In 1996, I wrote a paper on this topic:
http://www.library.ucsb.edu/people/larsgaard/mulinher.html
from the standard-cataloging point of view.
It is important to differentiate between the various
types of relationships that are possible, and in
the paper I do try to do that.
Mary Larsgaard
UCSB
>Date: Wed, 11 Nov 1998 12:17:09 +1100
>From: Alex Satrapa <[log in to unmask]>
>Organization: tSA Consulting Pty. Ltd
>X-Mailer: Mozilla 4.07 [en] (Win98; I)
>To: Karen Coyle <[log in to unmask]>
>Cc: meta2 <[log in to unmask]>
>Subject: Re: More 1:1
>Sender: [log in to unmask]
>
>To me, the whole reason we're allocating metadata to an object is that, by
>expending 15 minutes of effort now, we're saving the rest of the world
half an
>hour each later.
>
>To paraphrase - when allocating metadata to a "thing", the metadata should be
>allocated in a way that "makes sense" from the point of view of Jo Average,
>who's looking for some information, but dosn't know where to find it.
>
>So a hierarchy of preserved copies of documents (pardon my niavete, but
I'm only
>guessing here) only relates to one parent document - the branching in the
>hierarchies might be due to particular types of replication (eg: was it
>photocopied or OCR-ed), or particular replicators (eg: John Smith writes
>Sanskrit differently to Phillipa Wright).
>
>Now, ask yourself the question - what is it that someone might be looking for
>that you might have the answer to? Perhaps someone is looking for a document
>with a particular title, which you are maintaining copies of. That would be
>covered by one simplistic DC metadata set, for the hierarchy itself.
>Perhaps the person is looking for information about people who can write (and
>preferably read) Sanskrit. You'd have at least two DC sets describing
versions
>of the document written by John and Phillipa. Perhaps you have surrogate
records
>for John and Phillipa, or at least have pointers to, say Vcards or a four11
>database (these pointers would be in the DC.Creator field of the DC set
for the
>trascriptions that John and Phillipa have been writing).
>
>So you see... the 1:1 issue makes it a lot easier for people to use your
>information for their really obtuse queries.
>
>A CD containing pre-recorded tracks would itself count as a collection. It's
>DC.Type might be "sound", with a list of "DC.Source" fields pointing to
(DC sets
>for?) the individual tracks on the CD. These in turn point to (surrogate) DC
>meta data sets for the actual performances.
>
>Trying to focus on the creation of DC meta data from the point of view of the
>creator is probably "wrong" or "bad" (at the very least, it's damned
>egocentric). That's not what meta data is for. Meta data is *supposed* to be
>about resource discovery. Well... that's my opinion anyway.
>
>So as you start allocating metadata to your
>documents/collections/objects/entities, make sure that the metadata you're
>allocating makes sense from a resource discovery perspective.
>
>Another point that raises its head (in my point of view anyway) is at what
point
>you stop allocating metadata. For example, it is obvious that it's your
>responsibility to catalog a CD that you have in your own collection.
However, if
>the CD publisher was doing their job, the only meta data you need to store is
>DC.Title (if that), DC.Description (if that), and DC.Identifier (probably the
>publisher's catalogue number - do CDs have an ISBN or URN equivalent?)
>
>In the case of a collection such as the Mark Twain papers... isn't it worth
>"cataloguing" the entire set as individual works? Then the entire set,
"The Mark
>Twain Papers" becomes one collection, with its own record, which then has
>two-way links with the individual documents. Yes, it's a lot of work. But
if you
>do it completely the first time around, it's easier for other to (a) find the
>resource, and (b) refer to the resource in their own works.
>
>Regards,
>Alex Satrapa
>
>Karen Coyle wrote:
>
>> At 08:29 PM 11/5/98 -0500, J. Trant wrote:
>> >
>> >The way I see it, 1:1 is just a case of describing the 'object in hand'.
>> >And since there isn't a way of linking repeating elements in DC, it is
also
>> >the only way to ensure that records make logical sense.
>> >
>>
>> It's kind of hard to talk about the "object in hand" in the digital
>> environment; ...
>
>>
>> For example,
>
>[I paraphrase as]
> - Audio CD, tracks by different composers performed by one or more sets of
>artists
> - Collection of works of a particular author
> - Set of preservation copies (I assume the purpose here is to keep the
>intellectual content alive even
> though the paper/plastic/metal/etc used to record the content is
degerating)
>
>> So the first thing that we have to concede is that the "1" that represents
>> the work in the "1:1" is highly variable, and that different communities
>> have different definitions for the unit with which they work. We can't
>> force them into an artificial atomization of their world; what they do
>> makes sense within their context.
>
>[note about some communities only working with heirarchies, rather than
>entities]
>
|