Damian Steer wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 19 Sep 2008, at 13:39, [EMAIL PROTECTED] wrote:
On 19 Sep 2008, at 13:07, [EMAIL PROTECTED] wrote:
but by that token you could probably wipe out most of foaf and doap
space from the diagram
Most of that data is not very linky and many primary resources being
described don't have uris
Sure, I am guessing you are talking about foaf:knowing b-nodes, but
the use of a rdfs:seeAlso with a IFP, should satisfy the notion of
being "linky", or am I wrong in thinking this?
I hope you're right, otherwise I'm not sure what 'linking' would be :-)
Well for sure, FOAF was designed as a linked information system, with
two equally important styles of linking:
1. Reference by URI (ideally well known; ideally de-referencable; and
nice if derferencable to a description by everyone through common
protocols (http/303, ftp etc) into common formats (rdfa, rdf/xml,
xml/html+grddl, rif, owl2, ...)
2. Reference by Descriptions (ideally by IFP and FP; also perhaps by
multi-property keys, and also heuristic matching is important).
We can't build the Semantic Web without both of these pieces. Anyone who
thinks that linking is just (1.) might consider picking another word (I
suggest 'hyperlinking').
Dereferencing is a privilege and not a right. There's nothing in FOAF
against reference-by-URI, just an acknowledgement that often things
don't have well known URIs, and that the business of asserting
identifiers for people is heavily politicised.
cheers,
Dan
--
http://danbri.org/
