On Wed, 30 Jun 1999, Renato Iannella wrote:
> Dear all, we would welcome feedback and comments on
> an update to the A-Core Metadata set (was ADMIN Core):
>
> http://metadata.net/ac/draft-iannella-admin-01.txt
>
> A-Core is used in the process of metadata management
> to associate the instruments (who, what) with the events
> (when) to provide simple verification of the integrity,
> ownership, and authorship of content metadata retrieved
> from networked resources.
Four high-level comments on the draft...
1) The model. You appear to base this work on the assertion that
there are resources (data)
there is 'content metadata' that describes those resources
there is 'admin metadata' that describes that 'content metadata'
Why not deal with the general case...
there are resources
there is 'content metadata' that describes those resources
both of these are data that can be described using 'admin metadata'
I.e., your linear
resource <-describes-- Content metadata <-describes-- A-Core metadata
model becomes triangular
A-Core
/ \
/ \
/ \
describes describes
/ \
/ \
/ \
v v
resource <-describes-- content metadata
We need to be able to make statements about the 'provenance, management or
administration' of both resources and metadata and we need to be able to
make such statements separately. But I see no reason why we shouldn't use
the same mechanism to make such statements?
2) Semantics. Wouldn't it be better to re-use existing DC semantics where
possible? E.g. I suspect all the 'dates' you propose are covered by the
semantics of dc:date ?
3) HTML Syntax. In your proposed HTML syntax, some embedded metadata
(e.g. DC) describes the resource in which it is embedded, some embedded
metadata (e.g. A-core) describes some (or perhaps all?) of the other
metadata embedded in the resource in which it is embedded. This feels odd
(to me). I can see why you've done it - because HTML META tags give you
little option! But it still feels odd!
4) Syntax - general. You have adopted an 'event' based model. The
consequence of this is that most of your embedded AC META tags do not mean
anything when taken in isolation. You can take any DC META tag in
isolation and it tells you something about the relationship between the
value (content="value") and the resource in which it is embedded. On the
other hand, an AC META tag, e.g.
<meta name="AC.name" content="Oldman, Sam">
only means something when read in association with an AC.activity META
tag.
In the RDF/XML syntax you could have opted to structurally associate the
AC:activity with the AC:name. Instead, your encoding 'does not provide
any element grouping mechanism'. I don't understand why? Draw the
arc-node diagram for your example RDF/XML - isn't it pretty much
meaningless? Aren't you are asserting things like
Description '#id0001' has a property 'AC:name' with a value 'Crystal,
Jacky'.
The general thrust of this is
- we can and should describe DC metadata using DC metadata where
possible
- we should only invent admin-related elements where the
semantics fall outside of the DC element definitions.
- if A-core is needed, then it should be capable of describing both data
and metadata (because one person's metadata is another person's data)
- we have a problem with this in HTML, because of the limitations of the
HTML META tag - but we have a solution in the form of RDF.
Andy.
--
UK Office for Library and Information Networking
University of Bath, Bath, BA2 7AY, UK Voice: +44 1225 323933
http://www.ukoln.ac.uk/ukoln/staff/a.powell/ Fax: +44 1225 826838
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|