Re: ISSUE-130 / ACTION-194 Come up with a proposal for conformance

Bijan Parsia wrote:
> On 3 Sep 2008, at 14:34, Ivan Herman wrote:
>> Ian Horrocks wrote:
>>>> Just for my understanding: what would that require for an
>>>> implementation? Would it mean that the RDF graph has to be converted
>>>> into the functional syntax and check against RDF-RL?
>>> That is how it is *defined*, but tools are free to *implement* it in any
>>> way they choose -- it might be possible, e.g., to implement checks that
>>> operate directly on the RDF graph.
>> Sure, but that would not make it simpler. The huge advantage of the
>> OWL-RL is that it can be implemented (o.k., with scruffy edges here and
>> there) in an afternoon on top of an existing RDF environment.
> Actually, this is true for all the profiles and maybe a week for OWL DL.
> Whether these implementations would be *production quality* is an
> interesting question (the OWL DL one, definitely not :)).
> I have to say that catering to the "implement in an afternoon" base
> probably isn't really sensible. After all, you can implement (most of)
> it in an afternoon either way.

O.k. I was not precise. By 'Implement in an afternoon' I meant for
somebody who does not really know much more than having good programming
skills and knowing how to use some sort of an RDF environment. I do not
believe that would be true for OWL DL, EL++ or DL Lite, which do require
a knowledge of DL algorithms that most mortals do not have (I certainly
don't). For that to be acquired is probably more than a week...

Anyway. The issue is that implementing OWL-RL/RDF is very easy, way
easier than the others. Can we agree on that?

>>  Such an
>> extra 'must' check would make it way more complicated. Hence my
>> preference of leaving it as a 'may'
> Technically, I think they could claim conformance by providing a
> separate check tool, such as will soon be freely available both for
> download/distribution and as a web service :)
> Is there anything wrong with "SHOULD" here? I think it ought to be
> encouraged, at the very least. A reasonable argument for violating the
> should could be "too great performance impact".

... which is a relative notion to the rest of the tool to be
implemented. And I fear this is the case here.


> Bijan.


Ivan Herman, W3C Semantic Web Activity Lead
PGP Key:

Received on Wednesday, 3 September 2008 14:42:11 UTC