Thomas Baker wrote:
>
> On Fri, 23 Feb 2001, Stefan Kokkelink wrote:
> > That sounds like a precise and simple definition. So
> > the dumb-down of
> >
> > <dc:subject>
> > <rdf:value>12dget4564</rdf:value>
> > <rdfs:isDefinedBy>some strange schema</rdfs:isDefinedBy>
> > <dc:subject>
> >
> > is '12dget4564'? If this is recommended as best practice,
> > I don't see what do we gain by this? (But, I am quite new
> > to this, so please be forgiving ;-)
>
> The idea is that one should ask whether the strings are "useful for
> resource discovery" even in the absence of qualifiers. In this case,
> the string "12dget4564" would seem to be pretty useless as a piece of
> information unless you know where it comes from. So if this "some
> strange schema" were to be proposed to the Usage Board for
> certification as a Encoding Scheme qualifier, I would vote against it.
I should have used a real example, e.g. the Mathematical Subject
Classification (MSC). The code would look like this:
<dc:subject>
<rdf:value>19D10</rdf:value>
<rdfs:isDefinedBy>MSC</rdfs:isDefinedBy>
<dc:subject>
I could have used the PACS classification also. I hope
you would not vote against these two classifications, because
then we would have real problems to describe mathematical
documents at all. (That would frustrate my colleagues ;-)
I think the requirement for a classification scheme to
provide human readable strings as values is too strong.
> The idea is not to say that every subject heading or classification
> scheme in the world should go into dc:subject -- just the ones that
> seem useful for discovery.
MSC is usefull for resource discovery, see the
preprint index MPRESS
http://MathNet.preprints.org/
that is based on qualified
DC and allows searching and browsing in the
MSC classification.
> Instead of dc:subject, in this case, one
> could use an element like <my-app:sss>, where my-app is a namespace
> other than dc.
I strongly disagree here. Dublin Core should provide a framework
that allows the flexible use of classification schemes as objects
of dc properties. The framework only has to ensure that simple
resource discovery is always possible without knowing all
existing classification schemes.
> One counter-example is "Languages -- Phonetics", which is a useful
> string for indexing and discovery even if you do not know it comes from
> the Library of Congress Subject Headings.
You are right, the dumb-down algorithm works for this example.
> An intermediate example is Dewey Decimal Classification, which was
> approved as qualifier because (in my recollection) the text labels are
> useful strings even if the decimal numbers themselves are not helpful
> in their unqualified form.
I don't know the DDC systems in detail. Are the values of this
scheme the numbers or a combination of numbers and 'labels'?
Greetings,
Stefan
|