W3C home > Mailing lists > Public > public-owl-wg@w3.org > January 2008

Re: Punning discussion

From: Bijan Parsia <bparsia@cs.man.ac.uk>
Date: Thu, 24 Jan 2008 11:46:45 +0000
Message-Id: <894FD5C6-9AB9-4E26-B2A9-EAAE25BA3307@cs.man.ac.uk>
Cc: OWL Working Group WG <public-owl-wg@w3.org>
To: "Michael Schneider" <schneid@fzi.de>

On 24 Jan 2008, at 10:48, Michael Schneider wrote:
[snip]
>> From: Alan Ruttenberg
>> Sent: Tuesday, January 22, 2008 9:21 AM
>
>> 3) [...] How does punning effect OWL Full?
>
> Punning is a DL-only feature,

This doesn't seem right. If you take the equality free fragment of  
OWL Full, you cannot distinguish between punning and hilog semantics.  
RDF, the language, doesn't have equality, so it is compatible with  
punning.

> and cannot affect the definition of Full, as
> long as Full is understood to be an RDF compatible language.

The semantics specs something in a style that is *suggestive* of how  
to extend it to OWL, but that's a different thing. It doesn't say  
*how* to extend the semantics. It seems as legitimate to add  
parameterized extension relations as anything else.

> In RDF, two
> occurrences of the same name denote the same entity.

Again, you seem to be inflating the style of the spec into a general  
principle. It's very straightfoward to give a semantics where this is  
not true but is equivalent to the RDF semantics.

> Full cannot be adjusted
> to punning semantics without losing its compatibility to RDF's
> model-theoretic semantics.

First, it's clear that you are working with a very loaded and  
specific notion of "compatibility". The qualification to  
"compatibility...semantics" is esp. telling. This is fine *if* we can  
agree 1) what that is and 2) that it is binding (or, rather, a  
significant consideration). At the moment, there are variants I can  
imagine that I would find reasonable to try to respect and those  
which I don't find even remotely in the running.

There is precedent: The SPARQL query language semantics are not given  
in terms of the RDF semantics (even simple entailment). It's probably  
possible, but it proved to be so difficult and not useful that we  
gave up.

> Maybe the "punning for metamodelling" usecase suggests this, but  
> punning is
> not a feature in the same sense that QCRs or IFDPs are features.  
> The latter
> can easily (at least in principle) be given proper semantics in Full,

Consistent semantics? That doesn't seem easy even in principle.

> without the danger of getting into conflict with the core RDF  
> semantics.

Let me *grant* that there is a "conflict" (though, again, I'm not  
convinced): Even then we have to weigh *all* the considerations.

The HTML working group published a design principles document which  
contains a very interesting one that we all work with but rarely  
articulate:
	http://www.w3.org/TR/html-design-principles/#priority-of-constituencies

(Although, one part of this principle is missing: If the implementors  
don't get on board, then it doesn't matter what the users, speccers,  
etc. *say*. Similarly, if the users won't user it, nominal  
implementation doesn't matter.)

I think "theoretical compatibility with prior specs" falls way down  
on the priority ladder.

> Punning, however, makes more fundamental assertions about what  
> semantical
> interpretations are, by allowing the same name to denote more than one
> entity.

As I've said before, this isn't true. RDF semantics already separates  
classes as individuals and classes as sets (in the interpretation  
domain). It's very easy to see that instead of mapping to distinct  
individuals contextually, you vary (in the semantics) the mapping to  
sets. I.e., you change IEXT.

Now you might say, "You can't change IEXT!!!" But I don't see why.  
OWL is a *semantic extension* of RDF. It *already* changes many  
aspects of the RDF semantics (it *has* to or they'd be the very  
same :)).

[snip]

> Anyway, for RDF, the form of semantical interpretations is
> specified in the RDF semantics spec [1],
[snip]

Sorry, the RDF semantics document specs *RDF*, not "the form of  
semantical interpretations". It even points out alternative ways of  
specing the semantics (via translations) and even *gives* alternative  
formulations (the entailment rules).

In fact, there is no need for OWL Full to *be* a semantic extension  
of RDF. That's what it *was*, but one could sensible argue that that  
was a mistake and that the OWL WG should not have gone down that road.

Note that I am here arguing against your arguments, not for any  
conclusion. It's obvious what conclusions I prefer :) but one could  
make an argument for old style "kitchen sink" OWL Full.  
Unfortunately, it is, on the merits, rather weak (no implementation,  
users, surprising semantics, consistency problems, etc. etc. etc.).

OTOH, there are plenty of aspects of OWL Full which are reasonable  
and justifiable. I believe punning is one of them. (Yes, punning *is*  
in OWL Full :))

Cheers,
Bijan.
Received on Thursday, 24 January 2008 11:45:03 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 24 January 2008 11:45:04 GMT