Hi Jon,
On 22 June 2012 14:37, Jon Phipps <[log in to unmask]> wrote:
> Hi Karen,
>
> Your Use case isn't actually a dilemma for the DCAM model. DCAM requires
> properties to be defined as either literals or no literals, as does OWL-DL.
Oh... I don't think there is anything in (current) DCAM which requires
this. Or speaking for myself, it wasn't my intention that it did/does
say this, anyway.
The DCAM just says that each individual "statement" contains a
"literal value surrogate" or a "non-literal value surrogate", but it
allows for the possibility that a single property may be referenced in
two different "statements", with a literal value surrogate (RDF
literal object) in one "statement" and a non-literal value surrogate
(RDF URI or blank node as object, plus other triples with that object
as subject) in a second "statement".
> You can certainly have a property that isn't DCAM compatible that can accept
> either strings or things, like the original DCMES, but it wouldn't be DCAM
> compliant (one of the motivations for assigning ranges to the original DCMES
> in DCTERMS).
As I understand it, the original DCMES elements/1.1/ properties are
"DCAM-compliant", and there are examples of their use in e.g. the
(DCAM-based) DC-HTML specification
http://dublincore.org/documents/dc-html/
The tendency towards highlighting the use of the DC Terms properties
over the DCMES properties in our documentation emerged primarily,
IIRC, out of an interest in trying to establish greater consistency in
patterns of use in RDF (the primary focus of discussion was the
thing-agent properties, I think), not because DCAM "ruled out" such
variation.
Pete
--
Pete Johnston
|