Tom Heath wrote:
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Kingsley Idehen
Sent: 12 July 2008 21:43
To: [EMAIL PROTECTED]
Cc: public-lod@w3.org; [EMAIL PROTECTED]
Subject: Re: Ordnance Survey data as Linked Data (RE: How do
you deprecate URIs? Re: OWL-DL and linked data)
Kingsley Idehen wrote:
I also forgot to mention obvous use of RDFa in the HTML doc
which broadens the range of rdf aware user agents tha
commence RDF discovery from HTML
Question: is it worth creating a duplicate RDF graph by using RDFa in
HTML documents, when there is also RDF/XML available just one <link
rel=".../> away, and at a distinct URI? Doesn't this RDFa + RDF/XML
pattern complicate the RDF-consumption picture in general if we assume
agents will want to do something with data aggregated from a number of
sources/locations, i.e. doesn't it increase the cost of removing
duplicate statements by creating more in the first place? Does it not
also complicate the picture of making provenance statements using named
graphs, if the subject of the triple could be both an HTML document and
an RDF graph?
Dunno the answers to these questions, but interested to hear what people
think.
Tom.
Tom,
I believe we should spread the net as wide as possible re. RDF aware
user agents since we cannot assume which methods they will be using to
discover RDF. On our part (re. our RDF aware user agents), we use all
the methods (Content Negotiation, <link />, GRDDL, RDFa, and POSH) when
sniffing, which also implies that we
take on the burden of normalization.
On the part of publishers, I encourage the use of at least one of the
RDF exposure methods mentioned above, when known associations between
HTML and RDF representations exist.
--
Regards,
Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen
President & CEO
OpenLink Software Web: http://www.openlinksw.com
