[openbiblio-dev] JISC openbib update

[Mark MacGillivray]

(copied in to openbiblio-dev list now)

JSON-LD, RDF/JSON, RDFlib

Yesterday we looked at writing a parser and serialiser for RDFlib to
handle JSON-LD, as JSON-LD was expected to be the best of various JSON
linked data formats. Here are some updates:

* We can write a parser for RDFlib, it is not that hard
* but we need good info on the algorithm to implement to create a
successful parse
* unfortunately, JSON-LD is insufficiently specified to enable this
* JSON-LD does not clearly explain how to encode an entire document -
rather it describes how to encode snippets
* There are advantages of JSON-LD over RDF/JSON, although they are not
critical requirements
* in particular, whilst JSON-LD provides a format that is easier to
manage at the front end (e.g. via javascript) it is not worth trying
to implement an underdefined spec just for that advantage
* JSON-LD does give namespace handling which we do not get in
RDF/JSON. This would be great, but requires more specification
* Hence, we believe our time may be better spent doing an RDF/JSON
parser, and building on that simple spec where necessary to
incorporate the functionality that JSON-LD would offer (more explicit
namespace handling)
* This could be done within a week, (bringing up the RDF/JSON parser
written for an old version written by Rob Sanderson of RDFlib to meet
3.1 requirements) as opposed to a JSON-LD implementation, which would
require a few weeks of additional speccing before any coding began

So, we would now like some feedback; should we proceed with RDF/JSON?
Have we missed details about JSON-LD that would help us successfully
parse and serialise entire documents? Here are some more details about
where the JSON-LD spec is lacking:

* sections 7 and 8 are incomplete and have warnings
* we cannot clearly define how to parse / serialise an entire
document, because the way JSON-LD specifies serialising records does
not build actual JSON - therefore it is unclear how we build these
fragments together into a reliably parseable JSON document

Please let us know what we are missing!

Mark



On Tue, May 3, 2011 at 10:17 AM, Peter Murray-Rust <pm286 at cam.ac.uk> wrote:
>
>
> On Tue, May 3, 2011 at 9:43 AM, Rufus Pollock <rufus.pollock at okfn.org>
> wrote:
>>
>> Good to hear. Please, please can this sort of discussion start going
>> on openbiblio-dev ;) (only use private alias when we absolutely feel
>> it is necessary!)
>
> Agreed - I nearly reposted it.
>
> Looking forward to it
>
> P.
>>
>
>
> --
> Peter Murray-Rust
> Reader in Molecular Informatics
> Unilever Centre, Dep. Of Chemistry
> University of Cambridge
> CB2 1EW, UK
> +44-1223-763069
>