Re: RDF Syntaxes 2.0

Hi Dave,

Great to hear your thoughts on these matters! I certainly know that
you've been at these topics many times before.

> People moan about RDF/XML and have for years.  I even wrote down in great
> detail the flaws in [1].  Over all that time nobody has come up with a
> credible and complete XML syntax alternative that stuck, even myself.  Let
> me summarize the ones I know:
> - TriX: had little takeup
> - RXR[1]: ditto
> - GRIT: new, but flawed since it can only represent trees (no named bnodes)

In defense of Grit [1]: there is currently no specified support for
named bnodes, that's true. Apart from that, it's true to the RDF graph
model.

If Grit is to be a full RDF format, I'd be happy to eliminate the
"unnamed bnodes only" restriction. The easiest way is to just allow
references of the form "_:xyz" where only URI:s are allowed now (@uri
and @ref). (This mostly in order to keep it uniformly transformable --
I could just as well go with e.g. separate @id/@idref or similar. But
sparse syntax is a goal.)

I'm not sure how much interest there is for Grit as a general format
though (as said, it started out as a way to distinguish a
normalization from RDF/XML). I certainly do think it has potential,
with its goal to be simple, normalized and usage friendly.

(It can also be fitted with support for named graphs without much
effort. Apart from defining what they are..)


> Right now I'd dismiss any XML format for any "simple" or "obvious" way to
> write down RDF graphs that will be accepted by new users.

Could you elaborate on this? Are there any conditions under which
you'd think Grit would have merit as a general format? I'd be happy to
collaborate on it in any way required.


> So I'm happy with how Turtle turned out and that should be the focus of RDF
> syntax formats *for users*.  It does need an update and I'll probably work
> on that whether or not a new syntax is part of some future working group - I
> have a pile of fixes to go in.  Adding named graphs (TRIG) might be the next
> step for this if it was a standard.

I couldn't agree more. Turtle should certainly be standardized. (And
TRiG, perhaps as a distinct superset.)

I also do think Turtle is far more important to the existing RDF
community than a *new* XML format. An endorsed JSON format may also be
higher up that list.


But I think RDF should to be approachable and sellable to new, even
"casual" users. Who haven't yet bought the conceptual model, nor begun
thinking of the merits of the "upper layers". A lot of the industry
(at least here in Sweden) just marshal idiomatic data records to/from
"some kind of text" (hopefully a standard one like XML or JSON), and
hope that encodings, timestamps, maybe even the idea of "a name" just
won't garble in the numerous, sometimes dubious, translations at the
system borders..

Linked Data is getting traction, and catering to people who have
recently opened their minds to the relative benefits of Atom, JSON
etc. is crucial, IMHO. Thinking in "framed records" is often good
enough (even the best approach, at least if they're linkable), but
there are few RDF formats "packaging" the triples in such a friendly
form. Turtle does a good job of making RDF easy to handle. But it's
neither XML nor JSON, which is a requirement for many.

(And showing up to the party in RDF/XML just might not make you seem
like the most environmentally friendly participant.. ;])

Best regards,
Niklas

[1]: http://code.google.com/p/oort/wiki/Grit

Received on Tuesday, 26 January 2010 22:20:53 UTC