----------------------------------------------------------------------
2006-02-06
From: General DCMI discussion list on behalf of Diane I. Hillmann
Sent: Mon 06/02/2006 13:48
To: DC-GENERAL@JISCMAIL.AC.UK
Cc:
Subject: Fwd: Issues in Applying RDA in Non-MARC Metadata Communities
Folks:
At the recent American Library Association meeting in San
Antonio, I attended meetings concerned with the new Resource
Description and Access (RDA) standard being developed as
a replacement for the Anglo-American Cataloging Rules. The
opinions represented below are mine only, intended for the
purposes of inciting discussion about the issues.
Regards,
Diane
--- begin forwarded text
Date: Mon, 6 Feb 2006 08:45:39 -0500
To: RDA-L@INFOSERV.NLC-BNC.CA
From: "Diane I. Hillmann" <dih1@cornell.edu>
Subject: Issues in Applying RDA in Non-MARC Metadata Communities
Cc:
Bcc:
X-Attachments:
Issues in Applying RDA in Non-MARC Metadata Communities
Diane I. Hillmann
Jan. 31, 2006
Having come rather late to the RDA discussion, I recognize
that the points I bring up in this document may not be the
comments that CC:DA or JSC expects or desires at this stage,
but I hope the concerns expressed may be useful nonetheless.
First, I'd like to describe the issues I see in applying the
underlying assumptions made in RDA (most of which emanate from
traditional cataloging practices) to the world of Non-MARC
Metadata (NMM) Communities. Clearly, these issues are less
apparent to traditional libraries whose preponderant exposure
to digital resources are digital versions of published
materials, but once outside that familiar boundary the
environment is much less comfortable.
I believe the primary issues that concern me lie in the
following areas:
Transcription as Identification
In the world of traditional cataloging and static published
resources, the notion of consistent transcription as an
important method to assure predictable access, from a variety
of agencies handling exactly the same resources, made a great
deal of sense.
However, digital resources carry no such assumption of
stability-change is part of the package. In that environment,
relying on use of consistently transcribed information as
the primary method of identifying a resource makes much less
sense. Resources in this environment are most often unique,
and usually identified by a numeric or alpha-numeric string.
In traditional cataloging, such identifiers are also used, of
course: ISSNs and ISBNs are the most obvious examples, but they
are generally not the primary identification of the resource.
As we all know, the current methodologies for identifying
digital resources uniquely and unambiguously are still in flux
and almost no one is satisfied with the current situation.
But whatever the ultimate answer, it will not rely on
transcription, nor will decisions about what constitutes a
"new" resource likely be susceptible to the rules defined for
editions or versions. It should also be noted that the gold
standard of infallible identification in a metadata description
is not always necessary for digital information, where the
resource itself can often be viewed easily and quickly.
Reliance on Notes
Oftentimes, the RDA (like traditional cataloging) herds
catalogers to make decisions about what is "primary" or
"secondary" and relegates the latter to the notes area.
This is a significant problem for many NMM communities,
who may either have no place to put this kind of descriptive
"notes" or who rely on repetition of elements (with or without
a notion of order) to capture information of the same kind
within a single description, thus focusing more on access
than descriptive integrity.
In most delivery systems for metadata (including OPACS, it
must be noted), only the information in a small number of
specified fields is actually displayed to the user (and we
know few users actually look at full records). Additionally,
because notes can contain so many different categories of
information, they may not even be indexed (when they are,
only as keywords). For systems using NMM, notes information
is even less likely to be displayed, and may indeed be entirely
ignored, since its "human-friendly" character makes it useless
for machine processing and marginal for access.
Reproductions
I brought up the issue of reproductions on the RDA-L list
and was dismayed to see how many catalogers were still
trying to make the case for describing an original and a
reproduction on the same record. If FRBR is truly underlying
RDA, I believe this bullet must be bitten firmly and these
practices explicitly marginalized within the context of the
rules. In an environment where metadata of different formats
created using different rules (or no rules) must be shareable,
these residual practices keep us all from benefiting from
our common enterprise.
Yes, it is certainly true that most vendor systems do not
display multiply versioned resources acceptably, but we
undercut the usefulness of our data by manipulating it to
overcome system inadequacies; rather, we should address those
problems with our vendors.
Source of Information
Specification of sources of information from which to record
information grew logically from the reliance on transcription,
the goal being consistency. Vital to this approach is the
idea that resources have commonly identified and named parts
that are similar within a specific category of materials,
something that is not generally the case in the digital world.
Similarly, notions of whether information comes from the
item itself or is supplied from somewhere else are often
less important in NMM communities, even those who still deal
primarily with physical, published items. In ONIX for example,
information about the author (from the book jacket, reviews,
or other marketing sources) is specifically tagged based on
the function of the information, and it's often not explicitly
descriptive in nature.
