Ergonomic output for JSON-LD?

I working on a JSON based API where I want to turn the JSON somebody sees
into an RDF graph,  then I make a new RDF graph and send it back as JSON.

Internally the system uses the JSON-LD transformation,  and it takes just a
@vocab statement to handle the import from a format that looks like it was
design natively for JSON.

On the way back,  however,  it seems not so easy to get output that JSON
users would recognize.  For instance you can channel N-triples pretty
easily,  but you need some kindof plan to create JSON-LD which looks like
the kind of JSON they would put into Mongodb or Marklogic and the tools I
have seen so far do not have it.

Is there something better out there?

-- 

On Beyond Owl
http://www.slideshare.net/jahendler/on-beyond-owl-challenges-for-ontologies-on-the-web

Making the semantic web work
http://www.slideshare.net/paulahoule/making-the-semantic-web-work

:BaseKB -- Query Freebase Data With SPARQL
http://basekb.com/gold/

(607) 539 6254    paul.houle on Skype   ontology2@gmail.com