Quoting "Myers, John F." <[log in to unmask]>:
>
> But, to launch a larger dialogue, I am not unconvinced that
> conducting a zero-sum analysis of the perceived relative worth of
> individual data elements in a data format is a bit of a straw man.
> We can lob personal and anecdotal assessments around all we want,
> but the data element needs are going to come from the respective
> descriptive rules -- a conversation for another list or at least
> another day. Any prospective upgrade or replacement to MARC needs
> to be sufficiently robust and flexible to accommodate the
> descriptive requirements for any number of communities.
I agree in part, John, but there's an issue here: the descriptive
rules exist to produce the data for the catalog. They aren't an
abstraction. Behind every element in the rules is an assumption about
technology and usage. Headings (now "access points") assume a linear
order:
Smith, John
Smith, John, 1836-1889
Smith, John, 1927-
For systems that do not provide a linear order, a different form of
heading might be better.
Transcribed elements from the title page have a an assumed surrogate
role that translates into a user display. We shouldn't try to get
multiple functions out of that data, but that means that some other
data elements might be needed (like a coded version of the place of
publication that disambiguates places.)
For each element, we really need to look at what its purpose is in the
catalog. Here's an example of something that might be a surprise to
some people:
245 with filing indicators: This is really two different data elements
- a transcription of the title from the title page, and a title, sans
initial articles, that is needed for sorting. If you look at this from
a systems point of view (and we all should, because this data WILL be
processed by computer systems) it might be better to break that into
two data elements. The filing title might have other differences as
well, and therefore has some overlap with the 246. I assume it was
kept as a single field originally because the needs were different
when MARC was developed, since sorting was the only processing that
took place on titles. (And we were closer to the card catalog in our
assumptions.)
Cataloging has to support the functions of the catalog. The cataloging
rules that we have had in the past had as their goal the creation of a
card. There was no system that came between the cataloger and the
final catalog entry. Even RDA does not acknowledge that there are
system needs, and no systems experts were included in the JSC work.
The creation of cataloging rules today needs to be an iterative
process between describing the thing and providing user services over
systems and networks.
I especially need to note that RDA (and FRBR, on which it is based) is
silent on subject access, and our methods of subject access are now
over a century old and yet still get little of our attention.
This all means that the cataloging community needs to be actively
participating in system design and how systems will make use of the
catalog data. It is not helpful if cataloging produces data that can't
be used in systems. There is a huge difference between what a person
might have in mind for some data and what you can do with it with
programs, systems, networks. The latter are not magic; they are
methods with constraints and possibilities that must be understood.
More below, believe it or not...
>
> Regardless of the mechanism for the resulting communication format,
> I think there is a need to address the larger issues of:
> * transcribed vs. controlled data,
What do you see as the issue?
> * controlled data rendered in representational (i.e. text) and
> non-representational (e.g. URIs) forms (because regardless of the
> disadvantages of the former, there will likely be instances where
> that is the only option available),
I think we need to note those elements that we will not attempt to
further identify, such as transcribed data.
> * how to connect those respective representations of the same data element,
> * how to connect the individual data elements applicable to a given resource,
> * how to connect those out to other resources and entities,
As you probably know, linked data does all of these very well.
> * extensibility, as much as possible, with respect to future data
> needs (witness the cited regret of Avram in developing different
> bibliographic formats for specific media, and the implications of
> FRBR that alters the bifurcation between authority/bibliographic
> data and replaces it with data for entity groups).
> (And all while keeping the data relatively compact!)
Extensibility is more of a social problem than a technical one. There
are extensibility mechanisms that work quite well but the idea of
allowing portions of the community to have their own extensions gets
great push-back. Again, I think linked data has an answer here but we
need for more people to understand how that works and how the
extension of data can work in an non-disruptive way. (I cover this in
my talks on linked data, but don't have it written up. I should do a
blog post..)
>
> Somewhat farther afield, how to realize all of that into an interface that:
> * translates computer friendly, language-neutral coding (element
> labels and data) into something intelligible for those performing
> the data entry in the context of multiple languages and descriptive
> codes,
> * won't require full double entry of transcribed and controlled
> data, and ultimately
Again, I think all of these can be done with today's technology. A
good start would be some mocking up of what people would like the
input screen to look like. Perhaps of a wish list of features.
> * renders a coherent and consistent display to end-users of the data.
.... or, allows users to have their own views, or to have community
views (I always think of music librarianship as having some specific
needs, but there may be others). There doesn't have to be just one way
to do things, although there should be a default that makes system
development and all of that detailed decision-making as simple as
possible for folks who do not want to go through the effort of
customizing.
etc.
kc
>
>
>
> John F. Myers, Catalog Librarian
> Schaffer Library, Union College
> 807 Union St.
> Schenectady NY 12308
>
> 518-388-6623
> [log in to unmask]
>
>
>
>
--
Karen Coyle
[log in to unmask] http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet
|