[Fwd: RDF Issues]

This was originally sent privately to Brian McBride as accessibility 
comments - just making sure people see it.

cheers

Chaals

--
Charles McCathieNevile             Fundacion Sidar
http://www.sidar.org/

-------- Original Message --------

There are two issues. The first is the lack of a standardised
"abouteachprefix" in RDF. The second, and I think more serious, is that there
is no defined way to talk about a view of a document, where those views are
defined by using a URIRef for a particular MIME type.

I have tried to lay them out below, but am not writing terribly clearly
today, so i hope this is enough to go on with. It might be useful to try and
talk about this some more tonight or tomorrow - I will be around. I thought
it was better to send something sooner than the perfect version later.

(and finally due to various process wierdness please note that this email is
supposed to be from me as an invited expert to the WAI PF group, representing
La Fundaciòn Sidar, as charles@w3.org)

cheers

Chaals

Issue 1 - aboutEachPrefix

"aboutEachPrefix" seems to be a particular instance of a class of things.
There are many cases where people want to talk about a set of objects without
having to enumerate the list:

 - things in a particular namespace (the original aboutEachPrefix case) are
published by the owner of that namespace;

 - the infinite set of times within a particular range for one calendar fall
within the range of a particular date in another calendar (think about how
many nameable times on the 1st of Ramadan 1476AH fall on a particular day in
the gregorian calendar, as measured in Paris)

 - the homepages of staff at an organisation represent people at that
organisation - that list can be provided by a Web service at any given time,
or it is possible to determine whether a given resource is in that set, but
it cannot be enumerated cleanly in a static document.

(EARL statements about resources of this type are an example of a use case in
accessibility - for example that these have been tested and found to contain
appropriate structure or be valid before they were published).

I am told by Jeremy Carroll that this problem can be dealt with by Jena and
modelled using OWL. The issue is why this is not something that a "basic" RDF
processor should be able to deal with. In the aboutEachprefix case it was
(theoretically) available in basic processors which did not implement other
"optional" specifications.

===

Issue 2 - referring to a particular view of a resource

According to the RDF Concepts document, a statement that
<http://www.example.org/foo#svgView(viewBox(0,200,1000,1000))> <foo:isLike>
<bar:something>

refers to something which is defined in the version of
http://www.example.org/foo that has an RDF MIME-type. So if there is nothing
returned with that MIME-type then the statement doesn't have a defined
subject.

It seems you also can't rely on content negotiation to say something like

<http://www.example.org/foo#svgView(viewBox(0,200,1000,1000))>
<foo:viewableAs> <mime:image/svg+xml> .

<http://www.example.org/foo#svgView(viewBox(0,200,1000,1000))>
<foo:describedInHTML> <http://www.example.org/foo#someDesc> .

An accessibility use case is describing particular views of documents under
particular conditions - for example giving some information about what kinds
of 'delivery context' can make sense of that part of a resource, or pointing
to another resource which can be used to understand a resource which a person
with a disability can't use directly. It is often important to talk about a
part of a document, because some parts will be accesible to people and others
won't, and they want to know which is which. In the presence of
content-negotiation, statements will be about fragments of versions other
than the RDF one.

Received on Wednesday, 12 March 2003 09:58:35 UTC