- From: Markus Lanthaler <markus.lanthaler@gmx.net>
- Date: Mon, 21 Jul 2014 20:49:55 +0200
- To: <public-linked-json@w3.org>
On 20 Jul 2014 at 16:19, james anderson wrote: > On 20 Jul 2014, at 16:02, Markus Lanthaler wrote: >> I assume you meant this instead, otherwise the JSON snippet below doesn't >> make much sense as the two blank nodes aren't connected to each other (at >> least not directly): > > correct. my error. > > i had intended something more like > > _:thing1 <http://example.com/id> <http://example.com/thing#1> . > _:thing1 <http://example.com/type> <http://example.com/ImportantThing> . > _:thing1 <http://example.com/feature> _:thing2 . > _:thing2 <http://example.com/id> <http://example.com/thing#2> . > _:thing2 <http://example.com/type> <http://example.com/Feature> . > _:thing2 <http://example.com/characteristic> "low power consumption" . It's basically the same. >>> [ { "key" : <http://example.com/thing#1>, >>> "class": <http://example.com/ImportantThing>, >>> "features" : [ { "key" : <http://example.com/thing#2>, "class": >>> <http://example.com/Feature>, "detail": "low power consumption" } ] } ] >> >> You can use the following frame >> >> { >> "@context": { >> "key": "@id", >> "class": { "@id": "http://example.com/type", "@type": "@id" }, >> "features": "http://example.com/feature", >> "detail": "http://example.com/characteristic" >> }, >> "features": {} >> } >> http://tinyurl.com/maezvet > > your playground example works on the prepared json, but it is not clear how > it applies to the given graph. It works on any JSON-LD document that serializes that graph. The processing algorithm is specified in terms of JSON-LD and not an abstract triple representation if that's your question. > that is, the "json-ld input" requires an initial transformation from the original > rdf graph which i do not yet follow as a consequence of the frame declaration. No, not really. The input document is basically flattened (it creates a so called node map internally) so that it has a deterministic shape. Then that JSON structure is simply filtered etc. and subtrees are reconstructed. > the frame also appears to rely on not having specified a type. I don't understand this question. The doesn't rely on the presence/absence of a type; only on the presence of the "features" property to find the desired top-level node. -- Markus Lanthaler @markuslanthaler
Received on Monday, 21 July 2014 18:50:26 UTC