RE: SUO: Monolithic ontologies (was ontology as science)
Matthew,
I'll have to let Ian respond on the broader points of philosophy, but I
can address a few of the points below:
At 02:01 PM 7/1/2003 +0100, West, Matthew R SITI-ITPSIE wrote:
[snip]
> > >
> > > In the SUMO, change is always represented as a 'Process' from one
> > > 'Attribute' or 'Relation' to another. For example, 'Melting'
> > > (http://ontology.teknowledge.com:8080/rsigma/SKB.jsp?req=SC&na
> > > me=Melting&skb
> > > =SUMO) is a change of state from 'Solid' to 'Liquid', and
> > > 'Translocation'
> > > (http://ontology.teknowledge.com:8080/rsigma/SKB.jsp?req=SC&na
> > > me=Translocati
> > > on&skb=SUMO) is a change from one location to another.
> >
> > MW: This is only half the story. How do you deal with change
> > with respect to the objects changed by the activities?
> > >
The objects would just have a new statement, true after a given time, such as
(holdsDuring
(ImmediateFutureFn
(WhenFn Process777))
(attribute Object333 GreenColor))
> > > >
> > > > But how about:
> > > >
> > > > - what it means to be an individual (process or physical
> > > > object - or both?).
> > >
> > > In the SUMO, to be an instance of 'Process' or 'Object' is to have a
> > > location in space/time.
> >
> > MW: This is close to a 4D statement. Now by location, do you mean
> > something that is extended in space and time, or space but not time,
> > or just a point location?
Here's the relevant axiom
(<=>
(instance ?PHYS Physical)
(exists (?LOC ?TIME)
(and
(located ?PHYS ?LOC)
(time ?PHYS ?TIME))))
Any Physical necessarily has both a physical and temporal location. The
location can be a point or a region.
> > >
> > > >
> > > > - What sorts of things physical quantities are (pressures,
> > > > temperatures etc)
> > >
> > > I don't see that the ontology has to take a stand on what pressures,
> > > temperatures, and other physical quantities *really* are.
> > > All we need to do
> > > is to offer a formalization which allows different people and
> > > different
> > > systems to say all they need to say to communicate with one
> > > another and
> > > perform certain standard inferences.
> >
> > MW: The view you take determines what aparatus you need, and the
> > complexity involved in achieving this.
> >
> > MW: However, lets try a simpler question. Does the membership of
> > classes change?
> > >
Generally, no. Now, this slight hedge may be unsatisfying to some, but
personally, I think it's needed. A stronger statement gets one into all
sorts of contorted boundary cases like the classic "castle and bricks"
example. That entities do not change their class membership over time, and
that classes themselves are reserved for those things which do not change,
is an important guideline.
Adam