I am trying to answer the question in my own mind as to whether the notion
of records should be in the abstract model or not.
In theory, the 50 million records in the OCLC union catalog could be
represented completely in a single RDF-XML file. As a practical matter,
that would be absurd. Thus, records remain quite important at the
application layer, and for exchange of information across collections.
We conventionally partition our metadata stores with a particular view...
generally based on that-which-we-think-users-are-most-interested-in, and I
think this will not change, even though we want to be able to look at our
collections in many views.
show me all the books whose authors include dostoyevsky
show me all the books on the subject of japanese gardening
show me names of people with a postal exchange of xxx zzz
Roughly speaking, a metadata store is a collection of assertions that relate
entities, properties and values. We partition those assertions in
particular ways for database performance reasons, for functional reasons,
for administrative reasons, for economic reasons.
It seems rather important to be able to aggregate a set of assertions in a
particular way, even though we may want to get query results that span many
Where is the abstract representation of this aggregation, or is it needed at
From: Mikael Nilsson [mailto:[log in to unmask]]
Sent: Tuesday, September 09, 2003 7:06 AM
To: [log in to unmask]
Subject: Re: [DC-ARCHITECTURE] Dublin Core Abstract Model
On Mon, 2003-09-08 at 18:16, Andy Powell wrote:
> If a Qualified DC metadata record can include external properties then the
> logical conclusion is that a metadata record that contains 50 IEEE LOM
> propertis and 1 DC property is a Qualified DC metadata record. That
> feels non-sensical (or, at least, non-intuitive) to me.
> A metadata record made up of 50 LOM properties and 1 DC property is a LOM
> record that 'incorporates' qualified DC (IMHO).
I really think your examples demonstrate quite clearly that the concept
of "record" does not belong to the DC abstract model at all. It only
makes sense in certain encodings (XML, notably). In RDF there is no
canonical way of identifying the borders between records anyway, so why
make this distinction at the abstract level?
I strongly feel the concept of a metadata record or instance creates
more problems than it solves (I even wrote about it:
> That said, I think there's enough in the current document to allow people
> like IEEE LOM to adopt the same model if they desire and write it up as an
> IEEE LOM abstract model.
Indeed, that would be great... The problem is of course that LOM has a
completely incompatible metamodel :-)
Plus ça change, plus c'est la même chose