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

Re: A thought on fragments and rec-track

From: Jim Hendler <hendler@cs.rpi.edu>
Date: Thu, 14 Feb 2008 10:41:59 -0500
Cc: public-owl-wg@w3.org
Message-Id: <A662BABA-D45F-497C-94C0-E1FBB0D08207@cs.rpi.edu>
To: Carsten Lutz <clu@tcs.inf.tu-dresden.de>

I think Carsten may be onto something - a couple of quick responses
  1 - I think OWL DL and OWL RDF (whether by that name or not) works,  
and just as in OWL 1.0 we tried to keep these in line, but not  
identical, I think that would be the win strategy here as well (i.e.  
the "vocabulary" should be releatively consistent, and the semantics  
wants to correspond as best it can to the "natural" meaning of this  
vocabulary -- that's not meant to be obscure - I just mean we try to  
use mnemonic names that correspond to the formalization)
  2 - I think the "OWL DL minus" and the "OWL Full minus" should also  
be made to correspond to some degree, but I agree w/Carsten that  
keeping them identical is less important - I think a lot of the  
constructs we want in each would map well, so I don't think this would  
be so hard.
  3 - I agree having a note, not a rec, that defines useful properties  
of particular language restrictions in DL would be very useful (note:  
this would be more or less our current fragments document renamed in  
some way to disambiguate the idea of fragments in the rec from useful  
things for implementors to know, in the note)
4 - The one place I think there is a problem w/the below is with the  
current OWL Lite.  We cannot "take it off the rec track" - it is  
already a W3C Rec for better or worse.  To understand this, think of  
RDFS instead of OWL Lite -- we could suggest adding some things to  
RDFS, we could suggest some clean up (backward compatible) of RDFS  
syntax or semantics, but we couldn't say "no one use RDFS anymore" --  
OWL Lite has exactly the same status (with respect to W3C and the  
outside world) as RDFS (or OWL DL or Full for that matter) - so our  
hands are a bit more tied with respect to that.   I think that the  
solution Alan was thinking of - doing the minimum to keep the  
currently defined LITE consistent w/the new recs, would work here --  
basically, we would be suggesting, implicitely that the new OWL DL  
minus and OWL Full minus are better replacements for Lite and the Full- 
version-of-Lite (whih is not actually a rec, but some people refer to  
in talks and documents) - and we'd be able to better motivate them.    
But as best I understand the world, Sandro and Ivan correct me if I'm  
wrong, if we say nothing at all about OWL Lite, then it remains a Rec  
exactly as it is - so it would be the "fifth" fragment (but less  
compatible) than if we actually clean it up, but motivate the new ones  
better
  I hope that makes sense, some of these things are hard to word right  
-- just in case anyone is unsure:

Summary: I like Carsten's idea, and am just puttering around the edges  
to make it more consistent with the outside world that I interact with.
  -JH


On Feb 14, 2008, at 2:55 AM, Carsten Lutz wrote:

>
> Dear WG,
>
> yesterday's discussion on fragments and rec-track showed once more
> that, simplifying a lot, the WG is split into two groups. Let's call
> them the RDF group and the DL group. Each of them has its own view and
> valid arguments that support it. I believe that neither the WG nor the
> two groups benefit from a confrontative way of dealing with this
> situation.  Instead, we should try to have peaceful coexistence
> whenever possible. Here is a simple way how this could be achieved for
> the fragments/rec-track issue:
>
> - We give OWL Full a rule-based semantics (it seems likely to me that
>  this will be decided anyway)
>
> - Then, there are two fragments in rec-track: one OWL-Full fragment
>  (for the RDF group) and one OWL-DL fragment (for the DL group)
>
>  So we have : OWL Full, OWL Full Minus , OWL DL, OWL DL Minus
>
>  Never mind the naming scheme (IMHO, we should even rename OWL Full
>  to OWL RDF).
>
> - As OWL-Full Minus, we choose OWL Prime. Unless I overlook something,
>  it is easily defined as a fragment of OWL-Full. Here, "fragment"  
> means
>  that it selects two things: a) a subset of the OWL-Full vocabulary
>  and b) a subset of the OWL-Full semantic rules.
>
>  This means that we do not need Boris' bright but complicated
>  construction to capture OWL Prime as a syntactic fragment of OWL DL.
>  Neither do we need conformance  levels (but we could still have them
>  if we want).
>
> - As OWL DL Minus, we choose a real *syntactic* fragment, as this is  
> what
>  the DL group seems to imagine. I think this must be EL++, and am
>  happy to again provide motivation for why this is the case, and what
>  are the serious and commercial applications. But in general, that's
>  another issue.
>
> - All the other fragments also remain, but don't go into rec-track.
>
> greetings,
> 		Carsten
>
> PS: As for OWL 1.0 Lite, I suggest that we don't make it rec-track  
> any-
>    more, but explain in rec-track why this is the case and that OWL  
> 1.0
>    Lite users are now simply OWL 1.1 DL users, and thus fully  
> supported.
>    This is what I understood Alan was proposing. I don't view the  
> issue
>    of punning as problematic.
>
> --
> *      Carsten Lutz, Institut f"ur Theoretische Informatik, TU  
> Dresden       *
> *     Office phone:++49 351 46339171   mailto:lutz@tcs.inf.tu-dresden.de 
>      *
>

"If we knew what we were doing, it wouldn't be called research, would  
it?." - Albert Einstein

Prof James Hendler				http://www.cs.rpi.edu/~hendler
Tetherless World Constellation Chair
Computer Science Dept
Rensselaer Polytechnic Institute, Troy NY 12180
Received on Thursday, 14 February 2008 15:42:32 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 14 February 2008 15:42:32 GMT