Future Considerations
As I mentioned in my comments at the Monday CC:DA meeting
at ALA, we may increasingly be thinking less about the
cataloging record as the lowest unit of description and more
as the "statement" as the optimal unit. In that context,
"Who says?" or "When said?" or "In what language?" is likely
to be more important information to know in order to manage
the information than where in a resource the information was
found, and the current RDA doesn't support these notions at
all. I suspect we'll begin to see this change in thinking
more as we discuss common authority files, where explicit
specification of language and form of heading are critical to
making appropriate choices for usage in different catalogs,
in the context where the concept of an individual "statement"
has already taken root.
Some of these attributes are easier to manage outside of MARC
(XML, for instance, supports language specification at various
levels), but it's really important that we start thinking
along those lines sooner, rather than later.
The ideal of the current RDA still seems to be the anonymous
cataloger acting objectively using a commonly understood
set of rules, providing consistent records suitable for
sharing. Clearly, the sharing and integration pieces are
still critically important, but we may not be able to afford
the levels of consistency and predictability that we've had
in the past. Other mechanisms may be available to improve
access in ways we don't understand fully at the moment, but
we should probably at least explore some of the possibilities
at this juncture.
I'm not entirely sure how to where to go from here, but
it might be useful to examine some strategies whereby the
most basic level of RDA instruction might be more generally
useful outside the traditional library environment, given
the dissonances noted above.
----------------------------------------------------------------------
2006-02-06
On Mon, Feb 06, 2006 at 06:13:13PM -0000, Robina Clayphan wrote:
> This message is to ensure that you have all seen the message
> sent by Diane Hillmann to the DC-General and RDA lists
> (see below). Diane attended the CC:DA meeting at ALA with
> a DC hat on and raises some fundamental points about RDA and
> non-MARC- metadata creation that we could usefully consider.
>
> As a non-AACR person I have been struggling to review
> the current draft of RDA from a DC perspective. There is
> a mismatch that is not easy to express but which Diane
> has begun to articulate. What I felt was the need for
> a specification of DC requirements for content level
> rules in order to assess RDA against those requirements.
> This specification of requirements will need to be based on
> a statement of assumptions or principles which this group,
> possibly together with other DC groups, should undertake
> to create. Diane's paper points the way to such a statement.
> I would be interested to hear you thoughts on this.
Dear all,
I have spoken with Robina about the DCMI role in providing
input to the revision of RDA. We agreed that it would be
useful to have a two-page position paper with something like
the following:
-- DCMI principles (e.g., one-to-one, "values" and "value
representations" in the Abstract Model, etc).
-- Requirements: what do DC implementers look for from
a standard like RDA?
We feel encouraged by the willingness (as I understand
it) of the RDA maintainers to make it easier for users
to extract selected sub-sets of the entire standard for
particular applications. I'm wondering if we (i.e., DCMI)
could focus in on a sub-set of rules of likely applicability
to DC metadata implementers instead of trying to review RDA
as a whole.
Ideally, we would have a draft two-pager on the table for the
Usage Board meeting in Seattle for discussion and, possibly,
some sort of official UB approval. Diane's posting to
dc-general [1, and copied below] provides a start.
Does that seem like an achievable objective, and are there
any volunteers?
Tom
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0602&L=dc-general&P=56
----------------------------------------------------------------------
Date: Fri, 10 Feb 2006 22:39:58 -0500
From: "Diane I. Hillmann" <dih1@CORNELL.EDU>
Subject: Re: Input to RDA process
Comments: cc: "Clayphan, Robina" <Robina.Clayphan@bl.uk>
To: DC-USAGE@JISCMAIL.AC.UK
Tom:
I think your second point is the strongest need,
at this stage. One of the legacies of AACR is the
important distinction between "description" and
"access." I think that to some extent the DCAM
principles are not particularly relevant (except
for "one-to-one) to the cataloger notion of
"description", though there will likely be more
relevance when the discussion moves from the
"descriptive" rules to the "access" portion. To a
great extent the "one-to-one" rule is reflected
in RDA by the adherence (such as it is) to FRBR
principles.
Reviewing RDA as a whole, particularly for
someone not conversant with AACR2, is likely to
be a pretty frightening experience. There is an
expression of principles at the head of the
current Part I, which is only marginally helpful,
since it does not address some of the hidden
assumptions that are carried over from AACR2
(some of which I tried to address in my comment).
I suspect one of the most useful things that we
could do, as a group, is bring to the discussion
our experiences with interoperability issues and
the uses (and misuses) we find when applying
machine-based understanding to human created data.
