- From: Dan Brickley <danbri@w3.org>
- Date: Wed, 4 Aug 2004 05:53:24 -0400
- To: Mark Birbeck <mark.birbeck@x-port.net>
- Cc: public-rdf-in-xhtml-tf@w3.org, 'Masayasu Ishikawa' <mimasa@w3.org>
Thanks for the fix. I was talking with Max Froumentin about possibility of XSLTing this notation into standard RDF/XML. It would seem that every element encodes a property and that the XSLT would spend a lot of its time inserting 'padding' elements that correspond to nodes in the graph (untyped and hence written as rdf:Description of rdfs:Resource according to taste). An alternative idiom would be liberal use of parseType="Resource" attributes where we have a property pointing to a more detailed description of something. People seem to have different instincts about which style of RDF is the ugliest ;) If I understand correctly, the basic transform we're after is: [[ <link rel="foaf:maker"> <meta property="foaf:name">Dan Brickley</meta> <link property="foaf:homepage" resource="http://danbri.org/"/> <link rel="foaf:knows"> <meta property="foaf:name">Dan Connolly</meta> </link> </link> ...into: <rdfs:Resource> <foaf:maker> <rdfs:Resource> <foaf:name>Dan Brickley</foaf:name> <foaf:homepage rdf:resource="http://danbri.org/"/> <foaf:knows> <rdfs:Resource> <foaf:name>Dan Connolly</foaf:name> </rdfs:Resource> </foaf:knows> </rdfs:Resource> </foaf:maker> </rdfs:Resource> ]] A contracted form of the latter is: <rdf:Description> <foaf:maker rdf:parseType="Resource"> <foaf:name>Dan Brickley</foaf:name> <foaf:homepage rdf:resource="http://danbri.org/"/> <foaf:knows rdf:parseType="Resource"> <foaf:name>Dan Connolly</foaf:name> </foaf:knows> </foaf:maker> </rdf:Description> ...perhaps this would be a natural target for an XSLT conversion? Talking of which, more questions: i) re GRDDL, if a transformation _does_ require XSLT 2.0 (although you argue in [1] that it can be implemented in 1.0), do we have any way in GRDDL of expressing this? ii) are there XSLTs already which consume this new notation and emit RDF/XML? I was encouraging MaxF to have a go at it, but perhaps the work is already done... iii) test cases. Getting a syntax for something like this right can be tricky, with lots of corner cases. Does the HTML WG have infrastructure for hosting test cases, perhaps something using the same approach we used in http://www.w3.org/TR/rdf-testcases/ ie. pairs of files, one in the relevant XML syntax, the other showing the canonical RDF graph serialized using the N-Triples notation... If the HTML WG aren't likely to do this, perhaps this is something the SWBPD WG could assist with in some way? iv) xml:base, relative URIs in ns declarations etc? eg. test case: <link xmlns:foaf="/foaf/0.1/" xml:base="http://xmlns.com/" rel="foaf:maker"> <meta property="foaf:name">Dan Brickley</meta> <link property="foaf:homepage" resource="http://danbri.org/"/> <link rel="foaf:knows"> <meta property="foaf:name">Dan Connolly</meta> </link> </link> ...does this generate the same triples? I think I've also some question about XML literals but need to re-read first. On first glance it wasn't clear that this notation could maintain the distinction RDF has, between XML literals and plain literals that happen to look like markup. cheers, Dan [1] http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2004Mar/0007.html ...re xslt 2, qnames in attributes etc.
Received on Wednesday, 4 August 2004 05:53:41 UTC