- From: Timothy Lebo <lebot@rpi.edu>
- Date: Wed, 13 Feb 2013 10:21:04 -0500
- To: Graham Klyne <GK@ninebynine.org>
- Cc: public-prov-wg@w3.org
On Feb 12, 2013, at 1:28 PM, Graham Klyne <GK@ninebynine.org> wrote: > On 12/02/2013 13:44, Timothy Lebo wrote: >> Luc, >> >> On Jan 15, 2013, at 5:51 PM, Luc Moreau <L.Moreau@ecs.soton.ac.uk> wrote: >> >>> Hi Tim, >>> >>> The prov-o document has several examples with blank nodes. >>> Some of them are difficult >>> to express in prov-n/prov-xml. >>> >>> Consider: >>> https://dvcs.w3.org/hg/prov/file/5495d990f17b/testcases/provo/prov-o-property-hadUsage-PASS.ttl >>> >>> The usage has no identifier we can use in the derivation. >> >> Any identifier will do; you may choose a new one for each bnode you find. >> >> > Not *any* identifier: you have to pick one that almost certainly won;t be used for something else - i.e. a UUID or similar. Sorry, any "well-behaved choice" of an identifier. Don't choose <http://www.w3.org/People/Berners-Lee/card#i> or <http://data.semanticweb.org/person/luc-moreau> since those already refer to something. c.f. "Keep out of namespaces you do not control" in http://linkeddatabook.com/editions/1.0/#htoc35 > > Use or non-use of blank nodes is something of a religious issue in RDF circles. Agreed. But I'm not bringing that to this group, merely trying to communicate how they should be treated when it comes to transforming to other representations. > They can be problematic in some environments, but such environments would be free to allocate UUIDs for themselves. Also, using them can sometimes lead to a proliferation of RDF graph size, but I don't think that would be a problem in the case of examples. > > Personally, I wouldn't worry about using blank nodes in examples - they are a valid aspect of RDF, and I don't see that changing. > -Tim > #g > -- > > > >
Received on Wednesday, 13 February 2013 15:21:32 UTC