SUO: RE: A proposed SUO content outline
Pat,
Comments follow.
-Ian
> -----Original Message-----
> From: pat hayes [mailto:phayes@ai.uwf.edu]
> Sent: Wednesday, February 28, 2001 5:55 PM
> To: standard-upper-ontology@ieee.org
> Subject: Re: SUO: RE: RE: A proposed SUO content outline
>
>
>
> "West, Matthew MR SSI-GREA-UK" <Matthew.R.West@is.shell.com>,
> to Ian Niles:
>
> > > On the basis of Pat Hayes' and Chris Partridge's
> comments, the classes
> > > 'Occurrent' and 'Continuant' were replaced with the classes
> > > 'OccurrentType'
> > > and 'ContinuantType', which are classes of classes of
> occurrents and
> > > continuants, respectively. This change was motivated by the
> > > fact that the
> > > distinction between occurrent and continuant is more stable
> > > with respect to
> > > type descriptions than with respect to individual things
> > > themselves. Note
> > > that 'OccurrentType' and 'ContinuantType' are both immediate
> > > subclasses of
> > > 'Entity' in the merged ontology.
> >
> >MW: This does nothing to help. It only means that instances
> of those classes
> >have no members in any part of the ontology I have presented
> an outline for,
> >since the members of members of continuent type must be continuents.
>
> I agree.
I don't understand why you're agreeing here, Pat, since you outline below a
way of interpreting continuants that is consistent with Matthew's
ontological framework. That confusion aside, the larger issue, it seems to
me, is that the whole point of replacing 'Continuant'/'Occurrent' with
'ContinuantType'/'OccurrentType' is that the latter distinction doesn't
require us to ascribe these properties directly to individuals in the world.
For example, given the new dichotomy, 'Process' is an instance of
'OccurrentType' and 'Object' is an instance of 'ContinuantType'. Some
things that are both process-like and object-like (e.g. clouds) could be
instances of both 'Process' and 'Object'.
>
> >MW: You may recall that I have said a couple of times that
> at least the 4D
> >ontology I am working on should be mapped to and not
> combined with the
> >workyou are doing. The reason is that in a 4D ontology there are no
> >individuals that are continuents. That is because the basis
> for what an
> >object is is different. On the other hand, from a 4D
> individual, you will
> >always be able to infer an occurrent (it might even be the
> same object - I'm
> >not sure)
>
> I think it is more likely to be the lifetime of the object.
>
> >and sometimes you will be able to infer a continuent, but this
> >will definitely not be the same thing as the object from which the
> >continuents existence is inferred.
>
> <snip>
>
> > >
> > > (documentation Object "A Physical Continuant which retains
> > > its identity over
> >
> >MW: Continuant does not compute in "my" ontology.
>
> Matthew, I think you could get continuants into a 4-d ontology by
> thinking of them as a certain class of temporal cross-sections of a
> 4-d history. Given a history, consider the set of all spacelike (ie
> perpendicular to the time-axis) 'slices' of it: that set always
> exists, and it is 1:1 with the history but not identical to it. For
> histories of continuants, the things in this set will 'be' the
> continuant at the various times in its lifetime (to speak for a
> moment in continuant-talk) ; for other histories they won't, of
> course, but they might still be worth allowing to exist. The
> characteristically 'continuing' nature of continuants, to which those
> wedded to that way of thinking attach so much importance, may be
> phrased in terms of what Nicola calls identity criteria on the things
> in these slice-sets. For example, from the history (lifetime) of a
> person, this would construct the (single) person at each moment in
> that person's life. Each of these would 'be' that (single) person at
> a particular time, so if we were to (re)construct a temporal or modal
> logic by thinking of times - temporally possible worlds - as slices
> through the 4-d universe, then these things would be the (individual)
> continuant in each temporally possible world.
>
> This move would allow a 4-d ontology to accomodate
> continuant/occurrent talk quite naturally, I believe, even though it
> would not be a mapping that those who like that way of talking would
> approve of.
>
> > >
> > > some interval of time. Although no physical entity is ever
> > > permanent, an
> > > object
> > > can have stable properties over its lifespan. The type Object
> > > corresponds
> > > roughtly
> > > to the class of ordinary physical objects.")
> >
> >MW: Anything that has anything to do with continuant might
> have a related
> >object in a 4D ontology, but it would not be the same
> abject. Thus all the
> >classes that you are talking about are different classes
> because they have
> >continuants as members and not 4D object, so they have different
> >memberships.
>
> Indeed, these types would have to be taken apart more carefully in a
> full 4-d ontology.
>
> <snip>
> >
> > > > > >
> > > > > > Temporal Boundary
> > > > > >
> > > > > > Contains axioms and relations that apply to all individuals
> > > > > > that have a zero temporal extent.
