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

Re: ISSUE-111 Proposal to Resolve

From: Bijan Parsia <bparsia@cs.man.ac.uk>
Date: Tue, 19 Aug 2008 16:39:34 +0100
Message-Id: <A04B21E0-B4D7-4462-80DE-4A025087061C@cs.man.ac.uk>
Cc: Ian Horrocks <ian.horrocks@comlab.ox.ac.uk>, public-owl-wg Group WG <public-owl-wg@w3.org>
To: Jim Hendler <hendler@cs.rpi.edu>

On 19 Aug 2008, at 16:21, Jim Hendler wrote:

> but isn't that a circular argument

No.

> -- you say people will use sameAs sameAs sameAs to signal intent,

No. sameAs sameAs sameAs makes the ontology OWL Full.

> but it will be bad for them to have to have something to signal  
> intent.

I think think we revealed enough issues with signaling user intent in  
generally that we don't have good grounds for moving forward.

> Basically, my problem is that I firmly believe that in language  
> design the overloading of terms is a bad idea

We aren't.

> .  SameAs has a very specific meaning and it is highly used -  
> someone seeing sameAs sameAs sameAs in an ontology seems to me to  
> be much more likely to be confused than someone seeing  
> "owl2:OWLFullOnly" or whatever semantics free tag

It, by definition, isn't semantics free. It affects the semantics of  
the document.

> we use -- we could even put it within the ontology header --- seems  
> to me you still haven't really shown me a downside

It clearly involves 1) additional complexity to the language and 2)  
potential cause for misunderstanding of its actual effect.

I would have been much more sanguine about this (since its what I  
proposed) if several people didn't immediately start wondering how it  
affects imports, other tools, etc.

The experience thus far hasn't been good. That seems enough for me.

> -- you say the downside is that people would need to signal intent,  
> but then propose a solution where they still have to signal intent,  
> just in a less transparent way

They don't signal *intent*. They just have made their ontology  
actually unambiguously OWL Full. That's a big difference.

>  btw, I'm also fine with a solution where there is no signaling at  
> all for OWL Full, which might be the compromise -- the thing I  
> don't like is the overuse of existing syntax - so another  
> alternative is to simply leave things as they are in OWL 1.0 (i.e.  
> that ontologies are what they are) - would that be more acceptable?

Since it's only advice I don't see that we've changed things. The  
point is that we *haven't* changed things! We've just pointed out a  
possible work around for the few folks (like Sandro) who care. This  
is why Ian doesn't want to make the triple mandatory.

>  -JH
> p.s. And again, let me stress that I would suggest we only create  
> one term, and it is only intended for OWL Full only

But then you get people asking, "Why only OWL Full? Why not for EL?"  
and away we go :(

> -- I do indeed agree that generally signaling intent is a bad idea  
> - but if the WG feels it is needed for some reason in this case  
> only, then we should do something explicit.

The situation is that some people (e.g., Sandro) have indicated  
discomfort with the fact that some ontologies are not syntactically  
distinguishable as OWL DL or OWL Full. We considered adding a general  
intent signalling mechanism but it ran into a lot of problems. Thus  
we pulled back and said, "IF you care about this, THEN you can ensure  
that your OWL Full ontologies are definitely syntactically OWL Full  
by including a trivially entailed triple such as sameAs^3."

This involves *no change* to the language. It is not required. It  
doesn't change the tool chain in any way. It doesn't change the  
conceptual model of the language in any way.

Cheers,
Bijan.
Received on Tuesday, 19 August 2008 15:41:18 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 19 August 2008 15:41:18 GMT