Re: Today's equivalent of CBD+URIQA? (was: Re: Schism in the Semantic Web community.)

On 2009-01-27 13:53, "ext Danny Ayers" <> wrote:

> 2009/1/27 Manos Batsis <>
>> Danny Ayers wrote:
>>> i.e. maximally exploiting HTTP
>> Ah, nailed. It's too bad I have fallen behind regarding the state of the art,
>> but the most horizontally useful idea I've ever stumbled upon re semweb was
>> CBD [1] plus URIQA [2]. Essentially CBD defined graph limits (what we do
>> with,
>> e.g., lazy collection loading in the object-relational world). URIQA was a
>> fitting HTTP extension anyone could implement to acquire/exchange/CRUD those
>> graphs.
>> So, what is today's equivalent of those two?
> I'd love to hear Patrick Stickler's retrospective opinion on these, I remember
> him forcefully (and pretty convincingly) arguing the case.
> But I suspect now we have named graphs in the picture, along with SPARQL (and
> proposed update protocols), and the TAG finding on httpRange-14, it has become
> a non-question.

I actually find URIQA and CBD to be just as relevant today, with named
graphs, SPARQL, etc. and don't consider httpRange-14 to either support or
deprecate those concepts.

In fact, I consider CBDs (and their various cousins, e.g. SCBDs which we use
a lot at Nokia) to be optimal for the SPARQL DESCRIBE result form, with a
default to CBD and the ability to specify some other result form in the

(it's perhaps worth noting, that URIQA and CBDs/SCBDs/etc. are not
inseparable; one can take one and not the other)

At the risk of making a potentially offensive statement, I think the lack of
uptake on URIQA was essentially political, since I never saw any convincing
arguments why it was not an excellent solution to the general problem of
authoritative and third party knowledge discovery. There were too many folks
with vested commercial and other interests in other approaches. And that
said, I really am disinclined to enter into further debate/promotion of
those worthy concepts in this particular context, as I don't see that the
general attitudes or viewpoints have changed.

We continue, regardless, to do what we know works well, scales well, and
provides excellent real-world results.




Patrick Stickler
Chief Architect
Forum Nokia Developer Infrastructure & Operations
+358 50 4823 878

> What still bothers me somewhat is there is a danger of reinventing the silo
> mentality for data, rather than exposing directly addressable resources, using
> the link whether encoded in RDF/XML or whatever. We now have a much better
> understanding of how follow-your-nose through URIs can allow enormous
> scalability, and to some extent how this material can be exploited locally
> through inference and/or Web 2.0-inspired modes of visualization.
> In other words, although we can expose big data on the Web through mapping
> techniques from traditional DBs or more cleverly through OWL(2), we may be
> losing sight of the succesful bits of the Web as it stands. I say this with
> the caveat that I honestly don't know for sure, and these kind of issues have
> a tendency to sort themselves out in the massively distributed environment.
> Cheers,
> Danny.
>> [1]
>> [2]
>> Cheers,
>> Manos

Received on Monday, 23 February 2009 07:15:54 UTC