> > > > >
> > > > > It sounds like "Temporal Boundary" corresponds to the class
> > > > > of 'Abstract',
> > > > > another direct subclass of 'Entity'. Some axioms regarding
> > > > > 'Abstract' can
> > > > > be found in the section "General Axioms" of the
> merged ontology.
> > > >
> > > > MW: No. These are things that exist in space time, so
> they cannot be
> > > > classes. They just have a zero extent in the time dimension,
> > > > they are at a
> > > > point in time.
> > >
> > > OK, so the items in this category would be indivisible slices
> > > of space-time
> > > worms?
> >
> >MW: In the sense that anything that has zero thickness is
> indivisible.
>
> Yes, but be careful: there might be indivisible things which do not
> have zero thickness. For example there might be an nonzero atomic
> time-duration ('planck time'?) . Many temporal databases use
> date/time stamping which makes this assumption implicitly. I discuss
> this issue in the 'time catalog'.
>
> >On
> >the other hand it is not indivisible spatially. Further the
> objects that
> >they are slices of do not have to be contiguous either spatially or
> >temporally.
>
> Ah, then I withdraw my earlier suggestion that they be thought of as
> spacelike boundaries: they must be collections of parts of spacelike
> boundaries. Interestingly, I have found a need for a purely spatial
> notion like this (a set of pieces of surface all in some relation to
> one another): it is precisely the general notion of a shape; so one
> could characterize your notion as the temporal equivalent of a 3-d
> surface shape.
>
> <snip>
> > > > > Currently, we have 'Proposition' as a direct subclass of
> > > > > 'Abstract' in the
> > > > > merged ontology. However, as Pat Hayes has pointed out, it
> > > > > is likely that
> > > > > the composite notion of 'ContentBearingThing' will prove to
> > > > > be more useful.
> > > > > In any case, axioms about these notions have yet to
> be written.
> > > >
> > > > MW: I don't think proposition is the same as information
> > > > pattern, unless you
> > > > think your DNA is a proposition.
> > >
> > > I agree, and the interesting example of a DNA strand fits
> > > into the current
> > > structure of the merged ontology. 'ContentBearingThing'
> is a direct
> > > subclass of 'Abstract' (not 'Proposition') and 'Object'.
> Hence, by
> > > subclassing DNA somewhere under 'ContentBearingThing' we
> capture its
> > > informational quality, without claiming that it states a
> proposition.
> >
> >MW: That's interesting. So a piece of paper with writing on
> it is abstract.
> >Now I always thought you could pick a piece of paper up?
> >
> >MW: I think you have a problem. A piece of paper is either
> abstract or not.
>
> Right. I hadnt noticed this before: but obviously one cannot simply
> put ContentBearingThing both under Abstract *and* under Object, since
> these categories are mutually exclusive.
Well, as I pointed out to Matthew before, we could just remove the
disjointness condition on 'Abstract' and 'Physical'. This condition is
probably just a hold-over from Plato's days and/or a product of an obsessive
concern that you philosophers have about not allowing the real world to
contaminate your realms of pure thought. In any case, perhaps the best way
of resolving the problem is to make 'ContentBearingObject' a subclass of
'Object' and then to write an axiom stating that every
'ContentBearingObject' has a part which is an instance of 'Abstract'.
>
> Ian, I think you are wanting a semi-lattice, but you are trying to
> make one out of a tree. That won't work. Let me repeat a query I made
> a while ago: what are the basic inheritance rules of your overall
> structure? It seemed to be a classification tree, but you are no
> longer using it in a way consistent with its being a tree. So what is
> it? If it is a semilattice then you need to say both what is under
> what and *also* what is disjoint with what, and do so explicitly. But
> I would urge that you consider the approach of formal concept
> analysis http://php.indiana.edu/~upriss/fca/fca.html also well
> described by John Sowa
> http://www.bestweb.net/~sowa/misc/mathw.htm#Lattice which allows a
> considerable degree of freedom to assign concepts, and will compute
> the 'smallest' appropriate inheritance structure automatically.
Right, there are cases of multiple inheritance in the merged ontology, so it
has the structure of a lattice rather than of a tree. Furthermore, I have
been trying to state as many disjointness assertions as possible in the
merged ontology, but probably I haven't been as careful about these as I
have been about the subclass assertions. In any case, thanks for the
pointers to the stuff on formal concept analysis.
>
> Pat Hayes
>
> ---------------------------------------------------------------------
> IHMC (850)434 8903 home
> 40 South Alcaniz St. (850)202 4416 office
> Pensacola, FL 32501 (850)202 4440 fax
> phayes@ai.uwf.edu
> http://www.coginst.uwf.edu/~phayes
>
>