W3C home > Mailing lists > Public > semantic-web@w3.org > September 2006

Re: "role" and inheriting "about"

From: Ben Adida <ben@adida.net>
Date: Mon, 11 Sep 2006 08:11:24 -0400
Message-ID: <4505526C.7090601@adida.net>
To: Yoshio Fukushige <fukushige.yoshio@jp.panasonic.com>
CC: semantic-web@w3.org, public-rdf-in-xhtml-tf@w3.org


Yoshio,

Good points. Let's see if I can help clarify things a bit more.

Yoshio Fukushige wrote:
> 
> Should @role be a way of declaring a type of a HTML ELEMENT,
> not a way of declaring a new resource?
> 
> Hmm.

In RDF, those two things are quite similar.

> So, although the Primer says in the 3rd paragraph of 2.2 Publishing An Event
> (excluding the examples):
> 
> [[[then Jo declares a new event
> (example code)\
> then, inside this event declaration, Jo can set up the event fields, ...]]]
> ,
> 
> the @role doesn't actually introduce a resource by itself.
> 
> Is this what you mean?

Well, it does introduce the resource by giving the current HTML element
a type. The question is whether every child node should then have this
element as a subject. Right now, the task force doesn't think so,
especially since it's really easy to do this already with ABOUT:

<div role="cal:vevent" id="event1" about="#event1">
...
</div>

> Then I can understand that @role should not block the inheritance,
> although the text in 2.2 may need to be changed and I got unsure about 
> the reason we need @roles...

We will definitely take this into account for the next round of editing.

> Yes, it embarrasses me a lot(who is a validation addict ;-))

Well it validates in XHTML2 :) We're working on XHTML1 in the coming weeks.

[...]

> I further do not want for @class to do any function in RDFa.

I hope to convince you otherwise. Let me know what you think after the
arguments below.

> I've used many @class'es for declaring the semantic classes (or functions ) 
> in the document, of the document fragments being described.

Exactly, and these are effectively RDF triples with local types.

> e.g. 
> <div class="section">
> <h2>Next BBQ Party</h2>
>  ...
> </div>
> ( What I want to talk about is the next BBQ party, not the section itself )
> 
> I can embed the same triples without using @class'es,
> by using (@about and) link with @rel="rdf:type" as in Elias's example, right?

So, let's take your example:

<div class="section">
<h2>Next BBQ Party</h2>
</div>

This would yield, in our current thinking about RDFa:

_:div0 rdfs:type <section>

where <section> is a relative URI to the current HTML base, and _:div0
is a blank node (meaning not addressable by the outside world.)

We think that's exactly the triple meant in that document, and since the
type is entirely local, it will not cause any weird inferences. In fact,
it's really expressing, in RDF, the semantics of that document. The
meaning of "section" is local to your document, so the type is scoped to
the current HTML base.

Then, this approach allows you to do:

<div class="cal:vevent">
meeting next week.
...
</div>

which does exactly what you want it to do. Plus, you can declare a few
classes, as usual:

<div class="cal:vevent section">
meeting next week.
...
</div>

> Thank you for saying so.
> 
> I'd be happy if I could help you by providing "unintended readings" ;-)

That is incredibly valuable, so please continue!

For specific RDFa feedback, I also recommend cc'ing
public-rdf-in-xhtml-tf@w3.org

-Ben
Received on Monday, 11 September 2006 12:11:33 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:41:53 UTC