- From: <noah_mendelsohn@us.ibm.com>
- Date: Thu, 14 Jun 2007 22:52:36 -0400
- To: "Booth, David (HP Software - Boston)" <dbooth@hp.com>
- Cc: "Rhys Lewis" <rhys@volantis.com>, www-tag@w3.org
David Booth wrote: > I have a problem with every spec thinking that it is important > enough that all processors that might encounter it should have > built-in knowledge of how it should be processed. Even if > an XML language has specifically designed its own mechanisms > for relating its constructs to triples, I would much prefer > that the spec writers be encouraged to express those mechanisms > using GRDDL. Exactly. While designing some features of particular media types so that they formally map to known triples, or designing the whole language in terms of triples may both be good things to do, the corresponding RDF can only be extracted if you know the particular media type. Consider a document in a mythical media type application/noahsformat+xml. The spec. for application/noahsformat+xml might indeed indicate how some particular attribute mapped to RDF, but general purpose software is unlikely to know that. If I use GRDDL to make the connection to RDF, and if we assume that GRDDL is widely deployed, then the chances that the RDF will be extractable go way up. Simillarly, with the slight glitch (mentioned at the F2F I think) that the self-describing story for RDFa really only works if the media type registrations for HTML are updated to account for RDFa, that too is likely to be a widely-deployed convention for integrating RDF into Web content. Noah -------------------------------------- Noah Mendelsohn IBM Corporation One Rogers Street Cambridge, MA 02142 1-617-693-4036 --------------------------------------
Received on Friday, 15 June 2007 02:52:50 UTC