Re: DOCUMENT: Layering document review (was Re: UPDATE: status of longer version of layering document)

Frank - the problem is that 1-4 as you define them below are 
APPROACHES, not yet proposed solutions to move on.  I mean we need to 
figure out how we operationalize these approaches and move to 
language features that support them.
It is fine and good for the group to decide "We will have a XXX 
syntax and YYY semantic extension" (or whatever), but that doesn't 
say which things will be the same or different from current DAML+OIL 
or why.  We clearly need to set philosophy as part of moving ahead, 
but the "task force" I suggested yesterday will need to go through 
D+O figuring out how we change it, if at all, to deal with the issues 
we've been discussing -- I'm trying to be hard-nosed here and remind 
everyone that we need to eventually get this all the way to language 
features.

  To date, I see nothing that specific on the table, and we need such.
For example - a differing syntax solution could be anything from 
"chuck the entire DAML+OIL and start again from scratch using C++ 
syntax" to "if we simply add the tag <OWL></OWL> around the OWL 
equivalent of daml:class statements then everything works again"  -- 
that's quite a range!  Also, it is complicated by the fact that some 
of our reasoning may need to go like this:

  If RDF were changed to allow XXX then our language works as is (or 
with the following minor modification)
  if not, then we need to drastically change our language to YYY (for 
example, largely abandoning RDF syntax)

which is the best way we could be able to communicate this sort of 
thing back to an RDF WG (either Core or future RDF 2.0) for 
consideration or joint solution.

by the way, in the space of solutions is to simply acknowledge the 
problem, leave DAML+OIL essentially as it is (modulo our 
requirements) and make the suggestion that some sort of new work 
(some future joint task force or such)  will be needed to make the 
modifications to fix things at some future date.  I do not advocate 
this, but am trying to make it clear that knowing the problem and 
deciding how to fix it are different, and that we are in a limited 
time environment to reach these decisions.

  -JH




At 10:12 AM +0100 2/14/02, Frank van Harmelen wrote:
>Jim Hendler wrote:
>
>>  Umm, I think the *next* step is to get some people working on some
>>  specific proposals - not to have a general discussion of approach, we
>>  need to have specific proposals on the table to discuss and move
>>  through.
>
>Jim,
>
>I think we are already close to this:
>
>The layering document lays out 4 options.
>
>1.Same-syntax semantic extension:
>2.Syntax and semantic extension
>3.Same-syntax, but diverging semantics
>4.Differing syntax and semantics
>
>Each of options 2-4 could be realised without further ado,
>
>Option 2 would lead to "extraordinarily-complex and difficult 
>patterns of reasoning",
>so is out.
>
>I get the impression option 4 is not really serious (it basically 
>ignores RDFS).
>
>These leaves 1 and 3.
>
>Option 3 has the nice feature that we can use RDF to reason about OWL syntax.
>
>option 1 is very nice but could only be chosen if at least one of 
>the following fixes was made to RDF(S):
>1a move rdf:type to the meta-theory
>1b stratify RDFS
>1c allow for un-asserted triples in RDF
>(and perhaps 1a and 1b are the same if someone could explain it to me)
>
>Looks like the next steps would be to get a sounding from RDF Core on 1a-c.
>If these are all out, we now where we stand.
>If at least one of these could be in, we have to choose between 1 and 3.
>
>Frank.
>    ----


-- 
Professor James Hendler				  hendler@cs.umd.edu
Director, Semantic Web and Agent Technologies	  301-405-2696
Maryland Information and Network Dynamics Lab.	  301-405-6707 (Fax)
AV Williams Building, Univ of Maryland		  College Park, MD 20742
http://www.cs.umd.edu/users/hendler

Received on Wednesday, 13 February 2002 09:42:56 UTC