Re: More direct conversion from microdata to RDFa?

On Oct 13, 2011, at 17:58 , Stéphane Corlosquet wrote:

> Hi Ivan,
> 
> On Wed, Oct 12, 2011 at 12:02 PM, Ivan Herman <ivan@w3.org> wrote:
> Well... I must admit that I do not like that. It is a very personal thing, but the aspect I see as a problem in
> microdata is the fact that the usage of itemtype conflates two very different concepts (at least for me), namely the
> choice of a vocabulary and the typing of a concept, in one place. I am pretty uneasy going down that road in RDFa
> solely for the purpose of a better translation...
> 
> yes, this proposal comes from the difference in modeling between the 2 syntaxes. I believe that namespaces and types are two different concepts for RDF folks, but I'm not sure the web developer community really care about making that distinction, hence the design decision in microdata.

This is a hypothetical reason. I am not sure our goal is to mimic all aspects of microdata without further proofs.

> I'm not advocating to remove this distinction from RDFa, but simply to let the processor deal with it.

But having them both does muddle the waters, at least in my view. There are a number of corner cases (some of them are actually discussed in conjunction with the microdata->RDF mapping, too) which we may have to deal with:

- If there are several items in @typeof, which one determines the vocabulary? The first one? Doesn't that create issues for authors as a possible source of errors (in my view it does)
- @typeof values may not be full URI-s (in contrast to microdata). Ie, a @vocab expansion is also valid for @typeof; what are exactly the rules there? 
- What happens if there is a @typeof but there is also a valid @vocab statement somewhere 'up' in the tree?
- What are the exact rules of deducing a vocab URI from a type value?

All these issues can be answered one-by-one, of course, and none of these demand rocket science. But they do make things way more complicated.

As you said in your earlier response to Toby, @typeof is already overloaded insofar as it also creates a new subject in some cases. That is true and, to be honest, this is one of the aspects of RDF1.0 that I really do not like at all. I would prefer to remove any overloaded features from @typeof (although, alas!, that ship has already sailed) rather than adding to it.

To make it clear: if the majority of the WG decides for this feature, I will not lie down the road. But you still have to roll over me:-)

As I said before: sorry, Stéphane, this time we disagree...:-)

Ivan


> 
> Steph.
> 
>  
> 
> Sorry:-)
> 
> Ivan
> 
> 
> 
> 
> On Wed, October 12, 2011 2:56 pm, Stéphane Corlosquet wrote:
> > Converting microdata to RDFa could be made easier by bypassing the vocab
> > attribute. Microdata incorporates the concept of vocabulary in the data item
> > type via @itemtype. Coming from a microdata perspective, the concept of
> > vocabulary URI is an extra element (different from the item type) which
> > could be avoided by the following proposal. Since there is a mechanism being
> > designed to infer a vocabulary namespace from the microdata @itemtype, RDFa
> > could use a similar mechanism, and use the first token of @typeof to
> > construct the namespace that we currently put in @vocab.
> >
> > Consider the following microdata snippet:
> >
> > <div itemscope itemtype="http://schema.org/Person>
> >    My name is <span itemprop="name">John Doe</span>.
> > </div>
> >
> > Currently to convert to RDFa, you need to split @itemtype into two
> > attributes: @vocab and @typeof. What I'm suggesting is a direct mapping from
> > @itemtype to @typeof without the need of @vocab, where @typeof would include
> > the full URI:
> >
> > <div typeof="http://schema.org/Person>
> >    My name is <span property="name">John Doe</span>.
> > </div>
> >
> > This would ease the RDFa to microdata and vice versa from a human
> > standpoint.
> >
> > Note that I'm not suggesting to drop @vocab, but simply to make it optional,
> > which would make the common microdata snippets a no brainer to convert to
> > RDFa by a simple attributes string replace. In the processing model
> > sequence, step 3 would have to include an extra step for the case where
> > @vocab is missing, and use the same mechanism as Gregg is defining for
> > microdata to RDF conversion. Here is a suggestion:
> >
> > [[[
> > 3. Next the current element is examined for any change to the default
> > vocabulary via @vocab. If @vocab is present and contains a value, its value
> > updates the local default vocabulary. If the value is empty, then the local
> > default vocabulary must be reset to the Host Language defined default. If
> > @vocab is not present but the first token of @typeof is an absolute
> > URI, construct the local default vocabulary by removing everything following
> > the last SOLIDUS U+002F ("/") or NUMBER SIGN U+0023 ("#") in the first token
> > of @typeof.
> > ]]]
> >
> >
> > Thoughts?
> >
> > Steph.
> >
> 
> 
> --
> Ivan Herman, W3C Semantic Web Activity Lead
> URL: http://www.w3.org/People/Ivan/
> FOAF: http://www.ivan-herman.net/foaf.rdf
> 


----
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