Re: Using RDF to describe DOM tests

> [...] Could you please indicate what the vocabulary would have
> to look like on order to make the assertions that Dublin core
> doens't allow? What would we need to think of?

I think that there are going to be three types of data that you'll
need:-

   1) Basic information
   2) Framework and structuring
   3) Scoped information

For example, stuff like DC and RDFS will provide 1), and EARL may
provide 1). 3) will be the information that is only applicable to DOM
tests, or that hasn't already been invented elsewhere. For a start, I
already needed to make up "javaId" and put it in the example - as a
sub property of dc:identifier. In other words, :javaId is a
specialization of dc:identifier.

Note that it is better to use other people's semantics if they've
already been used, or (which is what we do) to state the equivalences
and so forth between them. As you'll probably be usiong XML Schema, it
is quite difficult to do this, unless you take a look at the work done
on adding RDF to XSD... of which there has been very little. A better
idea would be to use RDDL at the namespace, and then point to the XSD
and RDF.

However, I think that you should also be very careful to create the
right depth of granularity. If another persons term doesn't quite fit,
make up your own! That's what I did with the javaId thing - Curt used
dc:identifier, which although adequate, does not provide any more
information to a processor than "this thing is an identifier". :javaId
on the other hand could be very specific about what it points to (you
could define that yourselves). Take a look at some of the stuff in XML
GL [2] for more details about exporting semantics and so forth.

[1] http://www.rddl.org/
[2] http://www.w3.org/WAI/PF/xmlgl

--
Kindest Regards,
Sean B. Palmer
@prefix : <http://webns.net/roughterms/> .
:Sean :hasHomepage <http://purl.org/net/sbp/> .

Received on Monday, 28 May 2001 08:51:39 UTC