Re: Labelled graphs

On 2012-04-13, at 11:22, Andy Seaborne wrote:

> What I saw in the concept of "labelled graphs" was that it was a loose association of label and graph in a dataset.  Minimal, and something to build on top of.
> 
> There had, prior to "labelled graphs", been a conversation about what name graph semantics could be and we (I?) concluded there was going to be no single choice we could agree on.

+1

> So no fixed useful semantics as the base case and build on that with additional information to strengthen the meaning of the labelling in the dataset.  Juts make sure RDF-WG devised vocabularies don't conflict.
> 
> 
> <u> { ... }
> 
> with no additional statements is a very weak relationship - either nothing, no implied RDF triples (in some extended RDF with graph literals), just the access to the graph in the dataset or a relationship that has null semantics.
> 
> <u> rdf:label { ... }
> 
> Read that as rdf:aaa to avoid natural language meaning.

I don't get that, but… 

> Semantics are "caveat emptor".  It's up the client to decide what can and can't be done.
> 
> 
> We had some discussion about the additional vocabulary - I have no opinion on class/property choice for that.
> 
> <u> { ... } .
> <u> a rdf:staticGraphContainer .
> 
> means that {...} is the value of (g-snap) of URL <u>.
> 
> <u> { ... } .
> <u> a rdf:Graph .
> 
> means that <u> denotes the graph {...}.
> 
> Ideally, I hope TriG files will have the addition information.  In the base case, which cover current practice, is that they don't and that all
> 
> <u> { ... }
> 
> says is that there is an association in this dataset.  Then build on top of that.

Yes.

- Steve

-- 
Steve Harris, CTO
Garlik, a part of Experian
1-3 Halford Road, Richmond, TW10 6AW, UK
+44 20 8439 8203  http://www.garlik.com/
Registered in England and Wales 653331 VAT # 887 1335 93
Registered office: Landmark House, Experian Way, Nottingham, Notts, NG80 1ZZ

Received on Friday, 13 April 2012 10:33:59 UTC