Re: Playing with the model

> Sean and Danbri, Libby said this was "implicit reification."

This doesn't seem to be "fake" or "implicit" refification going by
Libby's description of it in her "RDF Annotations" paper [1]. In that
model, things are grouped under bNodes and given properties. With this
model, things are just asserted in the model with no grouping or way
of preserving the context information.

> What are the implications on queries?

The implication is that you either can't have two different
evaluations in one document, or you can't query it at all. I'm not
sure how this would fit into the reification by hypertext idea [some
notes on that a bit further down].

Consider the following problem (which I've already outlined twice at
[2] and [3], but here it is again). You may have one document which
contains:-

   :Sean earl:asserts <http://example.org/> .
   <http://example.org/> earl:passes :WCAGA .

and then in another document you have:-

   :Wendy earl:asserts <http://example.org/> .
   <http://example.org/> earl:fails :WCAGA .

notwithstanding the fact that asserting a bit of WebContent doesn't
make much sense (and could be misconstrued as asserting the semantics
of the WebContent, which is certainly *not* what we want!), when you
merge the RDF documents you get a contradiction:-

   :Sean earl:asserts <http://example.org/> .
   :Wendy earl:asserts <http://example.org/> .
   <http://example.org/> a earl:WebContent;
      earl:passes :WCAGA; earl:fails :WCAGA .

It's impossible to query it usefully. Who said that it passes, and who
said that it fails?

And then there's reification by hypertext, treating the two documents
as things that we don't necessarily believe. The problem with that is
that XML RDF has no way to quote a set of statements other than using
log:quote as a parse type. It also has no way to identify the root
context of the document, unless you make a link from the document to
it's semantics using log:semantics or something.

Of course, the other big problem with that is that you can't have lots
of different evaluations (i.e. with different assertors) in one
document. We discussed that months ago, and the people on the call
(Jim and Nick, IIRC) didn't think it was a good idea.

[1] http://ilrt.org/discovery/2001/04/annotations/#fake
[2] http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2002Feb/0053
[3] http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2002Feb/0056

--
Kindest Regards,
Sean B. Palmer
@prefix : <http://purl.org/net/swn#> .
:Sean :homepage <http://purl.org/net/sbp/> .

Received on Thursday, 28 February 2002 15:42:27 UTC