On mån, 2007-02-26 at 22:49 +0100, Thomas Fischer wrote:
> Hi Mikael!
>
> > > We aren't actually putting the place in the CV. That is
> > > impossible to do. Rather we are putting a description of
> > > the place in our dc:subject.
> > > In fact, we're using Getty's description of the
> > > place, aren't we?
> >
> > All metadata implies at least one level of indirection
> > already (except for literal strings). So, the indirection you
> > are asking for is already implicit in the model.
> ...
> > In both cases, the referred value *is* New Zealand, but
> > referenced in different ways.
>
> I am wondering if this "implied level of indirection" shouldn't be made
> explicit and would thus make life and understanding easier.
> For me, it would be easier to follow the concepts if we had
> - objects in the real word ("outside" of the description)
> - references (pointers) to these objects.
In my opinion, the DCAM does define just that:
1. resource: Anything that might be identified.
2. Resource URI: reference to a resource
>
> If you define a resource as a reference to or an identifier of some object
> (material or not, "New Zealand", "Philosophy", a particular stone or human,
> or an XML-based expression), then the relations described in the DCAM become
> relations of these identifiers. There may still be problems to be dealt with
> in the relation between the identifier and the identified, but this would be
> out of scope for the DCAM. (In some sense, there are more or less precise
> identifiers of objects, like the name of an author (bad) or a reference to
> her authority file (good).)
>
> But this would remove the ambiguity mentioned by Joseph Tennis: If New
> Zealand is a resource, how can it be the value of a statement.
If you look closely at the DCAM description model, you will find that a
statement contains:
* zero or one value URI
* zero or one vocabulary encoding scheme URI
* zero or more value representations, which may be
* value strings, or
* rich representations
Nowhere in the definition of a statement will you find the value itself.
In fact, DCAM descriptions *never* contain any resource at all, but only
references (identifiers) and representation (string or other digital
ones). So I don't see a problem - I believe we *are* being consistent.
>
> resource: Anything that might be identified.
>
> then you cannot say anything about it before you identify it: If you have a
> heap of sand, you cannot say anything (create a statement) about a
> particular grain of sand before you _identify_ it (e.g. paint it blue and
> refer to the "blue grain").
There are many ways of identification - inverse functional properties
for example.
The FOAF vocabulary explicitly recommends *against* using identifiers
for people, and it has the exact same notion of resource as the DCAM has
(through the common RDF heritage). But still, using FOAF you can say a
lot of things about unidentified persons. So again, I'm not sure I
really see where the problem lies.
/Mikael
>
> All the best
> Thomas
>
> ––
> Dr. Thomas Fischer
> Metadata and Databases
> Göttingen State and University Library
> 37070 Göttingen
> Germany
> Tel.: +49 551 393883
> and +43 662 621498
>
--
<[log in to unmask]>
Plus ça change, plus c'est la même chose
|