RE: RE: SUO: Re: Ballot Comment
Matthew,
Comments below:
At 10:39 AM 9/3/2001 +0200, West, Matthew R SITI-GREA-UK wrote:
>Dear Adam,
>
>See responses below.
>
>
>Matthew West
>Principal Consultant
>Shell Information Technology International Limited
>Shell Centre, London SE1 7NA, United Kingdom
>
>Tel: +44 20 7934 4490 Other Tel: +44 7796 336538
>Email: matthew.r.west@is.shell.com
>Internet: http://www.shell.com
>snip of headers.
> > > >
> > > > Matthew,
> > > >
> > > > At 10:09 AM 8/29/2001 +0200, West, Matthew R SITI-GREA-UK wrote:
> > > > >Dear Adam,
> > > > >
> > > > >If you don't mind changing the meaning of the concepts involved
> > > > >so that they are 4D rather than 3D concepts, then your
> > axioms would
> > > > >probably work.
> > > >
> > > > Could you explain further what you mean by "don't mind
> > changing the
> > > > meaning..."? The only meaning the terms have is from
> > their axioms.
> > >
> > >MW: I admit I am reading the words rather than the axioms, and it is
> > >possible that the SUMO has not added those axioms that would make it
> > >specifically 3D. The only point is that you can't be both at the
> > >same time, you have to be one or the other, or have each in different
> > >places (what I am proposing).
> >
> > Well, it sounds like we just disagree on whether this is
> > possible. I'm
> > looking for an operational, computational definition of what
> > it would mean
> > that they "can't be both at the same time". The only operational
> > definition I can think of would be the one could prove a
> > contradiction
> > using first order logical inference. Since that's not the
> > case I have to
> > conclude that it is possible unless someone offers another
> > operational
> > definition. But forgive me, I've said as much before, and I
> > know my saying
> > it again doesn't move us along. I wish I knew what would and
> > I appreciate
> > that you're still trying.
>
>MW: There are two issues here.
>
>1. I have not had the time to go through the relevant parts of
>SUMO to see whether there are inconsistencies with 4D, so I cannot
>say with certainty whether there are or not, and if so what they
>are. I estimate it would take me a manday or two to do this.
>
>2. There are basically 3 possibilities for the SUMO.
>
>a) It is a 3D ontology with appropriate constraints that specify
>objects as continuants.
>
>b) It is a 4D ontology with appropriate constraints that specify
>both static and dynamic objects as spatio-temporal extents.
>
>c) It is vaguer than either of these and has no constraints that
>would distinguish 3D from 4D objects.
>
>The third of these is the worst possible case of course, since
>someone's intent of being a 3D object or a 4D object could not
>be discerned or acted upon.
>
>One important difference between 3D and 4D objects is that in
>a 4D world an object is uniquely defined by its spatio-temporal
>extent and nothing else. This is not so for a 3D object. For
>example the clay (substance) a vase is made out of is always
>considered a separate object from the vase itself.
>
>Normally this does not give a contradiction with the 4D view,
>because the spatio-temporal extent of the clay extends beyond the
>spatio-temporal extent of the vase, but they can coincide accidentally,
>and then 3D has 2 objects and 4D has one object. I think a
>requirement for at least 2 objects and at most one object is
>a contradiction.
>
>So if the SUMO doesn't give a contradiction, then there is
>something missing from the SUMO.
I believe that there may be a fourth alternative which is that there is
another theory which can represent our world in a reasonable way that is
neither purely 3d nor 4d and yet which has the expressive or predictive
power of either one.
In fact, after reading Loux's Metaphysics I'm even more convinced that this
issue is one of aesthetics rather than one of truly irreconcilable
views. That each argument has layers of argument and counter argument,
claim and counter-claim tells me that both are completely valid and it is
arbitrary to pick one or the other. If one paradigm were correct and the
other incorrect or unable to represent some practical phenomenon this would
likely already be clear.
I'll try to provide a "book report" on Loux later which I hope will help.
I hope you'll review the actual content of SUMO since although talking in
general terms can be (and has been) helpful in bringing us to common
understanding, the only way really to understand and critique the SUMO
model, is to critique the model itself rather than my informal explanations
of parts of it.
> >
> > > >
> > > >
> > > > > There are some details, like the date/time attributes
> > > > >really represent the events that cause the start and end of the
> > > > >states, but I'm sure we could get there.
> > > >
> > > > Could you explain further here as well? Maybe it's just my
> > > > 3d bias again
> > > > but I don't think of time points as being events that cause
> > > > anything. If
> > > > we need to represent events causing other events then we
> > > > should be relating
> > > > an event and another event, not a timepoint and event.
> > >
> > >MW: I would say that a spatio-temporalExtent is bounded by two
> > >things I would call events, but they are more precisely spatio-
> > >temporal boundaries, the start and end of the spatio-temporal extent
> > >in the time dimension, the state-change. The state change would be
> > >a part of a (physical) point in time, where a physical point in
> > >time is a slice through all of space at (dimensional) point in time.
> > >(spatio-temporal dimensions are really abstractions from the physical
> > >reality).
