Hello,
Seth Russell wrote:
> The draft document
> at http://dublincore.org/documents/2001/03/09/dcmi-namespace/
> states:
>
> As DCMI registry infrastructure evolves, it will become the repository of
> the canonical definitions of all DCMI terms (in the form of schema
> declarations that are machine-processable, but also can provide
> human-readable definitions). Thus, three currently approved DCMI metadata
> packages have the following namespaces:
>
> http://dublincore.org/2000/03/13/dces
> http://dublincore.org/2000/03/13/dcq
> http://dublincore.org/2000/03/13/dctype
>
> Seth asks....
>
> Yet if one dereferences those uri, there are no "machine-processable" schema
> there to be found. How do machines get to the official schema documents?
This is the dd-name vs. dd/name question we are floating. We (as authors of
this document) prefer the pattern http://dublincore.org/yyyy/mm/dd/name, but
what we are trying with this exercise is to determine if such a name is going
stop the world for a day, if not we'll adopt the pattern and get on with our
lives, if so well reconsider.
Therefore the listed links are not live (so to speak) to prevent further
confusion until the matter is settled.
We fear that folks like yourself have been introduced to draft rdfs (maybe by
myself) via an unofficial namespace and as we have been working though the
issues. We are suggesting the change (dd/name), will that hurt much?
>
> This is of some intereste to me because I have been attempting to automate
> the inclusion of the dc terms in a SWAG dictionary [1].
>
> [1] http://robustai.net/swag/dictionary.html
>
> Currently I am using these URI to arrive at the schemas that I use to update
> the dictionary.
>
> http://dublincore.org/2000/03/13-dces
> http://dublincore.org/2000/03/13-dcagent
> http://dublincore.org/2000/03/13-dctype
> http://dublincore.org/2000/03/13-eor
>
> What is the recommended method for people and agents to find your
> recommended schema documents ?
I would suppose this could be a set of published canned queries against the
registry database (still only a prototype), or maybe an RSS feed or some such
publishing mechanism. We have to get happy with the name pattern and crank the
crank for a while to make sure all the bugs are out of the process. As always,
we welcome any suggestion, comments or short debates you may have on the
matter.
>
> Also I note that the class "Event" is defined differently in dcagent and
> dctype. Is this intentional, and if so, is it wise?
>
> Seth Russell
These are unofficial drafts that need review and repair. I have noted you
collision and will investigate it.
We will distribute an update when this process has been completed (after the
end of March, the timeframe for accepting the namespace pattern proposed),
since the rdfs documents need to reflect the official namespace.
Cheers Tod...
|