We might also try and bring in Wayne Hodgins to
the discussion, who has also undertaken to
participate on behalf of LOM (I didn't get to
speak to him at any of the meetings but he was in
attendance at a few of them).
----------------------------------------------------------------------
Date: Thu, 23 Feb 2006 11:32:57 -0500
Reply-To: "Diane I. Hillmann" <dih1@CORNELL.EDU>
Sender: General DCMI discussion list <DC-GENERAL@JISCMAIL.AC.UK>
From: "Diane I. Hillmann" <dih1@CORNELL.EDU>
Subject: Re: [RDA-L] NLM/MLA View on Critical Deficiencies of RDA
Comments: To: Resource Description and Access / Resource Description and Access
<RDA-L@INFOSERV.NLC-BNC.CA>
To: DC-GENERAL@JISCMAIL.AC.UK
Alice, et al.:
I'd like to comment briefly on the contents of your document, which I
agree with in principle and disagree with strongly in some specifics.
To quote the document:
"As stated previously, NLM remains dismayed that, as presently
constituted, RDA represents a largely a cosmetic revision, rather
than a high-level reexamination of content. In order to make the huge
investment that adopting RDA entails (e.g., changes to documentation,
training efforts, programming changes, etc.), we stress that libraries
need a document that addresses critical bibliographic issues/problems,
some of which have plagued libraries since AACR2 itself was published,
as well as broadens the bibliographic perspective to cover today's digital
environment. We concur with others, specifically Paul Weiss, who stated
that it is appropriate and important to raise issues in the ALA response
to the JSC that need to be addressed either from AACR2 or in the current
RDA draft. In its strict constructionist view, we believe that the JSC
is failing to meet its own stated goal that 'RDA =8Abe a new standard
for resource description and access, designed for the digital world.'"
I think this point is absolutely correct, and concur with the concern that
the rush to meet deadlines has to a great extent overpowered discussion
of first principles. We know our world has changed and will continue to
change dramatically, but you'd hardly know that from the highly specific
discussions going on at present.
That said, I see a real disconnect between the general and the specific
recommendations in the document, as it goes on to suggest:
o Incorporation of rules to allow use of a single record to describe
multiple entities in certain instances
Since the beginning of the rule revision process, NLM has stressed the
need to provide guidance on the critical cataloging question of multiple
versions. We continue to view it as imperative to incorporate, in some
fashion, the highly-used and pragmatic approach of using a single record
for multiple entities in certain instances.
o Treatment of the description of a reproduction based on its
original
We included this in our comments in
confluence as a major problem and concur with the statement of others
that this is a long-standing practice that likely will continue even
if RDA cements its current position. We contend that, at a minimum,
RDA should provide an option that would recognize what is now and what
is anticipated to be U.S library practice in the future.
o Rules on Changes Requiring a New Description
=46irst, we believe that RDA should address what changes require a new
description for all modes of issuance. The rules in the current draft
relate solely to serials.
The first two points seem to me to contradict the general thrust of the
earlier paragraph. We seem to be in deep denial that the "highly-used
and pragmatic approach" is in fact a serious kludge, and one that will
not allow us to either share data effectively or take advantage of some
of the =46RBR-based research on improved display of manifestations and
expressions. Consider for a moment some of the work OCLC has done with
=46ictionFinder--all based on the notion that there is for the most part
sufficient data in the records to associate them in this way and create
a much more useful display than current library vendors now offer.
In my opinion, the more we cling to these practices, the more we hobble
ourselves in the future. Instead, we should be examining the conclusions
of projects such as FictionFinder and others to determine what should
be included in records to allow the associations we need to create to
make useful displays possible, and then =46LOG OUR VENDORS to provide
that capability. Instead we insist on manipulating our essential data to
overcome the deficiencies of our software, and thus mortgage our future
options. This state of affairs is, in my opinion, a direct outcome of
the lack of discussion on first principles that we've skated over in our
rush to talk about specific rules, and it is exactly the reason why the
goal of RDA as "... a new standard for resource description and access,
designed for the digital world" is looking increasingly out of reach.
The third point about "what changes require a new description" is an
important one, particularly in the digital world. What I'd suggest
though, is that rather than proliferate the notion that there might be
differences in the rules for what constitutes sufficient difference based
on mode of issuance or genre, that we try to come up with something more
general that can be applied across the board.
For those of you who don't believe that the powers-that-be
in our libraries will over-rule any rearrangement of
the deck chairs on the Titanic, I'd suggest you take a
look at the new report from the University of California:
http://libraries.universityofcalifornia.edu/sopag/BSTF/Final.pdf
This report is being discussed at some level in most large academic
libraries in this country, and the implications for an RDA not based on
a significantly broader view of the world are quite chilling.