Bernard,
One thing that I would like to see clarified in the LC work is what it
means to "replace MARC." I'm assuming that the goal is not to create a
set of triples that replicate the fields, subfields, indicators and
fixed fields of MARC, but instead to make sure that all of the
*information* (I guess I could say "useful information") in MARC can
move forward to this new data format. In my opinion ([1], [2]) this
means more than just creating a one-to-one of MARC in RDF; it means
converting the *intention* of the data to a new metadata model. So where
in MARC we might have had text strings representing a related work (and
perhaps an record number in $w), in a new model that might be a triple
that represents the specific relationship between the focus resource
(the thing being described or cataloged) and the resource it relates to.
As I've said before, RDF is not just another serialization of our data,
like coding MARC in MARCXML - it really is a new way of thinking about
our data, and it has possibilities we haven't had before. I see no
possibility of enhancing MARC to achieve this. The example of the $0 is
essentially proof of how that would *not* work, and you can find the
arguments in the MARBI discussions of that subfield. (The gist is that
there are many MARC fields that are not one-to-one with an entity, and
MARC does not have a way to associate a subfield with particular other
subfields.)
The big question here is whether LC and its partners will decide that
this new framework must not only accept the information once carried in
MARC but must be able to round-trip with MARC records. That, I think,
would be a hindrance to moving forward.
and, p.s. Everything that Ed said. And here's the SW work that is
designed to support the vision he provides:
"Named graphs" http://www.w3.org/2004/03/trix/
and p.p.s. Agree with Simon on inappropriate tone of original post.
kc
[1] Coyle, "MARC21 as data; a start"
http://journal.code4lib.org/articles/5468
[2] http://futurelib.pbworks.com/w/page/29114548/MARC%20elements
On 6/4/12 12:29 AM, Bernhard Eversberg wrote:
> 04.06.2012 04:21, Simon Spero:
>>
>> Eric Miller <http://zepheira.com/about/people/eric-miller/> "[...] is
>> co-founder and president of Zepheira [...] Previously, Eric was a [...]
>> Research Scientist at OCLC in Dublin, Ohio". I believe that 8 years at
>> OCLC is a sufficient period of time to allow one to gain some
>> familiarity with libraries.
>
> The LC press release does indeed say:
> A major goal of BFTI is to replace the MARC 21 exchange format with a
> new Linked Data model ...
>
> I just wonder if this is even conceptually possible. More precisely:
> Can a MARC record be replaced by a set of Linked Data triples? How could
> the data exchange then actually work? Exchange, up until now, is based
> on records. So what would replace them - or is the new idea and the
> intention now to make the very concept of the bibliographic record
> obsolete? What does that mean for ILSs that used to receive MARC
> records? What will it mean for the processes of cataloging and the
> catalogers' user interface?
> OK, there may be features of MARC that need not be dealt with since
> they are obsolete or useless in other ways. Yet, would leaving these out
> make triplification any easier?
>
> But the LC statement, I think, may also be read as calling for an
> addition or enhancement for MARC, not its wholesale replacement.
> For the press release statement goes on to say
>
> ... while retaining as much as possible the robust and beneficial
> aspects of the historical format.
> See
> http://www.loc.gov/marc/transition/news/modeling-052212.html
>
> And this will indeed - if I'm not entirely misinformed - not be
> possible with the Linked Data concept and triplestores alone.
>
> What one might envision is something like this:
>
> 1. In regular MARC records, add something like the subfield $0 or $w
> to those fields that need to be linked to authority records or
> other bib records, each containing a URI XYZ.
>
> 2. Set up a central triplestore for the authority data allowing to be
> queried for XYZ to yield the information necessary to enhance the
> MARC record at that particular point, say, with information about
> the person or subject heading.
>
> 3. OPACs may then use that remote triplestore to enhance their
> record displays without the need to restructure their MARC records
> into an entirely new way, except adding those URIs or constructing
> them from existing data on demand.
>
> B.Eversberg
--
Karen Coyle
[log in to unmask] http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet
|