Re: ACTION-97: comments on link relations

I actually think that most of the new HTML5 rel values are nonsense, and 
I am happy accepting ONLY the values in our existing vocabulary 
document.  If people think there are a couple in HTML5 we should add, I 
guess we can do that.  The valuespace for link types is unbounded.  
Anyone can put anything in there that they want.  What we choose to 
generate triples for is up to us.

I would be open to dropping support for alternate, but it has semantic 
meaning when NOT used with stylesheet and I think it might be bad to 
lose that.  I am NOT open to special-casing 'alternate stylesheet'.  I 
know that HTML special cases it, but I don't care.  We had this debate 
years ago and I don't think anything in the environment has changed.  
Steven, what do you think?

On 10/13/2011 5:42 AM, Ivan Herman wrote:
> On Thu, October 13, 2011 10:28 am, Toby Inkster wrote:
>> On Thu, 13 Oct 2011 07:26:36 +0100
>> Ivan Herman<ivan@w3.org>  wrote:
>>
>>> minus those whose specification does not fit the 'semantic' aspect of
>>> RDFa (e.g., nofollow)
>> Why isn't nofollow semantic? OK, it's badly named, but properties in
>> RDF are URIs, and URIs are opaque.
> Maybe nofollow is a wrong example. But are we sure that all of the HTML5 rel values are semantically kosher for us? If
> yes, that is fine, but that is something that we have to decide.
>
> (I must admit that I always found the fact that we automatically use stylesheet, for example, irritating. An RDFa web
> page usually speak about a more general concept and not the HTML source itself, and stylesheet is on the source. Most
> of the case this means that that triple is noise for me...)
>
>> Let's imagine an agent which takes the RDFa of a page, and also adds
>> some extra triples along the lines of:
>>
>> 	<page>  sioc:links_to<dest>  .
>>
>> For every<a href>  and<link href>  link on the page.
>>
>> It's reasonable to run the following SPARQL on that data:
>>
>> 	CONSTRUCT {
>> 		?author ex:endorses ?page .
>> 	}
>> 	WHERE {
>> 		?source dc:creator ?author .
>> 		?source sioc:links_to ?page .
>> 		FILTER NOT EXISTS { ?source xhv:nofollow ?page . }
>> 	}
>>
>> Personally for HTML5+RDFa I just accept the union of HTML5's rel
>> values, and RDFa's rel values.
> Do you mean the XHTML rel values? I do not really really mind, but the fact is that, for good or bad reasons, HTML5
> threw out most of the XHTML defined @rel values. I wonder whether it is a good policy for us to keep them
> nevertheless.
>
> Ivan
>
>
>
>> --
>> Toby A Inkster
>> <mailto:mail@tobyinkster.co.uk>
>> <http://tobyinkster.co.uk>
>>
>>
>

-- 
Shane McCarron
Managing Director, Applied Testing and Technology, Inc.
+1 763 786 8160 x120

Received on Thursday, 13 October 2011 11:09:08 UTC