I would agree with these concerns completely, and it's one of these
"well I wouldn't start from here" type situations.

I haven't really been involved in the RDA discussions to date, but I get
the impression that this work is too far along for us to feel optimistic
about changing some of the fundamental decisions here. Which means (for
me) making the best of the situation as is, and look at ways we can see
to move incrementally forward and demonstrate the strength of a
different approach. 

So, I suppose the possibility I see in these vocab lists is that if we
create it for the documentation, we also create the possibility of using
it in a cataloguing environment. So I can imagine a record like:

300 ## $a 324 <a
href="http://RDVocab.info/termList/RDAextent/1001">p.</a> $b <a
href="http://RDVocab.info/termList/RDAillustrativeContent/1001">illus.</
a> (incl. 24 <a
href="http://RDVocab.info/termList/RDAillustrativeContent/1002">ports.</
a>, 30 <a
href="http://RDVocab.info/termList/RDAillustrativeContent/1003">maps</a>
, 5 <a
href="http://RDVocab.info/termList/RDAillustrativeContent/1005">plans</a
>, 9 <a
href="http://RDVocab.info/termList/RDAillustrativeContent/1006">facsims.
</a>) $c 28 x 33 cm.

Which would provide a reasonably good machine-parsable record, language
independent etc.

I wouldn't ideally structure like this, but I can see this working - and
I think it would be reasonably easy to build a cataloguing client that
could exploit the vocabs in this way once they are in place

Owen

Owen Stephens
Assistant Director: e-Strategy and Information Resources
Imperial College London Library 
Imperial College London 
South Kensington 
London SW7 2AZ 

 
Tel: 020 7594 8829
Email: [log in to unmask]

-----Original Message-----
From: List for discussion on Resource Description and Access (RDA)
[mailto:[log in to unmask]] On Behalf Of Thomas Dukleth
Sent: 18 June 2008 18:35
To: [log in to unmask]
Subject: Re: New RDA Vocabularies available (plus other info)

Reply inline:

On Wed, June 18, 2008 3:41 pm, Karen Coyle wrote:
> I think the main areas where use of URIs is being contemplated is for 
> 1) headings that are defined in authority files (e.g. name and subject
> authorities) and 2) links to related records (such as the ISSNs in 
> series headings). Can anyone confirm or deny this?

I had presumed that URIs in RDA had been significantly limited to what
has been the domain of authorised string forms, especially in standard
MARC 21.  (UNIMARC bibliographic has $3 for the authority record number
as part of the standard which has been an advantage over mere authorised
string forms in standard MARC 21.)  I presumed that Diane Hillmann had
misunderstood the extent to which I was advocating applying
linguistically neutral identifiers and hoped that greater use of such
identifiers was being contemplated for RDA.

I thought I understood from previous discussion in this or another
thread that the possible use of URIs was being contemplated on this list
for standard vocabulary which is applied to the type of material, format
of the carrier, physical description, etc.  A major question was whether
a subject vocabulary in SKOS or OWL could be used outside the context of
topical content to cover also terms used for physical attributes of the
material, etc.  Should I understand that JSC has no realistic
possibility of endorsing any such expansion of the use of URIs for
attributes of the material beyond the few which Karen Coyle listed from
the message quoted above?

A standard vocabulary bound to a particular natural language is
problematic for indexing across multiple languages where cataloguing
rules are used in translation and the standard vocabulary is also
translated.

Writing scripts to parse records for machine use can be great fun. 
However, the less normalisation is needed to make machine use of every
element in the records the more effort can be applied to doing something
interesting with the records in already normalised machine actionable
form throughout the record.

There is still a need for a few transcribed elements to use for record
matching when multiple manifestations exist sharing too many common
elements.  Outside that minimum, natural language should not be used
where identifiers can be applied more universally whether for authority
forms or descriptive vocabulary.

Presenting machine actionable identifiers in a format for human readable
consumption is easy throughout the record.  What we have to do currently
to normalise record elements for machine use is a waste of cataloguer
entry time creating records inefficiently, and a waste of programmer and
machine resources parsing the string variations which are catalogued to
partly reverse the effort originally expended when record elements were
originally catalogued.

If we have an ever worsening crisis in cataloguing, why do we exacerbate
the problem by wasting such precious resources?


Thomas Dukleth
Agogme
109 E 9th Street, 3D
New York, NY  10003
USA
http://www.agogme.com
212-674-3783

[...]