Re: DAML+OIL and OIL

>>>>> "Tim" == Tim Berners-Lee <timbl@w3.org> writes:

    Tim> I think that actually in the big wide wold, you can't make
    Tim> that distinction so simply.  What you can do in the web, when
    Tim> the identifier for something is in a URI space which has
    Tim> concepts of ownership and dereference (such as http:), is to
    Tim> dereference the term.  If this can be done (not at all always
    Tim> the case), then you know from the http transaction that you
    Tim> have information about the thing which was authorized
    Tim> directly or indirectly by the owner of the identifier.  In
    Tim> that sense such informatoin is definitive.  So anything you
    Tim> learn about http://www.w3.org/People#danc in a document
    Tim> provided by the www.w3.org service as a represnetation of
    Tim> http://www.w3.org/People can be taken as definitive.

    Tim> However, I don't see myself why there is any difference
    Tim> beween the "definition" information or other infomation about
    Tim> danc in that file.  The same inferences can be made from
    Tim> either and they have the same level of trust.

    Tim> Across the web, you may in fact trust some documents which
    Tim> are not the definitive ones for a term more than you trust
    Tim> the definitive ones, so the idea that one bit of informatoin
    Tim> about what identifier refers to is the "definition" doesn't
    Tim> in the end mean anything.

Tim -

I agree with you in this context. If the only thing that happened with
DAML+OIL was that agents used the ontologies to infer information
about resources, then there would be no problem.

    Tim> What practical use do you need to make of such a distinction,
    Tim> apart from to preserve it for its own sake?

My concern here is with the activity of building those ontologies that
will then be used across the web.

    Tim> You express a concern that in RDF there is no distinction
    Tim> between information about something and its definition.  This
    Tim> is a rather philosophical question.

Sure, but if I'm building a model (or ontology) of how I think a
particular piece of the world fits together I can make a distinction
between axioms and definitions. When building (and exchanging)
ontologies, I believe it *is* useful to preserve the distinction, as
this is part of the conceptual model that you're building. Although
the two collections of statements (a trivial example I do admit):

dog => (animal AND hasTeeth Big)

dog => animal
dog => hasTeeth Big

may have exactly the same semantics, the different organisations of
the information may help the modeller to understand how it all fits
together. 
 
As I said in my earlier message, if necessary, I can elect to add
extra information to my DAML+OIL in order to represent this additional
structure, but that does require extensions that are then proprietary
in some way. The "three roots of OIL" (frames, DLs, web languages)
made it very attractive as a modelling and exchange language, with the
frame modelling primitives backed up by a well defined semantics. To
my mind, losing some of that structure means that DAML+OIL is fine for
deployment of ontologies across the web, but will not provide such a
good platform for the modelling process.

Of course, that may not be what DAML+OIL is intended for, in which
case I needn't worry :-), but it's my suspicion that it *will* end up
as a mechanism for exchange of ontologies as well as deployment, in
which case I will worry :-(.

Cheers,

	Sean

==========================================================================
| Sean Bechhofer               |                                         |
| Research Fellow              |                                         |
| Information Management Group |                                         |
| Computer Science Department  |                                         |
| The University of Manchester | email: seanb@cs.man.ac.uk               |
| Oxford Road                  | Tel: +44-161-275-6145                   |
| Manchester M13 9PL           | Fax: +44-161-275-6236                   |
|------------------------------------------------------------------------|
| WWW: http://www.cs.man.ac.uk/~seanb                                    |
==========================================================================

Received on Wednesday, 7 March 2001 04:35:37 UTC