Re: Requirements for a possible "RDF 2.0"

On 14 Jan 2010, at 09:05, Danny Ayers wrote:

> 2010/1/13 Chris Welty <cawelty@gmail.com>:
>
>> I suppose we don't really need to discuss whether we should  
>> investigate an
>> "RDF 2.0", but rather what kinds of requirements various RDF users  
>> have that
>> they would like to be considered
>
> Personally I don't think the time is right for an "RDF 2.0"
> (suggesting a major overhaul), though an "RDF Second Edition" (as Jiri
> suggests) may be desirable.

Agreed. Anything that broke back-compatibility at this point would be  
a huge mistake in my opinion.

> There's still a large amount of consolidation work needed (notably
> with the recent arrival of RDFa, OWL2 and RIF), and probably more
> significantly there aren't any major technical blocks to widespread
> deployment based on the current specs - i.e. if it ain't broke, don't
> fix it. If there are such problems, then let any spec changes be
> driven by real-world use cases rather than change for change's sake.
>
> Having said that, there are a few areas that could probably use
> attention given developments since the 2004 specs -
>
> Out:
> * containers (Bag, Alt, Seq) - quiet deprecation would probably be the
> best bet, supported by clear instruction on how to use Lists

Well... partly, I'm still not sold on Lists, they're hugely  
inconvenient to deal with in their triple form, and cause significant  
data-bloat.

> * reification - ditto, with named graphs as the modern alternative
>
> both features can cause a lot of confusion, seeming more useful than
> they actually are

Yup.

> In:
> * improved support for named graphs - essentially bringing the
> constructs included in SPARQL back into RDF core (including support
> for named graphs in RDF/XML, done in a manner that would be
> backwards-compatible if at all possible)

I'm not really sure how that fits all together. If you dereference  
some URI, and get back a RDF/XML document that includes other named  
graphs, what then? Surely the grph URI of the document you fetched you  
be the URI you dereferenced.

> * W3C-blessed subsets of the model & syntax - on the model side, e.g.
> a profile that excludes blank nodes, on the syntax side e.g. a profile
> that uses the triple-shaped rdf:Description style of RDF/XML
> (basically I reckon it's too late to make breaking changes to RDF/XML,
> but it still remains a significant hurdle for newcomers)

Perhaps, bNodes could certainly use some clarity. I find them very  
useful in omdelling, but the whole existential variable thing is a bit  
awkward.

> * an RDF/XML profile (as above) designed for maximum compatibility
> with XPath/XSLT/XQuery

That would be nice. I often thing that RDF/XML is not a terribly good  
way to encode RDF in XML.

- Steve

-- 
Steve Harris, CTO, Garlik Limited
2 Sheen Road, Richmond, TW9 1AE, UK
+44(0)20 8973 2465  http://www.garlik.com/
Registered in England and Wales 535 7233 VAT # 849 0517 11
Registered office: Thames House, Portsmouth Road, Esher, Surrey, KT10  
9AD

Received on Thursday, 14 January 2010 12:01:22 UTC