> >
> > Let me attempt to restate and give an example. Let's say we
> > had the 4d
> > entity of Joe. There is an event BirthOfJoe that "caused"
> > Joe (ignoring
> > the complexities of when life begins for the moment please!).
>
>MW: Just to note that you are using "event" in a different sense to me.
>You are using it in the sense of activity or process, both of which
>are spatio-temporal extents, and not temporal boundaries/state changes.
If you want to call a temporal boundary an event, that's ok (although it
seems a bit odd to me), but you still need two different things in any
ontology - the notion of an activity or process and the bounds of that
process on a time line. You'll also need some notion of causality although
coming up with a strong theory of causality is very hard.
> > There's also
> > an event DeathOfJoe. There are relations that we might call
> > startTime and
> > endTime on entities that relate them to points in time. So,
> > in this example
> >
> > (startTime 1980 Joe)
> > (endTime 1980 BirthOfJoe)
> > (endTime 2000 Joe)
> > (causes BirthOfJoe Joe)
> > etc
> >
> > I think this is still consistent with a 4d ontology, but let
> > me know if
> > not.
>
>MW: Well not really. I would say something like:
>
>(activity birthOfJoe)
>(temporalBoundary startOfJoe)
>(startTemporalBoundaryOf startOfJoe Joe)
>(cause birthOfJoe startOfJoe)
>(spatialPartOf startOfJoe 1980)
>etc
You've changed the names of the relations, and chosen to name the temporal
bounds rather than assign them points on a real time line.
>Note that in particular an activity does not cause the whole
>object, only the state change (temporal boundary) that brings
>it into being. It is not (necessarily) responsible for any
>aspect of the object after that.
Why doesn't the birth of Joe cause Joe himself as opposed to just his
beginning temporal bound? Note that in this particular model of the world,
we're not modeling belief at one time without knowledge of the
future. Even if Joe dies at age one day, isn't it true that (causes
BirthOfJoe JoesOneDayOfLife)?
> > My point is that the relations endTime and causes are
> > both distinct
> > and needed.
>
>MW: No, and yes (but not as stated by you).
Let me restate then with more general terms:
A predicate relating an activity or process to its temporal bounds as well
as a predicate relating an activity to the activity or process which caused
it are needed.
> >
> > > >
> > > > >But then it wouldn't say the 3D things you probably want it
> > > > to say as well.
> > > >
> > > > Well, I'd be happy with SUMO + spatioTemporalPart.
> > >
> > >MW: spatioTemporalExtent is better than part. There is at least one
> > >spatio-temporal extent that is not part of anything (all the universe
> > >for all time).
> >
> > So, have we converged? If we add the relation
> > spatioTemporalExtent to
> > SUMO, consistent with my earlier example usage of that
> > relation, are you
> > happy with SUMO as far as its ability to represent the 4d
> > view that you want?
>
>MW: I think spatioTemporalExtent is a predicate (class) e.g.
>(spatioTemporalExtent Fred)
>rather than a relation, but maybe that's what you mean.
No, I mean as in my previous message that spatioTemporalExtent would be a
predicate that relates spatio-temporal (4d) entities to their
spatio-temporal parts. So
(spatioTemporalExtent JoesLife JoeOnHis20thBirthday)
Adam
>MW: Otherwise, yes, introducing spatioTemporalExtent as a
>subtype of entity provides one of the necessary start points
>for being able to represent a 4D viewpoint.
>
> >
> > > > I think
> > > > that would
> > > > still enable us to state things in a 3d way but also allow 4d
> > > > expressions.
> > >
> > >MW: Precisely my intention too.
> > > >
> > > > >The point is that 3D and 4D concepts are not coincident,
> > > > even where you
> > > > >might naturally choose the same name for each. Recall
> > that I had to
> > > > >translate
> > > > >you 3D statements into 4D statements before I could answer
> > > > your question.
> > > > >
> > > > >Thank you for sticking with this.
> > > >
> > > > no problem. If only we could make the tone of this list more
> > > > friendly I
> > > > think all discussion of this sort would be fun and
> > > > enlightening. I look
> > > > forward to more.
> > >
> > >MW: I prefer it this way too.
> > > >
> > > > Adam
> > > >
> > > >
> > > >
> > > > >Matthew West
> > > > >Principal Consultant
> > > > >Shell Information Technology International Limited
> > > > >Shell Centre, London SE1 7NA, United Kingdom
> > > > >
> > > > >Tel: +44 20 7934 4490 Other Tel: +44 7796 336538
> > > > >Email: matthew.r.west@is.shell.com
> > > > >Internet: http://www.shell.com
> > > > >
> > > > >
> > > > > > -----Original Message-----
> > > > > > From: Adam Pease [mailto:apease@ks.teknowledge.com]
> > > > > > Sent: 27 August 2001 15:41
> > > > > > To: West, Matthew R SITI-GREA-UK; 'pat hayes'
> > > > > > Cc: standard-upper-ontology@ieee.org
> > > > > > Subject: RE: RE: SUO: Re: Ballot Comment
> > > > > >
> > > > > >
> > > > > > Matthew,
> > > > > > Did you have any further comments on the logical
> > > > representation I
> > > > > > presented in my message? That was the actual "meat" of the
> > > > > > message and
> > > > > > where I think we have hope of some progress. I may
> > be the eternal
> > > > > > optimist, but I think I found a solution that supports the 4d
> > > > > > view that you
> > > > > > want. I've copied that portion below.
