[Dini-ag-kim-bestandsdaten] Re: Holdings, Units, Items and URIs

[Jakob Voß]

Hi,

Thanks to Carsten for the good summary and collection. Thanks also to 
Jörg for showing the complexity of difference between "logical" and 
"physical" bibliographic objects.

In short, there is a n-to-m relationship between logical documents 
(which can be books, articles, series etc.) and physical items (which 
can be bound books, journal issues, PDF files etc.). For the holding 
ontology in RDF I propose to use an "examplarOf" relationship between
documents and items, as also shown at

https://wiki.dnb.de/display/DINIAGKIM/Scope+of+Holdings

In addition one should support "narrowerExemplar" and "broaderExemplar" 
relationships, for instance to connect a chapter as document with a 
physical book or with a PDF file of the book that contains multiple 
chapters.

Furthermore there is the "holding record", as found in current library 
data. I hope I don't disillusion anyone, but there is no general and 
clear relationship between holding records and neither documents, nor 
items. In our catalogs we can find holding records that refer to 
multiple documents, holding records that refer to multiple items, and
holding records that refer to one document with multiple items. Trying 
to get a clean mapping between holding records and document or item 
entities with URIs will require to bend the concept of a document and 
item as following:

If we want all holding records to refer to exactely one item, there will 
be items that consist of multiple physical parts (as noted in Carstens 
mail).

If we want all holding records to refer to exactely one document, there
will be documents that consist span multiple objects (for instance a 
series is build of multiple parts).

I think this bending is not a general problem, but it must be clear that
it can be against the normal concepts of documents and items when 
talking to other people.

By the way, I like Jörgs definition of an item, but I would remove the 
limitation to "the smallest thing" because Items can be connected to 
each other by part-of relationships.

> In general, an item is the smallest thing the librarian can hand over
> to a patron for reference, or for circulation. An item gets a single
> shelfmark, and a single barcode (or a number, for inventory).

Furthermore I would note that the single shelfmark, barcode, copy number 
etc. can be located in many different MARC fields or other locations. 
Carsten wrote:

 > In all of the formats and standards like MARC, Z39.71, ZETA etc.
 > there is always the possibility to subsume multiple Units/Items
 > within one Holding.
 >
 > For instance the MARC Holding field 852 (Location) is repeatable.
 > And in our union catalogue it is common that multiple Units with
 > different locations, callnumbers and access restrictions are
 > described in one holding record.

Yes, there is no 1-to-1 relationship between holding records and items - 
but this is no problem if items have other identifiers but 
holding-record-ids. Frankly, I doubt that modeling holding records is of 
much use at all. Library holding records are old artifacts that one 
should get rid of when mapping holding data to RDF. Nobody is interested 
in the way that libraries tend to combine single or multiple documents 
and items in records.

What really matters are documents and items. Or to put in the other way 
round: A holding record can be *anything* that describes or references 
one or multiple items, held by the same agent. For instance I can create 
an inventory of the books on my small shelf in the office as Excel list, 
so this list becomes a holding record.

Jakob

-- 
Jakob Voß <jakob.voss at gbv.de>, skype: nichtich
Verbundzentrale des GBV (VZG) / Common Library Network
Platz der Goettinger Sieben 1, 37073 Göttingen, Germany
+49 (0)551 39-10242, http://www.gbv.de