Fwd: Status of OWL 2 Conformance

I meant to cc: the below to public-owl-wg, sorry. Bijan replied
privately and should feel to forward his reply to me here as well.

Jonathan

---------- Forwarded message ----------
From: Jonathan Rees <jar@creativecommons.org>
Date: Wed, May 27, 2009 at 12:40 PM
Subject: Re: Status of OWL 2 Conformance
To: Bijan Parsia <bparsia@cs.man.ac.uk>

OK, it looks like the way 2.1 is written I missed out that document
datatype conformance is always *implied* by syntactic conformance. If
so then instead of saying that D is an "OWL 2 DL ontology document
using only literals having datatype URIs in the domain of the OWL 2
datatype map" as I had thought would be needed, one can merely say D
is an "OWL 2 DL ontology document".  Similarly for EL QL, RL. I.e.
depending on an extended map makes a document nonconformant (in this
particular way). That's good.

In my defense, commitment to the OWL 2 datatype map is not obvious
from the definitions of "OWL 2 ...L ontology document"; you have to
scan down to the second note in that section to discover it for DL,
and for EL and so on, you have to read the profile descriptions to
find out that use of extended maps is not conformant. I didn't do this
until your message led me to. Section 2.2 didn't help me out here,
since it says nothing about document datatype conformance, implying
that there's no such thing (there is, but it's called "syntactic
conformance" instead of "datatype conformance"); and the part about
conformance being relative to a datatype map threw me off.

So now I think it's just an editorial problem. However I have little
confidence that I've untangled this correctly.

Jonathan

On Wed, May 27, 2009 at 11:07 AM, Bijan Parsia <bparsia@cs.man.ac.uk> wrote:
> On 27 May 2009, at 15:43, Jonathan Rees wrote:
>
>> You're asking me to compare wiki version 16385 (Dec 2) to the current
>> version (21970), right?
>> There are a lot of diffs, and I don't think I can check them all, sorry.
>>
>> http://www.w3.org/2007/OWL/wiki/index.php?title=Conformance&diff=21970&oldid=16385
>>
>> I do see that the problem I complained about before, that the
>> conformance document
>> gives no name to any class of ontology documents that satisfy *both*
>> syntactic conformance of any kind and datatype conformance, is still
>> there.
>
> Is this really a problem? I don't see that it (practically speaking) is. The
> current situation (with some required and a bunch of optional datatypes)
> doesn't seem to be made worse by the lack of such names. On the contrary,
> I've never seen anyone look for such names. People just ask if e.g., Pellet
> supports such and such a datatype.
>
>> The problem is compounded by the fact that datatype conformance is
>> only defined for tools, not for documents.
>
> Again, in practice, this has been a nonproblem, and I don't see that
> changing.
>
>> People exchanging ontology documents who want to establish some kind
>> of agreement around what's being transmitted
>
> First, there's are lots of agreements people might want to establish. So
> please be precise.
>
>> will have to invent for
>> themselves a name for the kind of conformance they need
>
> Uh, no. I don't see that as necessary or likely.
>
>> - i.e.
>> syntactic conformance (pick one) plus the use of the OWL 2 datatype
>> map - and there's a serious risk they won't even think to do this at
>> all,
>
> To no harm.
>
>> and will get confused over exactly what constraints the exchanged
>> documents need to conform to.
>
> I don't see why it's a huge deal to say, "We're going to only use
> xsd:decimal" if that's what you want. It's easy to say, communicate, detect,
> etc. Trivial even.
>
>> This seems like a serious source of future headaches to me.
>
> Since it's the current status quo, and this bit hasn't been a source of
> headaches, per se, I'm pretty skeptical that it'll be a source of future
> headaches. I do know that getting into further nomenclature is a likely
> source of current and future headaches (judging from what we've
> experienced).
>
> As a likely "syntax advisory" tool developer, I intend to maintain a list of
> what reasoners support what datatypes and issue warnings based on that. It
> seems more than adequate for the problem, such as it is.
>
> Cheers,
> Bijan.
>
>

Received on Wednesday, 27 May 2009 17:29:35 UTC