Re: EXTERNAL: Re: validation of json-ld to context schema

Hi Dave,

> On December 5, 2014 at 9:18 PM "David I. Lehn" <dil@lehn.org> wrote:
>
>
> On Fri, Dec 5, 2014 at 11:36 AM, peter <peter.amstutz@curoverse.com> wrote:
> > It's not clear what the right answer to validating json-ld is.
> > json-schema is not ideal for the above reason, RDFS is not very
> > powerful, and OWL is not really intended for validation (so I'm told).
> > Are there other options?
> >
>
> Validation is a complex problem that hasn't been well addressed yet.
> Eventually tools and techniques will emerge. For now, you can get
> fairly far with json-schema tools. It's not semantic validation but
> can be good enough. The key is to get your data into a known JSON-LD
> structure then process it as if it were plain JSON. One way to do this
> is to just to restrict how the input data is structured. That's not
> very flexible but it can work for now. The other thing you can do is
> use the work-in-progress framing spec to frame your input data first.
> In either case json-schma tools can then be used. There are
> limitations to structural validation and this may not cover every type
> of validation but it works well. For more complex cases just write
> additional code. We've used this for the payswarm work and some other
> projects with some success.
>

Have done similar using normalized RDF/XML and validation using some XML schema
(take your pic from XSD or Relax NG).
Downside of these approaches is you inevitably end up working (at least
partially) within the constraints of the particular serialization format/syntax
rather than validating the RDF graph.
Hopefully the work of RDF Data Shapes WG will help to resolve this gap.

> -dave
>