Re: Review of Representing Content in RDF

2008/10/1 Johannes Koch <johannes.koch@fit.fraunhofer.de>:
> Hi David,

I've not been following list discussion very well, but looking over
the current doc I agree with most of David's comments -

>> The document seems to make sense.

+1

>> 1. I think it would be good to mention GRDDL as a way to get a
>> finer-grained RDF representation of an XML document.
>
> Can you please explain, how you think GRDDL can be used "to get a
> finer-grained RDF representation of an XML document"?

To put it loosely, the spec in question treats content as a blob,
whereas GRDDL can allow you to extract statements embedded in the
content. A simple ref would be desirable, IMHO.

>> 2. When showing RDF, please also show the RDF in a more humanly readable
>> format than RDF/XML, such as N3.
>
> Well this seems to be a matter of taste. We will discuss this.

I'd go along with Turtle (sorry to be pedantic David :-) for use in
spec text as informative material, with a link to the RDF/XML
representation as normative.

>
>> 3. What is the correspondence between class Content and terms like
>> "entity" used in RFC2616?
>
> cnt:Content resources can be used as object resources  for triples with
> http:body property as metioned in
> <http://www.w3.org/TR/2008/WD-HTTP-in-RDF-20080908/#bodyProperty>.

Tricky one, no comment.

>> 4. There is a DoctypeDecl class for DTDs.  What about other schema
>> languages, such as XML Schema or Relax NG?   Maybe DTD needs to be treated
>> specially because it can be embedded in the XML?
>
> You cannot include a document type _declaration_ in an XMLLiteral, can you?
> So we have to treat this part of an XML document differently. XML Schema
> uses attributes to reference external schemas. So that's no problem.

Off the top of my head that sounds right. But all the same, it seems
about time we should consider deprecating DTDs in these kind of
circumstances. XSDs hurt the rest of that head I mentioned, but aren't
difficult to use normatively being a W3C spec. Relax NG shema have a
lot of benefits over either specification, it'd be good at least to
see an informative version if feasible within the group member's
available time.

>I'm not
> sure whether there is a referencing mechanism for Relax NG.

Dunno.

[snip]

>>  Appendix B shows an "Allowable types" column, which might imply that
>> other types are not allowable.
>
> I think we should change this to "Suggested types".

Agreed. No reason to close doors unnecessarily.

Cheers,
Danny.

-- 
http://dannyayers.com
~
http://blogs.talis.com/nodalities/this_weeks_semantic_web/

Received on Wednesday, 1 October 2008 18:18:12 UTC