Re: RDF/XML and named graphs

On 18 Dec 2007, at 16:06, Noah Slater wrote:

> On Tue, Dec 18, 2007 at 08:01:37AM -0800, Chris Richard wrote:
>> But you are in control of the RDF/XML serialisation and do it in a
>> consistent way, right? You can't grab any old RDF/XML and run this
>> XSLT on it even if it the data uses the same RDF schema.
>
> Well yes, but this is a non-sequitur.

Actually it isn't. The point is that RDF/XML, by itself, isn't  
particularly XSLTable because of the syntactic freedom it affords.  
You basically need an RDF Parser in XSLT (which is doable). This  
feels like one is fighting the serialization.

> There are plenty of stylesheets
> which could be applied in a general case to many RDF/XML documents.

Your "general case" only applies to "many" RDF/XML documents? Doesn't  
that simply make Chris's point for him?

> Case in point:
>
>   http://silkpage.markupware.com/about/index.html.rdf.xml

This is a good example. For Chris :)

Change your serialization slightly, i.e., add a namespace declaration:
	xmlns:tax="http://www.markupware.com/metadata/taxonomy#"

and change your rdf:Description to a typed node:
	<tax:SilkaPage rdf:about="http://silkpage.markupware.com/about/ 
index.xml">
...
	</tax:SilkaPage>

(Delete the corresponding rdf:type if you like.)

Same graph, different renderings.

This doesn't mean that XML tools are worthless on RDF/XML, just that  
it's hard to hold the XML tool chain up as a *strength* of RDF/XML,  
given how difficult it is to work with.

Of course if you normalize your RDF/XML documents...things are  
easier. But that's not because of RDF/XML per se (or perhaps, alone),  
but because of your normalization. Typically you need an RDF/XML  
parser in the loop to do that normalization (in a general way).

Cheers,
Bijan.

Received on Tuesday, 18 December 2007 16:28:29 UTC