David,
In terms of the DCMI Abstract Model [1], a "description"
describes one (and only one) resource. The example below
refers to five "resources" in the DC-AM sense: the Described
Resource, its Author, the Point of Contact, the Owner,
and the Metadata Author [*].
In the DC-AM sense, the descriptions of these five things,
when grouped in a metadata record, would constitute a
"description set".
For ISO 19115, then, the agent properties (Name, Position,
Role, etc) are used in descriptions of the Author, Owner, etc
-- i.e., each of these entities is described with a separate
"description" in a "description set" (metadata record).
As dcterms:creator, dcterms:publisher, and dcterms:rightsHolder
are "properties", and properties by definition _relate_
two resources to each other, these properties relate the
Described Resource to the Author, Point of Contact, and Owner,
respectively.
As you say, this is more cleanly done using separate XML
elements than by trying to smush the information into a
string. If done cleanly and coherently, the XML data would
be transformable into RDF triples (which express the data
relationships in a generic, portable way) by using GRDDL
transforms [2], as is done with the syntax DC-DS-XML [3].
Tom
[1] http://dublincore.org/documents/abstract-model/
[2] http://www.w3.org/TR/grddl/
[3] http://dublincore.org/documents/dc-ds-xml/
[*] The latter arguably implies Described Resource Metadata
as a sixth resource...
On Wed, 2008-09-26, David Bromage wrote:
> The ISO 19115 discovery metadata profile for spatial datasets has a very
> granular way of identifying agents responsible for some aspect of the
> described resource. It has elements (in the XML sense of 'element') for
> the author (cf dcterms:creator), 'point of contact' (cf publisher),
> 'owner' (cf rightsHolder) and 'metadata author' (no DC equivalent). Each
> can be described with the elements:
> - Individual name
> - Organisation name
> - Position
> - Role
> - Address
> - - Delivery point (physical address)
> - - City
> - - Administrative area (state/province)
> - - Postal code
> - - Email
> - - Country
> - Telephone number
> - Fax number
> - Organisation URI
> - Email
>
> Is there a way existing ISO 19115 agent metadata could be re-used in DC
> descriptions without reinventing the wheel? In the past this could have
> been done using DCSV but an XML-based option would be preferable. A few
> possible options I thought of are:
> - Like-named terms in an application profile referenced as a related
> resource (the agent is the related resource)
> - Concatenating multiple ISO 19115 values into a DC text string value
> (less useful for searching)
>
> Any comments?
>
> Regards
> David
>
>
> David Bromage
> Policy and Strategic Projects
> Government Information Management Branch
>
> National Archives of Australia
> PO Box 7425
> Canberra Business Centre ACT 2610
>
> T (02) 6212 3731 F (02) 6212 3989 M 0418 394 778
> [log in to unmask]
> www.naa.gov.au <http://www.naa.gov.au/>
>
>
--
Dr. Thomas Baker <[log in to unmask]>
Director, Specifications and Documentation
Dublin Core Metadata Initiative
|