Print

Print


On Tue, 2004-10-26 at 08:08, Ray Denenberg, Library of Congress wrote:

>
>
>  >And I'm still confused
> > about how "organizations register objects" -- is there a registry for
> > the individual objects?
>
> Do you mean "objects" or "object types"?

It's whatever level is in the info URI. It seems that if the URI will go
all of the way down to the value level
  info:xv/1/mods/resourceType/cartographic
then you need to be able to get to that value. In other words, this URI
does not just identify the authoritative list, it identifies the *value*
in the authoritative list.

> Right.  (Except I would say "entirely up to the xv community to define,"  LC
> acting on behalf of that community.)

Does that mean that we're going to have another committee ;-)? In other
words, how will this xv community define itself and interact?

> We have an additional  proposal, not quite ready to post yet, that we think
> is complementary to the xv proposal.  (In fact, the xv proposal could be
> integrated into the new proposal.)  That is, we think this additional
> namespace, tentatively 'cv' for "controlled value", along with 'xv', neatly
> partition the problem space of controlled values.  We don't suggest that the
> xv proposal be approved (or submitted `for 'info' approval) until we can get
> this additional proposal out for discussion, which we hope will be soon,
> however we do feel that the xv proposal is a separate problem and can be
> discussed separately for now.

I should mention here at NISO is in the early stages of convening a
meeting on identifiers for the library and vendor community. It sounds
like this approach is one that should be considered. The first steps
will be to define the problem space, and it seems clear that
authoritative lists will be in that space.

> > So you are saying that the registry agency might prefer to have fewer
> namespaces registered.
> Yes, is that not a reasonable assumption?

Not necessarily. I think there's a trade-off between simple identifiers,
and many of them, vs. fewer but more complex identifiers. The decision
shouldn't be based on what's convenient for the registry agency but on
what works best for the users. The registry should serve us, not us it.

>
> Ok then, what's best for our community?

Well, that's not for you and I to decide, obviously. The question is:
how do you find out the answer to that? What is the mechanism for
convening the key players and working out a solution? And the hardest
part is: how do you define *our* community? Libraries? Library vendors?
MODS users?

> Both. OCLC has an authoritative list (and, presumably, rules for how objects
> are added, which may, but need not, include rules for delegation of
> subauthorities), which  to me fulfills the definition of "registered".

Well, then we have two definitions of "registered" -- one, that
something is "registered" in the info-uri registry, and one that there
is a list elsewhere (which one has to take on faith). I would prefer
that "registered" mean something definite -- that there be a registry
that I know I can go to and that I'll find the list or values that I
need. One of the advantages of standardizing our identifiers should be
that we also standardize our access to the many lists that we use in our
work. So if I see a standard identifier I should be able to go someplace
well-known to find out
  - who "owns" that name space
  - what are the valid values
So I guess to me it isn't "registered" unless we have an actual
registry.

> Each authority provides a list of subauthorites that it has assigned, and
> for each, a url.

And I'd want those urls in a standard location -- preferably in the
registry, or with a link from the registry to the standard location. In
other words, it has to be possible to go directly from the registry to
the lists, not hunt and peck around individual vendors' web sites. So
the layer of sub-authorities means that you'll need a similar layer of
pointers to the "real thing."
--
-------------------------------------
Karen Coyle
Digital Library Specialist
http://www.kcoyle.net
Ph: 510-540-7596 Fax: 510-848-3913
--------------------------------------