My understanding is that the Agent WG is working on this. Not a lot has
happened (I'm on the list), but this is such a huge thing, that I think it
needs to emerge from a WG with broader consensus than our little group. So
perhaps the Usage Group could request that the Agent group set some
deadlines?

Rebecca


On Mon, 11 Jun 2001, Thomas Baker wrote:

> Dear all,
>
> Jose Borbinha, copied to this message, is interested in renewing the
> momentum for a DCMI Agent Core.  According to our current process, any
> such proposal would land before the Usage Board, so I would like to
> examine some assumptions and pose some questions.  Depending on our
> discussion, perhaps we could move this to dc-architecture, where anyone
> could participate.
>
> -- I am assuming that a DCMI Agent Core would be similar to, but
>    orthogonal to, the current Dublin Core.  In other words, it would be
>    a set of core elements -- elements that could be qualified -- only
>    the properties would refer to the class People instead of the more
>    generic Resources.  The adoption of an Agent Core (DCAMES?), then,
>    would have no affect on the existing Dublin Core (DCMES).
>
> -- In my opinion, an Agent Core should be evaluated completely
>    independently of DCMES by asking: "Are these good, well-defined core
>    attributes, and are they useful for cross-domain searching"?
>
> -- In practice, people may want to mix and match from these two sets in
>    particular application environments -- people want to look at
>    Creators not just as names, but as entities (resources) that can
>    have arbitrarily large sets of properties of their own.  Roland's
>    DC-in-RDF draft explains how this can be done without compromising
>    the dumb-down rule.  However, I do not think the Usage Board would
>    need to rule specifically on the relationship between (say)
>    dc:creator and an Agent Core.  For that, we would need a construct
>    like Application Profile (Element Profile?), which we do not
>    currently have.
>
> -- If we go the route of One Big Namespace (as is being discussed now
>    in dc-architecture), would the elements of an Agent Core go into
>    this, or would we create for it a separate namespace (such as we
>    did, perhaps wrongly, for DCMES)?
>
> Would anyone have any reservations about having the UB approve a
> well-defined and sensible Agent Core?  How thorough would our review
> need to be?  Would we set a higher standard for these than for any
> other proposed elements?
>
> Tom
>
> _______________________________________________________________________________
> Dr. Thomas Baker                                            [log in to unmask]
> GMD Library
> Schloss Birlinghoven                                           +49-2241-14-2352
> 53754 Sankt Augustin, Germany                              fax +49-2241-14-2619
>