476@kevin>
Date: Tue, 17 Jun 1997 11:41:56 +0100
Sender: "Z39.50 Implementors Workshop" <[log in to unmask]>
From: Mike Taylor <[log in to unmask]>
Subject: Mapping Dublin Core to tagSet-G
Comments: cc: [log in to unmask], [log in to unmask]
To: Multiple recipients of list Z3950IW <[log in to unmask]>
Dear Z39.50 Implementors (again!),
One of the projects we're working on involves search for and
retrieving Dublin Core elements through the Z39.50 protocol. We're
aware that useful work has already happened on the retrieval side --
Dublin Core elements have been mapped onto tagSet-G elements in a
proposal which, for anyone else who's interested, can be found at:
http://lcweb.loc.gov/z3950/agency/aprilzig/tags.html
The other half of the equation, of course, is searching. Is anyone
out there working on a similar mapping of Dublin Core elements onto
BIB-1's USE attributes? or onto any other well-known set of access
points? Even if this is only "work in progress", it would be very
useful for us to see the state it's in now.
We need to start working with _something_ RSN, however deficient, so
if no-one else is doing this work, I may be reduced to making up a
mapping myself (despite lacking all the necessary background!) and
inviting people to take pot-shots at it. (I'm told that "this kind of
thing happens more often that I can tell you.")
Thanks in advance -- and mild apologies for flooding the list with so
many questions recently. (But that's what it's for, right?)
Mike Taylor
System Simulation Ltd.
[log in to unmask]
Tel. +44 171 836 7406
|