do we really need two languages/levels? [Issue 5.2]

On Wed, 2002-07-10 at 10:31, Guus Schreiber wrote:
> Issue 5.2 Language Compliance Levels
> Proposal to CLOSE these issues in conformace with the resolutions of the 
> Stanford face-to-face meeting [1], as reflected in the feature synopsis 
> documentt [2]:
> - OWL will have a named subset, called "OWL Lite"

I understood the decision to be about what to call the
language, provided we would have one, not a decision
to actually have a subset. That's why I abstained:
I don't care what we call the subset, for now, because
ultimately, I don't think we should have one.

If this was a decision to have two languages, the
record should show me as objecting, not abstaining,

I remain unconvinced that the value of having two langauges
outweighs the cost.

Regarding the cost of level etc. pls see

  # levels/options considered harmful
  Dan Connolly (Thu, May 16 2002)

and the following thread.

When last we discussed this, I was one of about 7 who
favored having just one language:

"A straw poll showed more
in favor than against (~11 vs !7) but no consensus.
 issue 5.2 Language Compliance Levels remains open"
  -- 2 May minutes

I think that yes, OWL should be "light"er than DAML+OIL,
but that we should measure this light language against
our requirements, and make the hard choices about whether
to add features or demote requirements.

Reviewing the list of what's in OWL but not in owl light,

I think disjointWith is necessary for our requirements
about finding inconsitencies. Hmm... I think I have
used stuff like unionOf and intersectionOf in applications...
I wonder if we could limit the semantics to something
tractable... prolog systems have unionOf and intersectionOf,
don't they? or maybe they do, but just for finite lists/sets.
Hmm... I'll have to think about some test cases that
sit on the border and see which way we think they should go.

But I think we make the maximum contribution to the community
by coming up with *one* language.

> - OWL Lite includes both universal and existential local range restrictions
> - OWL Lite includes min/max cardinality restrictions of 0 and 1
> [1]
> [2]
Dan Connolly, W3C

Received on Wednesday, 10 July 2002 16:33:09 UTC