Re: RDFa and Microformats

Hello Toby nice to talk to you again....

Toby A Inkster wrote:
> Martin McEvoy wrote:
>
>> My response would have been *If* the RDFa community decided that
>> Microformats are expressing semantics( which is Acknowledged ) and
>> should be somehow ported to RDFa, Microformats do  this using mainly
>> using just Class attributes If you are going to "bring them to the fold"
>> @class should be added to RDFa as one of its properties.
>
> I think you'll find it's actually a lot more complicated than that. 
> Firstly, in microformats different classes have entirely different 
> parsing models. In hCard for instance:
>
>   vcard = Look inside for properties.
>   fn = Use the contents as the property.
>   url = Ignore contents, use the link.
>   tel = Look inside for class="type" and class="value".

You over complicate microformats if you think of it in that way,  they 
have a very basic parsing model which goes a little something like ....

root classes can be determined easily by asking the question "does" this 
class have children? f yes then it is a root class name, If no then this 
is a property,  to which the contents will either be text, or a url, if 
the contents are text look for @title Its a little more involved than 
that but not by much by the way I am not talking about any @class name 
just ones used in a constrained vocabulary such as Microformats.
>
> There are all kinds of special cases. Even in hAudio, which you've had 
> a part in designing, and I think is certainly one of the better 
> authored microformat specs, there are some parsing oddities. i.e. 
> figuring out the type of audio being described (is it an album, is it 
> a recording, or is it a track on an album) requires looking for the 
> presence of two different properties and deciding based on the 
> combination of which exist; 
True,  oddities should be perhaps aired on the microformats-new mailing 
list ;)
> rel=tag is handled differently from other microformats that use 
> rel=tag (in hAudio the node contents are used, in others the URL is 
> used); 
@rel tag is the url value even in haudio, @category describes the key words.
> item opacity and propagation of values from parent to child item; 
this is a new model (inheritance design pattern) used in only haudio at 
the moment so its experimental I'm not sure it works though?. I think we 
maybe should have used something new like @class="part" or something 
because currently Item in haudio is inconsistent with the way Item is 
used in existing microformats...
> etc. Adding all these rules to the RDFa specification would massively 
> bloat it.
Not so much I don't think...
>
> Parsing existing microformats as part of the RDFa parsing model would 
> be massively complicated. And despite the fact that class names are 
> often re-used by new microformats, this would not help parsing future 
> microformats unless the mfo/hroot question were to be resolved. 
do you not think @class=item in haudio has gone part way to solving the 
Microformat Object debate.
> Such an endeavour would not only be bad for RDFa, but would seriously 
> constrain the future direction of microformats too.
>
>> All existing Microformated pages,
>> Millions of them,  could then potentially become part of RDFa in that
>> way almost instantly and So Supporting Microformats in the best way and
>> not *breaking* them.
>
> RDFa is "just" a representation of RDF. 
You know that's what I always thought, but I have been made to believe 
RDFa is a General Purpose Syntax used to describe semantics in XHTML, 
not limited to just RDF,

IF RDFa is just about RDF then I will leave you all here and never bring 
up this topic again because it is my view Namespaces/prefixes/CURIEs are 
not that well supported in modern browsers, not even well enough  in the 
W3C's own technologies add that to the fact that anyone can create a RDF 
vocabulary without using any kind of process encouraging website 
developers to build walled gardens around themselves in their own 
namespaces and Vocabularies... UGH! anti-social to say the least.
> And microformats can already be parsed as RDF - that's the point of 
> GRDDL. Despite the fact that XSLT is a horrible, horrible abomination, 
> I think that GRDDL, not RDFa, is probably the best hope for bringing 
> microformats into the "upper case Semantic Web". I believe 
> Microformats and RDFa can happily co-exist. They both have different 
> syntaxes, but once you've converted them both to the abstract RDF 
> model, you can use pretty simple rules to combine the data from each. 
> The aim to strive towards should be: different syntaxes, separate 
> parsing models, but at the end one data model.
>
 Best wishes

Martin McEvoy

Received on Friday, 12 September 2008 11:22:58 UTC