rdf:about and PURLs
Norman Gray
norman at astro.gla.ac.uk
Fri Jul 22 19:02:03 EDT 2005
Greetings,
This is partly an RDF question, but it's pretty immediately pertinent
in the context of writing CC metadata, hence the posting here.
Say I want to attach some licence declarations to a web document
referred to by a PURL -- what URL should I put in the rdf:about slot?
PURLs work by having a PURL resolver issue an HTTP redirect which
maps the PURL to the URL where the object is actually stored. Thus
the object has _two_ URIs which refer to it, with one being the URL
which actually retrieves it, and the other one being the one which
has the extrasyntactic/social/whatever committment to persistence.
It is this latter one which is the object's `real' name, and thus the
one which I, as its maintainer, wish to make statements about.
Naturally, therefore, it seems that I ought to make my RDF CC
assertions with reference to the PURL.
However the second, actually dereferencable, URL is a legitimate
alternative to the PURL, in both the practical sense that it works,
and the social sense that I can't legislate against that working one
being the one that some dunce passes around. If someone retrieves my
HTML document (or PDF or MP3) using this second URL, they are going
to find somewhere inside it an RDF block making a number of
assertions about a completely separate PURL, which has no connection
with the document they're retrieved other than those connections
supplied by common sense, native wit, and statements of the bleedin'
obvious (a most debased currency on the semantic web).
The problem is fairly obvious expressed in terms of PURLs. But it
also applies to DOIs, for example: statements about doi:10.1000/202
and <http://dx.doi.org/10.1000/202> are common-sensibly related; but
neither is in the least obviously related to <http://www.doi.org/
overview/041217DOIOverview.pdf>.
There's no RDFS isEquivalentTo property which I could use to assert
that statements about a PURL should also be taken to refer to that
PURL's mappings. There is in OWL, but dragging in the whole OWL
mechanism just to solve this problem seems the teensiest bit over the
top.
I'm firmly braced to be told that this is one of the most tediously
rehashed problems in this area (though inasmuch as it is a practical
problem of copyright assertions, it is not the same as the `what is
the referent of a homepage' hardy perennial). In which case, can all
of the above be turned into a request for a more obvious FAQ answer
on the creativecommons pages?
So which URl do I use?
All the best,
Norman
--
----------------------------------------------------------------------
Norman Gray : Physics & Astronomy, Glasgow University, UK
http://www.astro.gla.ac.uk/users/norman/ : www.starlink.ac.uk
More information about the cc-metadata
mailing list