- From: landong zuo <landong.zuo@googlemail.com>
- Date: Mon, 5 Dec 2011 15:38:35 +0000
- To: Mark Birbeck <mark.birbeck@webbackplane.com>
- Cc: Toby Inkster <tai@g5n.co.uk>, public-rdfa-wg@w3.org
Hi Mark, I totally agree with you regarding the graph container using "@typeOf=rdf:Graph" and separation between"<head>" and "<body>". This would give me the perfect solution to distinguish metadata triples from document. I am wondering if it is appropriate to have this feature included in the future release of RDFa processing model as "the recommended practice"? I don't know how many RDFa parsers support Named Graph yet. I am sure not every consumer or developer would love to write his/her specific implementation. Regards, Landong On Mon, Dec 5, 2011 at 2:26 PM, Mark Birbeck <mark.birbeck@webbackplane.com> wrote: > Hi Landong, > > On Mon, Dec 5, 2011 at 10:40 AM, landong zuo > <landong.zuo@googlemail.com> wrote:> Hi Toby,>> The working-around > style sheet you pointed out should work well in> this particular case. > Many thanks for your suggestion.>> My use case falls into the class as > you states "converting to RDFa and> injecting to HTML template". We > have to be very careful in this> process not to confuse "HTML > semantics" with "RDF semantics". This is> really painful.>> The case > could become even worse to have attached metadata like RDF> signature > or Provenance. This type of metadata could be generated by> different > organisations, and has to be carried all the way through> data > transformation process without change to the semantics. We have> no > control over RDF triples. Additional triples like xhv:stylesheet> > would definitely ruin the metadata.>> I think Named Graph will be a > promising answer for this type of> questions. What do you think? > IMO any web-page that is parsed for RDF is a named graph, whether it > wants to be or not. :) When implementing an RDFa parser I always used > to store the triples from any HTML page into a named graph identified > by the document's URI, as if the following were present [1]: > > <html> > <head about="" typeof="rdf:Graph"> > <!-- everything in here is about the named graph --> > </head> > . > . > . > </html> > > The fact that a named graph also has a CSS stylesheet is therefore no > big deal (and anyway, it's true). > > If authors then follow best practice and refrain from mixing their > people up with their web-pages, then all other triples should come out > fine: > > <html> > <head about="" typeof="rdf:Graph"> > <!-- everything in here is about the named graph --> > </head> > <body about="#langdong" typeof="foaf:Person"> > <!-- everything in here is about the object represented by the > named graph --> > </body> > </html> > > As you can see, I don't think anything needs to change in RDFa to make > your scenario possible. > > Regards, > > Mark > > [1] I'm pretty sure that Jeremy Carroll proposed rdf:Graph some time > back, but I can't find a reference. I'm sure someone like Ivan will > know where it comes from though. ;)
Received on Monday, 5 December 2011 15:39:12 UTC