- From: Richard Cyganiak <richard@cyganiak.de>
- Date: Wed, 23 May 2012 12:54:27 +0100
- To: Pat Hayes <phayes@ihmc.us>
- Cc: Andy Seaborne <andy.seaborne@epimorphics.com>, RDF-WG <public-rdf-wg@w3.org>
On 23 May 2012, at 12:32, Pat Hayes wrote: >> I'm reading the resolution as that - given datasets, we can address the use cases; we do not *need* nested graphs. > > ... OK, then I want to be convinced. How exactly do datasets provide the functionality that nested graphs were needed for? Either give a general construction, or a set of pithy examples. No use case that actually calls for nested graphs has been presented. The closest is this one from Sandro, “Loading Untrusted Datasets”: http://dvcs.w3.org/hg/rdf/raw-file/default/rdf-spaces/index.html#uc-untrusted What's needed for this use case is the ability to merge datasets in a way that allows us to track that some IRI-graph-pairs (state pairs) came from the one dataset, and others from the other dataset. This doesn't have to be done by the semantics, but can be done by application code that renames graphs if necessary and records additional triples in the default graph. In other words, there is no need that the merged dataset entails the original datasets. (Each graph taken individually should better still mean whatever it meant in the original dataset, but this is trivially true in Peter's, Antoine's and Sandro's proposed semantics.) Best, Richard
Received on Wednesday, 23 May 2012 11:54:58 UTC