- From: Andy Seaborne <andy@seaborne.org>
- Date: Wed, 12 Dec 2018 11:05:01 +0000
- To: semantic-web@w3.org
On 10/12/2018 17:04, Franconi Enrico wrote: > Hi, > I believe that, in order to properly/correctly solve this issue, SPARQL > implementations could provide a mean to get for each bnode in a SPARQL > query answer the data graph node /*bound*/ to it, e.g., for each SPARQL > query result there could be a table with bindings: <result-bnode, > data-graph-node>. > > IMHO it would be wrong (and not necessary, in this case) to give > persistent identifiers to bnodes. > cheers > --e. The rules for system identifiers would need to limits on their persistence (e.g. may change, non-reuse) but instead of dealing with the cases one-by-one, have one general mechanism and it should be applicable to network use which means some kind of serialization. Extend "IRI(bnode)" to be skolemization, for example. I'd prefer to introduce a separate space for bnode system ids, but skolemization/distinguished IRIs do seem to cover many uses and are RDF 1.1. Andy > > >> On 10 Dec 2018, at 17:30, David Booth <david@dbooth.org >> <mailto:david@dbooth.org>> wrote: >> >> On 12/10/18 5:34 AM, Michael Brunnbauer wrote: >>> I'd really like to see mandatory persistent blank node identifiers in >>> SPARQL - so I can easily do followup queries for blank nodes from a >>> SPARQL >>> result. Is this a big deal? I think some triples stores can already >>> do this. >>> Which issue do I upvote to get this? >> >> Blank nodes #19: >> https://github.com/w3c/EasierRDF/issues/19 >> >> David Booth >> >
Received on Wednesday, 12 December 2018 11:05:31 UTC