Re: better support for bnodes

Hi Ben,

yes, I like it! It looks nice. I have only two caveats:

1- technically it is nice. But it may be difficult to explain the
mechanism for a lambda user who does not really know what is going on in
RDF... It can be done, probably, but it will not be obvious.

2- a side issue, which goes actually against the example I gave.
Although the usage for <ul><ol><nl> for Bag/Seq/Alt seems to be an
obvious choice (and works well), what I miss is the encoding of lists
(maybe that was already discussed elsewhere, sorry if I missed it). As
you know, the Bag/Seq/Alt are not really part of the core semantics of
RDF(S), they are 'added on' features, so to say. Moreover, there are
lots of discussions about removing them from RDF altogether (the fact
that the _1, etc, bring in a potentially infinite amount of resources
was a significant problem on the theoretical side, for example). Lists
are much more important, they are semantically more stable. As such, I
would prefer the <ul>...</ul> to map on lists by default (and some extra
tweaks should be done for the containers). As I said, maybe this is
already settled, but I have not seen it.

Cheers

Ivan

Ben Adida wrote:
> 
> A followup on Mark's proposed solution to make it easier to have bnodes.
> Short version: I really like it.
> 
> Let's see if I can summarize the proposal: if @rel is present on an
> element without @href, then the element itself becomes the object of the
> triple. Thus, if the element has an @id, then the object is this named
> ID, but, if it doesn't, a bnode id gets generated and that's the object.
> 
> I don't mind the wrinkle Mark brought up about possibly orphaned bnodes
> if an @href is later added.
> 
> I now wonder if this can be extended to @rev. In other words, if @rev is
> present without an @href, then the *subject* of the triple is the
> current element, and the object is resolved using the normal upwards
> traversal to find an about. Are there wrinkles here that I'm not seeing?
> 
> Next, let's look at RDF containers. What if we wanted to have a list of
> authors for a page? I like the following syntax:
> 
> <ul rel="dc:creator">
> <li>Mark Birbeck</li>
> <li>Ivan Herman</li>
> </ul>
> 
> and this would exactly parse as expected using Mark's proposal and the
> existing RDFa Containers proposal [1]:
> 
> _:ul0 rdf:type rdf:Bag .
> _:ul_0 rdf:_1 "Mark Birbeck" .
> _:ul_0 rdf:_2 "Ivan Herman" .
> 
> "" dc:creator _:ul0 .
> 
> This is looking promising!
> 
> Now, let's take a look at the entire Turtle example provided by Ivan and
> try to write it up using Mark's proposal:
> 
> ===================
> 
> <div about="[:DJDKWBDADIH05]" class="bibtex:Article">
> <span property="bibtex:title">Do you see what I mean?</span>
> 
> <ol rel="bibtex:author">
>   <li><span property="foaf:name">David J Duke</span></li>
>   <li><span property="foaf:name">Ken W Brodlie</span></li>
>   <li><span property="foaf:name">David J Duce</span></li>
>   <li><span property="foaf:name">Ivan Herman</span></li>
> </ol>
> 
> <p rel="bibtex:journal">
>   <span property="bibtex:name">IEEE Computer Graphics &amp;
> Application</span>
> </p>
> 
> <span property="bibtex:volume">25</span>
> <span property="bibtex:number">3</span>
> 
> <p rel="bibtex:date">
>   <span property="bibtex:month" content="3">March</span>
>   <span property="bibtex:year">2005</span>
> </p>
> 
> <p rel="bibtex:page">
>   <span property="bibtex:startPage">6</span>
>   --
>   <span property="bibtex:endPage">9</span>
> </p>
> 
> </div>
> 
> ====================
> 
> 
> I'd say that's starting to look quite nice!
> 
> 
> -Ben
> 
> 
> [1] http://www.w3.org/2001/sw/BestPractices/HTML/2006-rdfa-containers
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
URL: http://www.w3.org/People/Ivan/
PGP Key: http://www.cwi.nl/%7Eivan/AboutMe/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf