On 1/4/12 10:42 AM, Yrjana Rankka wrote:
On 1/4/12 17:00 , Kingsley Idehen wrote:
Trouble is that there isn't consensus re. this matter.
For instance, one could assume that the URI / IRI of a named graph
resolves to a description of said graph. That wouldn't really imply
all the triples in the named graph :-)
Let's consider this situation:
GRAPH <x>
<s1> <p1> "val1";
<p2> "val2";
<p3> "val3".
GRAPH <metadata>
<x> dc:created "somedate";
dc:modified "someotherdate";
dc:creator "Zaphod Beeblebrox".
A client dereferences <x>
What would you expect to get?
Cheers,
SPARQL's use of GRAPH is the subject of intense debate as per Ivan's
comment. At the current time, its just a *label* with SPARQL engine
specific handling.
The example above demonstrates why Name Graph IRIs need clarity. Right
now -- as you know re. Virtuoso -- they serve as Identifiers for
dataset partitioning, by default. If one really seeks a de-referencable
URI/IRI for a named graph, then one should make triple based statements
describing said named graph using its URI/IRI. Then you have a route to
a Name that resolves to a description.
--
Regards,
Kingsley Idehen
Founder& CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen
