- From: Andy Seaborne <andy.seaborne@epimorphics.com>
- Date: Wed, 18 Jan 2012 15:16:35 +0000
- To: public-rdf-wg@w3.org
On 18/01/12 14:05, Richard Cyganiak wrote: > On 18 Jan 2012, at 10:58, Andy Seaborne wrote: >> Do any use cases *require* graph naming, or can they, maybe less >> conveniently, always be solved with identifying a notional >> representation (document, graph serialization) obtained from a >> place (graph container/g-box) at a point time (i.e. an >> observation). > > There are use cases such as “keep inferred triples separate from > asserted triples in an RDF store” that really seem to require graph > naming. Thanks - that is quite a good UC for naming graph values. > The use cases seem to fall into two groups: > > 1. RDF datasets as a cache/archive of collections of graphs obtained > from other sources > > 2. RDF datasets as a way of managing subsets of a larger RDF graph > > Maybe it is true that everything in the first group can be handled by > referring to documents or serializations. But I don't think it's > possible for the second group. Useful characterisation. At the risk of overemphasising "managing", the use of the naming is more towards local-application usage, not exclusively though. A certain amount of just "tagging" or "coloring" the graph (the tag may well be a proper name, owl:sameValueAs, the graph - it's the degree of importance of this I'm exploring) [1] Andy > > Best, Richard [1] I'm noting http://www.w3.org/TR/owl-ref/#sameAs-def says "owl:sameAs links an individual to an individual"
Received on Wednesday, 18 January 2012 15:17:13 UTC