Re: meta-information about assertions

On Mon Mar 03, 2008 at 08:13:16AM -0500, Frank Manola wrote:
> Have you looked at and the work 
> on named graphs (and provenance in general)?

of course (speaking for the original poster too)! he is asking, the 'proper' way.

theres many suggestions. like singleton named graphs (one per statement, or one per asserter), in-graph reification, and what not

i have questions how this relates to HTTP and URIs , particularly. it doesnt all quite fit together clearly, imo.

lets take the first solution, singleton named graphs..

the examples end up using the graph URI in the subject or object position of a statement... ok. but what if you GET that graph URI? normal servers just return a CBD or all the doc resources sharing a common trailing #hash. not an entire graph... 

you have a sort of confliction between 'resource' and 'graph' which is 'multiple resources' and how do you differentiate when fetching?

afaik there isnt even a 'Graph' header modifier when GETing an RDF resource.. how do you even specify?

on the second solution, 'reification'. 
you see stuff like <a> <statement:says> [<x> <y> <z>]

how do you serialize this? you need a URI for the inner statement object, afaik. wheres the spec on this?

Received on Monday, 3 March 2008 14:55:04 UTC