Unlike lucky Kass, I wasn't able to get to Helsinki, but what I have
heard so far of the 1:1 axiom concerns me, too. There is a tendency to
take abstract models to an illogical extreme, only to see them
disintegrate in actual practice. The library community has been dealing
with the description of and access to reformatted materials for decades,
and I hope we've learned something during that time. For example,
we know that for the purposes of resource *discovery*, no one cares who
ran the microfilm camera or when, and we have adjusted our practices to
reflect this reality. Let's use what we've learned as we build a model
to provide access to digitally-reformatted materials.
The 1:1 model, if I understand it correctly, posits complete,
independent metadata records for each instantiation with
relation links between them. This does not seem to allow for
inheritance of duplicative data (e.g. CREATOR and TITLE, in the case of
a reformatted work) nor for primacy of the characteristics pertaining to
the intellectual content rather than those of its erstwhile incarnation.
In contrast to Kass, the Source element as it stands concerns me for the
same reason: it buries the most important characteristics of a
reformatted work in an un-subfielded element, segregating the author,
publisher, and date users care about from those they will likely to be
able to search. E.g., if the user searches for works by an author in a
fielded search, he probably would not retrieve relevant (and only
relevant) info also from the source element. However, if we subfield the
Source element AND search engines treat CREATOR and SOURCE.Creator the
same way, we come closer to a workable solution. I still prefer a
grouping mechanism, myself, as being cleaner and more "ignorable" by
those who don't want/need to be aware of it.
--Robin
Robin Wendler ........................ work (617) 495-3724
Office for Information Systems ....... fax (617) 495-0491
Harvard University Library ........... [log in to unmask]
Cambridge, MA, USA 02138 .............
On Fri, 24 Oct 1997, Kass Evans wrote:
>
>
> On Thu, 23 Oct 1997, Ricky Erway wrote:
>
> > REPLY TO 10/23/97 12:43 FROM [log in to unmask] "Kass Evans": 1 to 1 Relation &
> > multiple metadata sets
> >
> > Kass,
> >
> > I think that if the cataloger and the likely users are interested in
> > the digital surrogate, that is what should be described. I
> > believe that the 1:1/relation group represents people who are
> > interested in original things (and probably have them "in-hand" in
> > their collections). The fact that there is a digital surrogate is
> > merely a reference convenience, not what is being sought by the users
> > of the metadata (e.g., the person searching Altavista for examples of
> > impressionistic painting will search for Type=painting not
> > Type=digital image). A person who has a digital image, but does not
> > "hold" the original, probably won't think of his image as a
> > surrogate.
> >
> > I think there should be no onus to describe any other
> > versions/instantiations/reproductions than those deemed important to
> > endusers trying to find the thing described. That is still the
> > point.
> >
> > Ricky
> >
> > Ricky Erway, RLG
>
> Ricky,
>
> I realize that I may not have been clear. Following Ralph Swick's
> presentation on the last day on Relation and RDF, I spoke with Ralph and
> Eric Miller about the implications. If I understood them correctly (many
> apologies if I did not) then what they were advocating was:
>
> A metadata set for a digital copy of a photograph may *only* discuss the
> digitized photograph. If you want to provide info on the actual
> photograph then you must make a second metadata set for the photograph and
> provide some type of pointer to it (a Mechanical Relation). And if you
> want to provide info for the original painting that was photographed then
> you will need to make a 3rd metadata set for the painting and provide some
> type of pointer to that (another Mechanical Relation). This new
> interpretation of Relation, Mechanical would effectively supplant the
> current use of Source.
>
> I have no concerns about each version having its own metadata set. My
> concern is that the person creating metadata for the digitized photo not
> be prescribed to create the other 2 metadata sets in order to convey that
> information when it fits so nicely into Source.
>
> I left DC5 with the impression that a number of the folks working on
> Relation were moving in this direction and wanted to register an early
> vote against this level of complexity. If I have totally misunderstood
> the point that Ralph was making with his diagrams then I will happily
> stand corrected.
>
> Kass Evans
> Florida International University
> Digital Library Project
> [log in to unmask]
>
>
>
>
|