On 2/1/09 3:36 PM, Sergio Fernández wrote:
After a short discussion with Michael by IRC about the issue he pointed
with my usage of void:target property, we got some conclusions.
Basically while voID is not extended, many dataset don't have a proper
URI. So there are two ways to refer them: URI vs Blank Node, a IFP
property (in this case foaf:homepage) does the magic in both cases. I
know that many people hate blank nodes, and I would like not to need to
use here, but in this case I prefer to use it and wait until the dataset
publish a proper URI.
IMO this is not just interesting for voID, but the scope of this
discussion is bigger for the whole Semantic Web in general: do we want
to create this (artificial) URIs? What's your opinion?
Cheers,
[1] http://semanticweb.org/wiki/VoiD#Using_voiD
[2] http://rdfohloh.wikier.org/about
Sergio,
Nothing wrong with IFPs, at the end of the day they are going to be the
key to bootstrapping the "Linked Data Web".
If you look across Twitter , Identi.ca, and all the other Web 2.0 data
spaces, you will notice that IFPs reign at User App. Level. Even better,
if we are trying to evolve the Web into a DBMS we have to be consistent
with how DBMS engines actually work: all DBMS engines have IFPs (Primary
Keys) and URIs (RowIDs). What makes the Web different is the fact that
the RowIDs (URIs) aren't locked at the application, operating system, or
network level :-)
VoiD is bringing important matters to the fore (e.g. describing data
containers and their relationships) that ultimately enable
juxtaposition of traditional DBMS and "Web DBMS" as the basis for
really understanding what the "Linked Data Web" is really about.
--
Regards,
Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen
President& CEO
OpenLink Software Web: http://www.openlinksw.com
