- From: Bruce D'Arcus <bdarcus@gmail.com>
- Date: Mon, 03 Sep 2007 12:22:07 -0400
- To: "Hammond, Tony" <t.hammond@nature.com>
- CC: Reto Bachmann-Gmür <rbg@talis.com>, semantic-web@w3.org
Tony, Hammond, Tony wrote: ... > The problem with XMP is that it does not allow for arbitrary RDF/XML but > constrains the defined schemas (DC being one of the defined schemas) to > contain the value types specified. See the dump below of a demo XMP using > the sample SDK app DumpMainXMP (and as RDF/N£ via cwm). Note that the > structured types (array - Alt, Bag, Seq) follow the Adobe spec. We've chatted about this a bit off-list earlier. Indeed this is an unfortunate situation. We have XMP, which in principle is a great implementation of RDF: the notion that one embeds metadata in the file and it travels with it. OTOH, XMP is a really weird subset of RDF circa 2000. And as the RDF worlds evolves, XMP is increasingly out of sync. Recent changes at the DCMI are perfect evidence of this, where they are deprecating the suggestion to use rdf:Bag and rdf:Alt, defining new base terms with explicit domains and ranges that conflict with how XMP does things, etc. I don't have any solutions for you, though I do wish someone at the W3C would put this issue on the radar screen. OpenDocument is getting full-blown RDF support, and I've seen barely a ripple in the semantic web community. What about a W3C-led effort to standardize RDF embedding in files, with Adobe invited to the table? Bruce
Received on Monday, 3 September 2007 16:21:36 UTC