In the last Platform release we expanded the response types for
CONSTRUCT & DESCRIBE SPARQL queries to include additional RDF
serializations, including ntriples, turtle and JSON [1]. This provides
for more flexibility in how clients can handle the processing of
results.

This is particularly true for the JSON output which conforms to a
draft RDF/JSON specification we've had support for in the Platform for
a while now [2]. As with XML, there are many ways to craft a JSON
based serialization of RDF. This particular format is a resource
oriented serialization whose design prioritizes consistency in the
structure of the response over, for example, syntactic short-cuts.
We've been considering submitting the specification for
standardization -- there are multiple implementations of it (e.g.
support in ARC and Redland).  Feedback on the format would be
appreciated.

With the availability of JSON output, clients and client libraries can
remove overheads of XML parsing, so may be able to get a performance
benefit from switching between the serializations. Ian Davis has
updated Moriarty [3] to use the new output format, and there is some
initial support for it in the trunk of Pho [4]. Obviously things will
"Just Work" for AJAX apps.

Cheers,

L.

p.s. there's a nice JSON viewer plugin for Firefox available from
here: https://addons.mozilla.org/en-US/firefox/addon/10869/

[1]. http://n2.talis.com/wiki/Store_Sparql_Service#Output_Formats
[2]. http://n2.talis.com/wiki/RDF_JSON_Specification
[3]. http://code.google.com/p/moriarty
[4]. http://rubyforge.org/projects/pho