The idea of such a registry has been discussed in a number of contexts.
My colleague Diane Vizine-Goetz provides the following note:
--->>
LC maintains a 'registry' of a sort for source names for terms and
classification codes used in MARC records. See: MARC code lists for term
sources (6xx) <http://www.loc.gov/marc/relators/relasour.html> and MARC
code lists for classification sources
<http://www.loc.gov/marc/relators/relaclas.html>
We adopted these codes to identify knowledge organization schemes in our
info:kos application
<http://www.oclc.org/research/projects/termservices/resources/info-uri.h
tm>. These lists could provide a good starting point for a registry of
source names.
<<---
JISC has had a slot in its thinking for 'terminology services' for some
time. I think that some of the issues raised in this discussion might
find a home in the broader context of a managed set of terminology
services if it is ever advanced.
Lorcan
Lorcan Dempsey [http://orweblog.oclc.org]
OCLC Research [http://www.oclc.org/research/]
-----Original Message-----
From: The CETIS Metadata Special Interest Group
[mailto:[log in to unmask]] On Behalf Of Andy Powell
Sent: Wednesday, March 09, 2005 6:18 PM
To: [log in to unmask]
Subject: Re: Questions re 9. Classification: First one
On Wed, 9 Mar 2005, Charles Duncan wrote:
> This is a good point because it may be important in future. Within the
> intraLibrary repository, administrators are able to use any string for
> the source of a taxonomy. However, when an object is exported from one
> repository and imported into another, the Source is used to try to
> identify if the same classification scheme is in use in the new
repository.
>
> We really need a taxonomy registry in which recognised source codes
> can be registered and variants of standard taxonomies can be
identified.
There are some difficult issues with running a registry like this, not
least scalability, trademark handling and conflict resolution. Is a
centralised registry the right solution for what is essentially a
globally distributed problem? What do you do when someone registers a
Source called 'Coca-Cola'? Who gets 'LCSH' - the Library of Congress
Subject Headings' or the 'Longwinded Classification for the Scottish
Highlands'?
Etc., etc.
DCMI spent a long while praparing the groundwork for building a
centralised registry of DCMI 'encoding schemes' but eventually gave up
when it became apparent that non-one wanted to take operational or legal
responsibility for its maintenance. More importantly, we realsied that
because DCMI uses URIs to name its encoding schemes (DCMI uses URIs to
name all its metadata terms) the Web effectively becomes the registry.
If I want to name a new encoding scheme then I simply create a new
globally unique URI for it (in the bit of URI space that I own) and
start using it! That leaves ICAAN with the problem of determining who
owns which bits of URI-space.
As others have pointed out, the failure of LOM to use URIs to name
stuff, using simple strings instead, is one of its weaknesses. And as
I've argued elsewhere, this is particularly the case with the 1.1.1
catalog and 1.1.2 entry elements, since you can't magically make the
string "10.1003/1234.5678" into a globally unique DOI by simply saying
that the catalog is "DOI" - since "DOI" is just a string that has no
global uniqueness. You can *only* make the DOI globally unique by
encoding it in a globally unique form like a URI (such as
"doi:10.1003/1234.5678"), registering the URI scheme with IANA and
having a syntax that guarantees that the string must be interpretted as
a URI.
However, I digress... :-)
Given that LOM uses strings rather than URIs to name the Source, I tend
to agree that despite the issues raised above, a registry of Source
names would probably be quite a useful pragmatic approach, particularly
within the context of, say, UK deployment of LOM.
Andy.
> Charles
> --------------------------------------------------------------
> Intrallect Ltd [log in to unmask]
> Braehead Business Park Tel: +44 (0)870 234 3933
> Braehead Road Fax: +44 (0)1506 50 5117
> Linlithgow, EH49 6EP, Scotland http://www.intrallect.com/
>
> Sarah Currier wrote:
> > Hi all,
> >
> > Yes, it's me again... more LOM related questions. I have two, which
> > I will do in separate emails.
> >
> > Firstly, in the LOM element 9.2.1 TaxonPath:Source we are supposed
> > to say what classification scheme, taxonomy, etc. is being used in
> > that instance. As we will have in our repository two or three
> > taxonomies which have not been previously used, I wondered if there
> > is a common or standard way of creating the code entered here (I
> > know MARC has a list of such codes maintained by the Library of
> > Congress). The taxonomies we will be using are not so much local, as
> > ones which haven't been used in LOM metadata yet, but may be in
future.
> >
> > What have other folk done and how important is this anyway?
> >
> > I note that in the CanCore guidelines they give an example where
> > they declare the National Library of Canada's version of Dewey as:
> > "DDC NLC-BNC http://www.nlc-bnc.ca/caninfo/" ... so they don't use
> > the MARC practice of small letters rather than capitals, and they
> > include a URL (which incidentally doesn't lead straight to the
> > taxonomy, it leads you to your first step (of several) in finding
> > their subject browse page where the taxonomy is used). I do think
> > that including a URL to the source of the taxonomy is a good idea
> > though- I'm not sure where else that information would go.
> >
> > Feedback, discussion, etc. welcome.
> >
> > Second question coming shortly..
> > Thanks
> > Sarah
> >
>
>
> --
>
Andy
--
Distributed Systems, UKOLN, University of Bath, Bath, BA2 7AY, UK
http://www.ukoln.ac.uk/ukoln/staff/a.powell/ +44 1225 383933
Resource Discovery Network http://www.rdn.ac.uk/
|