W3C home > Mailing lists > Public > public-lod@w3.org > July 2008

Re: RDFa + RDF/XML Considered Harmful? (was RE: Ordnance Survey data as Linked Data)

From: Bijan Parsia <bparsia@cs.man.ac.uk>
Date: Tue, 15 Jul 2008 14:44:55 +0100
Message-Id: <FD80E3E2-5472-41EC-9396-B3C90A7A5C9E@cs.man.ac.uk>
Cc: "Taylor Cowan" <taylor_cowan@yahoo.com>, public-lod@w3.org, "SW-forum Web" <semantic-web@w3.org>
To: "Mark Birbeck" <mark.birbeck@webbackplane.com>

On 15 Jul 2008, at 14:10, Mark Birbeck wrote:

> Hi Bijan,
>
>>> An important difference between embedded RDF/XML and RDFa is that  
>>> RDFa and
>>> the xhtml can use the same "literals".
>>
>> ? That seems true for RDF/XML as well. Hence parseType=Literal.  
>> You can also
>> hide literals, hence property attributes.
>
> I think what Taylor is describing, is something like this:
>
>   <span property="foaf:name">Taylor Cowan</span>
>
> which reuses the data.
>
> This is very difficult to do using embedded RDF/XML.

But not because it wasn't designed to do this.

> In RDF/XML, the two ways to express a predicate are via an element:
>
>   <foaf:name>Taylor Cowan</foaf:name>
>
> or an attribute:
>
>   <x foaf:name="Taylor Cowan" />

Yes.

> Since the first way would make the XHTML invalid (adding arbitrary
> elements is very difficult to do),

This is the part which missed the point. First off, it will be well  
formed and whatever extension you do to make it composable will work.

Obviously, today, we know that such extensibility is a barrier. But  
if we think of browser behavior esp at the time (i.e., ignore unknown  
elements and attributes, pass element content) you can see that RDF/ 
XML is exactly designed to do what Taylor says it can't do.

And, the second will be invalid too, so I have no idea what your  
point is.

> the second is preferred, since
> HTML/XHTML processors should ignore attributes that they don't
> recognise.

That's not the design or intent. Obviously, if you add additional  
constraints there's a problem with the execution, but to say that it  
can't do it or they can't use the same literals (yes, valid xhtml, if  
it doesn't have an schema with an open content model or RDF/XML  
permitted can't share the same literal, but uh, it can't embed the  
hidden things *either* and be valid).

> However, that would require a duplication of data:
>
>   <span foaf:name="Taylor Cowan">Taylor Cowan</span>

And how is foaf:name part of valid XHTML?

> RDFa is therefore more efficient than RDF/XML embedded in HTML.

Maybe, but not for the reasons you enumerate, afaict.

> (Hardly surprising, since as I point out in my other email, my first
> version of RDFa was exactly to embed RDF/XML.)
>
>
>>>  In other words, the text viewed by the human, and the text  
>>> stored as the
>>> literal object of a triple is the same.
>>
>> An option in RDF/XML.
>
> I don't follow what you mean. Do you mean that people can viewing the
> RDF/XML? Or do you mean that XHTML+RDF/XML is viewable? As I pointed
> out above, the problem is that this is not valid:
>
>   <span>
>     Welcome to the blog of
>     <foaf:name>Taylor Cowan</foaf:name>
>   </span>

If you presume that HTML browsers (or XHTML browsers) reject invalid  
HTML,  then you might have a point ;)

Nothing make RDF/XML *not* designed to accommodate in exactly the way  
RDFa does literal content with fall back behavior. RDF/XML embedding  
requires other infrastructural moves that didn't pan out, which is,  
in part, why it's not all that viable. But it's silly to deny its  
manifest design.

Cheers,
Bijan.
Cheers,
Bijan.
Received on Tuesday, 15 July 2008 13:42:47 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:20:40 UTC