Re: question re triples in xml literals

Hi!

I'm also interested in how to view this situation. As I wrote in
<http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2007Sep/0272.html>
(where at the end I suggest this Atom scenario), I feel ambiguous
about it.

Perhaps I am more for allowing processing xml literals than against.
To me it comes more naturally due to the "chimeric" nature of RDFa:
data is (potentially) both "in" the document and "about" it.

By contrast, XMLLiterals in RDF/XML (and N3 etc. for that matter) are
cleanly separated from the RDF. In XHTML+RDFa though, I don't
necessarily view adding @property to an element as formally escaping
its contents. It "exposes" its content to a graph (as XML or stripped,
depending on content and how @datatype is used), but it is still
"live" XML, if you will.

Perhaps the question can be whether there are dangers in processing it
further, and/or if this will be generally viewed as so odd as not to
allow it. It has been assumed that it is, but is that really the case?
What are the merits of allowing it, opposed to the cases where one has
RDFa within a literal and *don't* want it there (or intends to "expose
it in a subsequent step"..)?

The positive case is where an explaining text about a resource,
desirable to be given as one of the statements about it (with e.g.
@property="dc:description"), also contains more precise details in
themselves desirable to be expressed as statements. IMHO (and possibly
unscientifically), literals are where the "purity" of the RDF model
ends and pragmatic reality comes into play. The ability to have some
parts of that "raw" material more precisely stated is a bonus that
RDFa could provide by this. In fact, this is kind of how I view RDFa
in general; processing literals is "just one recursion down".. ;)

The worst case possible could be along the lines of every element
having @property="rdfs:comment" in it... And in general, I can see the
potential unnatural feel. Although the RDFa document will be "DRY",
the resulting graph will contain overlapping statements repeating the
same meaning in an increasingly fine-grained manner.

I can definitely understand if at this stage in the game it cannot be
worked in, but I think it is worth evaluating from different
perspectives. As you say, if nothing more then as food for thought for
future work.

Best regards,
Niklas


On 10/23/07, Mark Birbeck <mark.birbeck@formsplayer.com> wrote:
>
> Hi Benjamin,
>
> Yes, all of your use-cases are very good. The problem is that in most
> of them you'd want to _remove_ the RDFa from the 'escaped' data before
> storing it as an XML Literal, since it is not actually a part of the
> literal, but instead, it's part of the top-level document.
>
> I don't think it's impossible to solve these problems though, and it's
> good to start collecting these kinds of use-cases for a future
> version. In the meantime, I think structured blogging will be a huge
> use-case anyway. :)
>
> Regards,
>
> Mark
>
> On 23/10/2007, Benjamin Nowack <bnowack@semsol.com> wrote:
> > On 23.10.2007 10:59:30, Mark Birbeck wrote:
> > >But I can't see this being resolved within the _current_ version of
> > >RDF in XHTML, since it would take too long, I'm afraid--it's quite a
> > >large change.
> > ok, thanks, I see. Would be nice to see support in a future version.
> > Structured blogging could be a huge use case if RDFa supported
> > that.
> >
> > Another (personal) one is a lightweight RDF (Schema) editor
> > where term notes that contain markup (like the per-term hints in
> > the FOAF spec) could be used to generate an RDF Schema, e.g.
> > [[
> > <div about="#Student" property="spec:TermNote" instanceOf="rdfs:Class">
> >    Student is a subclass of
> >    <a rel="rdfs:subClassOf" href="http://xmlns.com/foaf/0.1/Person">
> >    foaf:Person</a>. It has the following properties:
> >    <a rev="rdfs:domain" href="#university">university</a>...
> > </div>
> > ]]
> >
> > Or other annotations:
> > [[
> > <div about="#review1" instanceOf="an:Annotation" property="rdf:value">
> >  The <a property="an:annotates"
> >         href="http://burgerking.com/whopper2000">
> >         new whopper website</a> is
> >  <span property="rel:rating" content="10">really cool</a>.
> > </div>
> > ]]
> >
> > I'm taking the opposite approach for the spec editor now (very
> > structured RDFa, with the backdoor to auto-generate human-readable
> > summaries from the triples at some later stage), which is fine,
> > but I'm sure there are many use cases where advantages of HTML
> > (formatting, order, semi-structure) could be combined with
> > structured markup. Stuff like recipes(+ingredients) or
> > manuals(+tools needed).
> >
> > I guess a work-around for now is to simply run another
> > extraction process on selected XML Literals after the main
> > processing is done, but it would be nice to do it in a
> > single-pass operation.
> >
> > Anyway, cheers,
> > Benji
> >
> > --
> > Benjamin Nowack
> > http://bnode.org/
> >
> >
> >
> > >
> > >Regards,
> > >
> > >Mark
> > >
> > >
> > >On 22/10/2007, Benjamin Nowack <bnowack@semsol.com> wrote:
> > >>
> > >>
> > >> Hi,
> > >>
> > >> I'm just reading the syntax doc and have a question regarding the
> > >> "[recurse] = false" rule, once an xml object is found. A sample
> > >> use case would be the description of an rss item ā la hAtom, e.g.:
> > >>
> > >> [[
> > >> <div about="post.htm" instanceOf="rss:item" property="content:encoded">
> > >>    <h2 property="rss:title">A post</h2>
> > >>    <p property="rss:description" datatype="">
> > >>       <a about="#foo" rel="foaf:weblog" href="http://foo.com/">Foo</a>
> > >>       said that.
> > >>    </p>
> > >> </div>
> > >> ]]
> > >>
> > >> i.e. I'm trying to generate
> > >>
> > >> [[
> > >> <post.htm>  content:encoded "<h2>....</p>"^^rdf:XMLLiteral ;
> > >>             rss:description "Foo said that." .
> > >> ]]
> > >>
> > >> without repeating the post body. And I don't want to lose the
> > >> triples from the rss:title/foaf:weblog info.
> > >>
> > >> The syntax doc says
> > >>
> > >> [[
> > >> "if an author indicates that some branch of the tree should be
> > >> treated as an XML literal, no further processing should take
> > >> place on that branch"
> > >> ]]
> > >>
> > >> which (to me) suggests that no "sub-triples" should be generated
> > >> whenever an XMLLiteral is encountered. How would I have to
> > >> change the code above to markup an XMLLiteral that contains
> > >> structured RDFa information?
> > >>
> > >>
> > >> Cheers,
> > >> Benji
> > >>
> > >> --
> > >> Benjamin Nowack
> > >> http://bnode.org/
> > >>
> > >>
> > >>
> > >
> > >
> > >--
> > >  Mark Birbeck, formsPlayer
> > >
> > >  mark.birbeck@formsPlayer.com | +44 (0) 20 7689 9232
> > >  http://www.formsPlayer.com | http://internet-apps.blogspot.com
> > >
> > >  standards. innovation.
> > >
> >
> >
>
>
> --
>   Mark Birbeck, formsPlayer
>
>   mark.birbeck@formsPlayer.com | +44 (0) 20 7689 9232
>   http://www.formsPlayer.com | http://internet-apps.blogspot.com
>
>   standards. innovation.
>
>

Received on Tuesday, 23 October 2007 15:37:53 UTC