RE: yet another strawman, was Re: generic XML to RDF triple mappi ng

On Fri, 8 Sep 2000, James Tauber wrote:

> > I generally agree, and would put forward that the 'mapping 
> > language' is an RDF Schema itself. 
> I'm not sure I follow. An RDF Schema places contraints on triples. I am
> talking about something that maps XML information items to triples.
> Here's an example:
> One of the serializations of 
> 	foo , bar , baz
> is
> <rdf:Description about="foo">
>   <bar rdf:resource="baz"/>
> </rdf:Description>
> So the mapping rule is (roughly)
> when you hit a
> 	rdf:Description[@about]/*[@rdf:resource]
> 	then add the triple
> 	../@about , name() , @rdf:resource
> This looks a lot more like XSLT than anything else to me.

Very like! Did you see Dan Connolly's RDF parser demo in XSLT? I'd love to
see this finished off...
	   Transforming RDF with XSLT, Dan Connolly	
		    -- an RDF parser in XSLTfor now, just the basic RDF syntax per2.2.1.

random excerpt:

	<xsl:when test='@rdf:about'>
          <xsl:call-template name='rdfp:propertyElt_s'>
            <xsl:with-param name='node' select='$node'/>
            <xsl:with-param name='subject' select='@rdf:about'/>
            <xsl:with-param name='parentSubject' select='$parentSubject'/>
            <xsl:with-param name='parentPredicate' select='$parentPredicate'/>

> I believe that most of the syntax of the RDF M&S could be expressed in
> similar fashion and FURTHERMORE that all the XML documents I have that are
> RDF-like but not quite (because they use ref instead of rdf:resource, for
> example) could be mapped to triples just as easily.

So... what's a good way of characterising the relationship between XSLT
resources (eg. Dan's parser) and various types of XML document? Given some
arbitrary XML, it'd be handy to be able to do a query to find appropriate
XSLTs that would perform the transformation to RDF. I've not come across
much work on metadata for XSLTs, though maybe I've been looking in the
wrong places. Are any conventions emerging?


Received on Friday, 8 September 2000 15:56:28 UTC