It is like DCMI five years ago (Re: Documents, Cars, Hills, and Valleys)

This thread sounds like a DCMI work shop five years ago... If you're not
careful, you'll reinvent the metadata concept.

These problems discussed in this thread aren't technical. They belong to
what my non-technical collegues call 'Cataloging rules'.  People doing
caloging rules have head problems of these kinds problems for more a
couple of millenia now, and do cope with them. For instance in one of my
projects are giving shelf-marks from local medieval monastic libraries as
alternative names. That is needed, since one medieval manuscript
may refer to another that way.

It should be obvious that you cannot use anything like a http url for
things things on the web that are also meant to be searchable on their
own. The simple way out is to use a good surrogate url, which could be
widely accepted. The obvious choice would be ldap urls for persons and
corporate bodies.

How to treat places, physical objects and the like, is more tricky. Again
this is not a technical problem, but a problem to strictly adhere to a
cataloging rule. The car industry may provide such a rule, but having done
that they will have to stick that rule for decades. If they change it,
they will have provide info on how to do the transition, or research into
the history of car industry will be unable to track what's been going on.

The problem concerning geography is that the functional requirements are
different for areas of use. Whatever vocabulary one would choose in ones
cataloging rule, one would have to stick to that and then provide also
rules for how to apply updates to the rule.

I tried to persuade Swedish authorities to provide stable URLs for
professions and Swedish universities for URLs for courses and programmes.
My project was to build a search engine that should (among other things)
answer the question:  Which courses should I take to get that job? It
didn't work, in spite of the fact that we didn't use RDF. The problem was
that those who cataloged the courses didn't follow the cataloging rules.

It is very unlikely that the World-wide web will accept anything that even
has a distant resemblance to a cataloging rule. Such rules are typically
specific to "closed worlds" like libraries and different branches of
science or industry. This is because of functional requirements differ.
If a piece of RDF escapes from a electronic engineering preprint archive,
it might damage a product database somewhere else.


Sigge

Received on Saturday, 20 April 2002 03:12:57 UTC