- From: Danny Ayers <danny.ayers@gmail.com>
- Date: Thu, 28 Jul 2011 19:21:44 +0200
- To: public-linked-json@w3.org
- Cc: public-lod@w3.org
[cc'ing public-lod@w3.org, this all seems to be drifting a little beyond JSON scope - see [1], [2], [3] ] "LD" meaning "Labeled and Directed" for JSON-LD works for me too. But I don't see a problem with defining linked data as being all-URIs (fully grounded, no bnodes or literals) just for spec purposes, it does at least emphasize the key feature (although I'm still a fan of bnodes :) Is a graph solely comprised of bnodes linked data? Presumably not. Is the result of merging an all-URI graph with an all-bnode graph linked data? In general parlance and practice yes, but it doesn't actually contain any more information than the first subgraph. So what happens with a graph which contains something like: <#uriA> :p1 _:x . _:x :p2 <#uriB> . ? It's tricky, the individual triples don't entirely fit with the 4 principles, together they kind-of do. But I certainly don't think we need to leap to skolemization to make sense of this. If the graph's on the Web as it should be, then it'll be named with a URI, so we could get a "quasi-entailment" along the lines of: <#graph> :contains <#uriA> . <#graph> :contains <#uriB> . or if you prefer to stay within the graph, something like: <#uriA> :p1 _:x . _:x :p2 <#uriB> . => <#uriA> rdfs:seeAlso <#uriB> . Dunno, this might all just be angels on a pinhead stuff... Cheers, Danny. [1] http://json-ld.org/spec/latest/ [2] http://lists.w3.org/Archives/Public/public-linked-json [3] https://plus.google.com/102122664946994504971/posts/15eHTC3FA4A -- http://danny.ayers.name
Received on Thursday, 28 July 2011 17:22:22 UTC