Re: json-ld and sparql

Hi Jean-Claude,

SPARQL SELECT results (regardless of in XML or JSON) do not represent RDF,
only tabular results.

To get JSON-LD, you need to use the results of a SPARQL CONSTRUCT query.
Some systems support returning that as JSON-LD, but perhaps only expanded
or using a predetermined context. Some may only return RDF/XML and/or
Turtle. In either case CONSTRUCT results represent RDF and you can turn
that into JSON-LD using any context you like on your end (and any method of
framing).

Best regards,
Niklas

(PS. I once made a thing that turned these tabular results into a tree
(with nesting determined by the variable names). That was fairly close to
the shape of JSON-LD (this was before JSON-LD came to be). I cannot say I
recommend this today, but you might want to take a look at:
https://github.com/niklasl/oort/blob/wiki/SparqlTree.md Along with a
JS-implementation here:
https://github.com/niklasl/oort.js/blob/master/src/oort/sparqltree.js)


On Tue, Mar 15, 2016 at 2:33 PM, Jean-Claude Moissinac <
jean-claude.moissinac@telecom-paristech.fr> wrote:

> I'm trying to find best practices in the use of a json output from a
> sparql endpoint
> For exemple the request on dbpedia :
> prefix schema: <http://schema.org/>
> select distinct ?p ?name ?lat ?lon where {
> ?p a schema:Place .
> ?p rdfs:label ?name .
> ?p geo:lat ?lat .
> ?p geo:long ?lon .
> } LIMIT 5
> give me as json output
>
> { "head": { "link": [], "vars": ["p", "name", "lat", "lon"] },
>   "results": { "distinct": false, "ordered": true, "bindings": [
>     { "p": { "type": "uri", "value": "http://dbpedia.org/resource/Cameron_Parish,_Louisiana" } , "name": { "type": "literal", "xml:lang": "en", "value": "Cameron Parish, Louisiana" } , "lat": { "type": "typed-literal", "datatype": "http://www.w3.org/2001/XMLSchema#float", "value": "29.85" } , "lon": { "type": "typed-literal", "datatype": "http://www.w3.org/2001/XMLSchema#float", "value": "-93.2" }},
>     { "p": { "type": "uri", "value": "http://dbpedia.org/resource/Cameron_Parish,_Louisiana" } , "name": { "type": "literal", "xml:lang": "ar", "value": "\u0645\u0642\u0627\u0637\u0639\u0629 \u0643\u0627\u0645\u064A\u0631\u0648\u0646 (\u0644\u0648\u064A\u0632\u064A\u0627\u0646\u0627)" } , "lat": { "type": "typed-literal", "datatype": "http://www.w3.org/2001/XMLSchema#float", "value": "29.85" } , "lon": { "type": "typed-literal", "datatype": "http://www.w3.org/2001/XMLSchema#float", "value": "-93.2" }},
> ... < cut for clarity >
>
>  }} ] } }
>
> We loose the fact that ?p is a Place. We can factorize the 'type' on the
> table of results. And finally get a table of objects like thos one
> {
>     "p ":  "http://dbpedia.org/resource/Cameron_Parish,_Louisiana",
> "name": "Cameron Parish, Louisiana",
> "pos": { "lat": 29.85, "long": 93.2 }
> }
>
> I suppose some results can be obtained by using a context and some others
> by framing.
>
> Can we point me on ideas for clarification?
>
> Many thanks in advance
>
> --
> Jean-Claude Moissinac
>
>
> Cet e-mail a été envoyé depuis un ordinateur protégé par Avast.
> www.avast.com
> <https://www.avast.com/fr-fr/lp-safe-emailing?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=OA-2109-A>
> <#-6989394509917606694_DDB4FAA8-2DD7-40BB-A1B8-4E2AA1F9FDF2>
>