Re: an alternative for microformat-like simplicity

Just like Michael and Toby, I do have my issues...



Ben Adida wrote:
[snip]

> Markup in the page:
> 
> <div about="#me" prefix="http://myvocab.org/#">
>   My name is <span property="name">Ben Adida</span>
> 
>   and my email is
> 
>   <a rel="email" href="mailto:ben@adida.net">
>     ben@adida.net
>   </a>
> </div>
> 
> 
> Vocabulary Definition at http://ben.adida.net/vocab, using RDFa:
> 
> <div about="#name" typeof="rdf:Property">
>     <h4 property="rdfs:label">name</h4>,
>     which corresponds to
>     <a rel="owl:sameAs"
>        href="http://xmlns.com/foaf/0.1/name">
>          foaf:name
>     </a>.
> </div>
> 
> <div about="#email" typeof="rdf:Property">
>     <h4 property="rdfs:label">email</h4>,
>     which corresponds to
>     <a rel="owl:sameAs"
>        href="http://rdfs.org/sioc/ns#email">
>          sioc:email
>     </a>.
> </div>
>

I presume that you messed up the example and you wanted to use
'http://myvocab.org' in the second part (instead of
'http://ben.adida.net/vocab') right? If not, then I too, like Michael,
have problems understanding how you get there...

However, I must admit that I am not 100% sure what is being proposed in
practice. The issue is what the requirement is on an RDFa processor.
Indeed, I can look at it with two different ways:

1. A comformant RDFa processor is required to dereference the URI
http://myvocab.org/#, extract the RDF data there, make the
identifications through, say, sameAs, and provide the output in terms of
the, in this example, SIOC triples. As far as I can see, this is NOT
what you propose, but I am just trying to make it explicit in case I
misunderstand...

However if that _is_ what you propose, I do feel uneasy. It does create
a significant extra load on RDFa processors, which are meant to be
relatively easy to implement and deploy. This proposal requires various
RDF parsing modules (we cannot _require_ that the vocabulary must be
described in RDFa, can we?), at least some sort of an OWL 2 RL level
inference on RDF/OWL graphs, corresponding queries, built into it... Of
course, all that is doable, but that is way too much to ask for an easy
deployment of RDFa (which is a major selling point...)

2. But, if that is _not_ the case, ie, RDFa processors are not required
to do all that, then I do not understand what the proposal really brings
that is not, in theory, part of the RDF/OWL infrastructure as of today.
If I take your example above, using simply @xmlns (to stay with the
current standard) I would produce with an RDFa processor

<#me> <http://myvocab.org/#name> "Ben Adida" .
<#me> <http://myvocab.org/#email> <mailto:ben@adida.net> .

And if I am a vocabulary maintainer, I would put an RDF file in whatever
syntax I wish (RDFa or not) to http://myvocab.org/# which says

<http://myvocab.org/#email>
          a rdf:Property;
          owl:equivalentProperty <http://rdfs.org/sioc/ns#email> .

and then a full-blown, OWL aware processor would do the right thing by
dereferencing <http://myvocab.org/#email>. That is, in theory, what
happens _today_ on an RDF/OWL environment. I fail to see what is new in
this case...

3. As a side issue, I fully agree with Toby. _If_ we use OWL terms, we
should be careful what we require (although users will get it wrong...).
owl:equivalentProperty is the right thing to do even in OWL (2) Full,
for example and not sameAs, stuff like that...

So: I am not sure I fully grasp the proposal...:-(

Cheers

Ivan

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf