Re: Requirements for a possible "RDF 2.0"


here are my two cents (and I've read down this long thread).

I agree that an incompatible review of RDF would be bad now. We have only know the basic building block, but not a killer application yet.

Anyway, this is what I have found in my experience with RDF.

There are a range a features that people needs when using, they are not in the specs, and then people try to find they way, sometimes with bad results.

One thing is validation. 

While this is not in the RDF "mindset", people that exchange data in RDF want to be able to say when a "file" is complete or correct. They sometimes wrongly refer to OWL restrictions, sometimes to custom code.
Perhaps a standard additional vocabulary for rdf validation would help.

URI rewriting. It is often the case that it is more than known that two URIs point to the same thing.
Extreme case: they are informed from URLs of a system that has an old and a new URl style.
Now, to use sameAs in this case is my opinion is an abuse, and an unnecessary complication: a declaration of equivalence of namespaces (or something like that), to be considered at parsing time, would go a long way in making things easy.

Containers: I agree they don't look nice. But they don't harm that much either...

Blank nodes: to me blank nodes looks "de facto" as some small URIs... if you consider that they are asserted somewhere and that this somewhere is like a namespace. In practice, many times one at the end uses the pointer of the node (in some coding), and that's essentially like using a URI, for that file content.
This may be explained in a very confused way, but I think the presentation by Pat Hays at IWSC was making things very clear in this direction.

Reification: in general, I want to describe sub-graphs, rather than single statements. So to me it looks that named graphs (or a similar mechanism) could take over reification. Is there some drawbacks in this ? I'm wondering especially from the practical side of triple-store implementation... 

Relations with OWL. 
Don't get me wrong, I understand the need for ontologies and the like but, at this stage, RDF and related technologies can be proposed in "semi" production environments. The same doesn't go for OWL yet.
And a similar situation applies for people's know-how.

When I talk about RDF to people, in a bio-domain, all the complains I get are about OWL and the fact that it's too complex (including staffing problems). They all agree that RDF is very simple and good. But at the end of the day, sometimes they turn away in front of the "overall complexity".

Perhaps avoiding unnecessary usages of OWL (when it's not needed) (sameAs, validation are two examples) could help.

There is an issue that I didn't see in any mailing list:

I think a good reason for the relatively low diffusion of RDF is that, at least until not much time ago, there was not much training available on RDF/SemanticWeb subjects.


On 13 Jan 2010, at 16:00, Chris Welty wrote:

> Without volunteering myself to be such a contact, I have (as both a 
> users of many RDF implementations and a W3C chair and I suppose a 
> self-declared semantic web expert) been the recipient of a lot of 
> complaints and suggestions regarding the design and implementation of 
> RDF, and at ISWC a few months ago I suggested to Ivan that we start 
> discussing starting a working group that would investigate a next 
> version of RDF.
> This discussion is happening in several places already, and we thought 
> this was the best place to house that discussion for now.
> A workshop on this subject is also in the planning, more news on that in 
> a week or two.
> 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 (I'd like this thread to be 
> less "+1" and "-1" messages, and more "I'd like to see RDF support x...")
> -Chris
> -- 
> Dr. Christopher A. Welty                    IBM Watson Research Center
> +1.914.784.7055                             19 Skyline Dr.
>                           Hawthorne, NY 10532

Andrea Splendiani
Senior Bioinformatics Scientist
Rothamsted Research, Harpenden, UK
+44(0)1582 763133 ext 2004

Received on Friday, 15 January 2010 16:25:55 UTC