Re: [All] Proposal: RDF Graph Identification (mutability & identification)

On 17/08/12 13:28, Sandro Hawke wrote:
> On 08/17/2012 07:37 AM, Andy Seaborne wrote:
>> RDF could just not mention mutability.  On balance, some text in
>> "concepts" to give an overview is useful; it does not need to be
>> continued.  RDF specs define what is, not operations. Implementers
>> seem to have got on just fine with RDF-2004, and that program language
>> "sets" are (with care) mutable does not cause problems.
> But we don't yet have interoperability in graph metadata/management.
> That's what IMHO we're trying to achieve here - the use cases I've tried
> to write up all seem to involve that. How can we reliably convey
> provenance metadata when there's no clarity about when/if different
> "graphs" will be changing?   Does metadata about one "graph" also apply
> to another "graph" that has the same triples?  (The answer depends on
> whether the metadata is really about a g-box or a g-snap -- I think
> people can get clarity on it when they consider whether the metadata
> would still hold if the "graph" changed.)

All this "graph" is gettign in the way :-)

RDF could just cover graphs (g-snap, values), and datasets (ds-snaps, 
values).  Other groups build on that e.g. PROV.

Adding in containers (g-boxes, ds-boxes =?= graph stores) adds useful 
language but RDF is really acknowledging it is grounded in the web.
RDF just defines the meaning of an observed state + acknowledging state 
can change (AWWW).

However, "mutability" can cover a lot more like saying what operations 
can be performed, how they relate, partial changes, continuous changes.

By the way:
Examples of an RDF space include but are not limited to the following:
+ the default graph or any of the named graphs available via a SPARQL 

The default graph is a value.  g-snap.

What can change is the endpoint to dataset relationship.  i.e. the 
"value" changes.

I think that fits better with

The state of an RDF Space at any time is an RDF Graph.

which ('space' aside) I'm comfortable with.  The "at any time" is key; 
you can't observe the change happening - only that change has happened.

>         -- Sandro

Received on Friday, 17 August 2012 16:32:40 UTC