Re: [Graphs] Proposal for Named Graph Semantics

Richard Cyganiak wrote:
> On 8 Apr 2011, at 16:32, Antoine Zimmermann wrote:
>> This is what Pat called "poking a g-box". Poking a g-box is sometimes HTTP dereferencing the IRI that identifies the g-box. Sometimes, it's just returning what's in the curly brackets in a TriG document. Sometimes it's whatever triples attached to a certain IRI in a Quad file. Sometimes, it's getting the graph represented by a Jena Model in memory.
> 
> +1.
> 
> That's why I'm in favour of defining *only* an abstract syntax that pairs IRI and g-snap, without constraining what the relationship is.
> 
> If I give you a TriG document, or tell you about my SPARQL store, I would probably say: “Look, I poked a bunch of g-boxes (via their IRIs), and here's the g-snaps I got from each.”

Agree, operative word there being "I", in different use cases the poking 
produces different results for different people, often at the same 
instant (auth* etc).

Nice to keep it as loosely defined as possible, it won't impede any 
functionality from any use case or spec, now or in the future, that way.

Best,

Nathan

Received on Friday, 8 April 2011 22:30:51 UTC