> > > > > >
> > > > > > Adam
> > > > > >
> > > > > > >If I understand you correctly, then you may be closer to the
> > > > > > current SUMO
> > > > > > >representation than I thought. Here's a
> > formalization in SUMO:
> > > > > > >
> > > > > > >(instance JoesArm OrganicObject)
> > > > > > >(instance JoesArmAtAge10 OrganicObject)
> > > > > > >(instance JoesArmAtAge20 OrganicObject)
> > > > > > >(spatioTemporalPart JoesArmAtAge10 JoesArm)
> > > > > > >(spatioTemporalPart JoesArmAtAge20 JoesArm)
> > > > > > >(equal (BeginFn JoesArmAtAge10) (BeginFn (YearFn 1990)))
> > > > > > >(equal (EndFn JoesArmAtAge10) (EndFn (YearFn 1990)))
> > > > > > >(equal (BeginFn JoesArmAtAge20) (BeginFn (YearFn 2000)))
> > > > > > >(equal (EndFn JoesArmAtAge20) (EndFn (YearFn 2000)))
> > > > > > >
> > > > > > >The key then is in the definition of a new relation for SUMO
> > > > > > which we
> > > > > > >could call 'spatioTemporalPart' which would allow us
> > to deduce
> > > > > > >
> > > > > > >(not (equal JoesArmAtAge10 JoesArmAtAge20))
> > > > > > >
> > > > > > >Now, at the danger of pointing out a flaw in this, we'd have
> > > > > > to assume
> > > > > > >that no one who wants to use a 3d viewpoint would ever
> > > > want to reify
> > > > > > >temporal sub-parts. Because from the above formalization,
> > > > > > they'd want to
> > > > > > >conclude
> > > > > > >
> > > > > > >(equal JoesArmAtAge10 JoesArmAtAge20)
> > > > > > >
> > > > > > >I'm hoping though that this might be an acceptable
> > > > > > compromise in terms of
> > > > > > >practical modelling support.
> > > > > >
> > > > > > At 02:42 PM 8/26/2001 +0200, West, Matthew R
> > SITI-GREA-UK wrote:
> > > > > > >Dear Adam,
> > > > > > >
> > > > > > >See comments below.
> > > > > > >
> > > > > > >
> > > > > > >Matthew West
> > > > > > >Principal Consultant
> > > > > > >Shell Information Technology International Limited
> > > > > > >Shell Centre, London SE1 7NA, United Kingdom
> > > > > > >
> > > > > > >Tel: +44 20 7934 4490 Other Tel: +44 7796 336538
> > > > > > >Email: matthew.r.west@is.shell.com
> > > > > > >Internet: http://www.shell.com
> > > > > > >
> > > > > > >snip
> > > > > > > > > >
> > > > > > > > > > Matthew,
> > > > > > > > > > Ian's just pointed out to me problems with
> > my response.
> > > > > > > > > > SUMO could
> > > > > > > > > > handle a 4d view but only in a trivial way by
> > a knowledge
> > > > > > > > > > engineer putting
> > > > > > > > > > everything he cares about under the notion of
> > Process and
> > > > > > > > > > thereby throwing
> > > > > > > > > > out the existing SUMO notions of identity that work on
> > > > > > > > > > Object. That's not
> > > > > > > > > > a good solution. I can see merit in both 3d and
> > > > 4d views and
> > > > > > > > > > the challenge
> > > > > > > > > > is how to let them coexist in a meaningful way.
> > > > > > > > >
> > > > > > > > >MW: Unfortunately, you need to start probably
> > above object.
> > > > > > > > >4D classes have unchanging extents, for example.
> > > > > > > >
> > > > > > > > I'm referring to Object in its sense in SUMO. The current
> > > > > > > > hierarchy is
> > > > > > > >
> > > > > > > > Entity
> > > > > > > > Physical
> > > > > > > > Object
> > > > > > > > Abstract
> > > > > > > >
> > > > > > > > etc.
> > > > > > > >
> > > > > > > > Do you feel a change is needed to axioms about Physical
> > > > > > > >
> > > > <http://ontology.teknowledge.com:8080/rsigma/SKB.jsp?req=SC&sk>
> > > > > > > b=Merge-WordNet&id=95>?
> > > > > > >
> > > > > > >MW: Physical is close to the 4D concept of spatio-temporal
> > > > > > extent. However,
> > > > > > >a 4D view would be that Physical Entities are parts of space
> > > > > > time, not
> > > > > > >located
> > > > > > >in it. I presume location is meant in some co-ordinate
> > > > > > sense, which is an
> > > > > > >abstraction. Before you can have an abstraction, you need to
> > > > > > have something
> > > > > > >that forms the basis of abstraction, these are
> > > > > > spatio-temporal extents,
> > > > > > >including
> > > > > > >one which is all of space-time (past, present, and future).
> > > > > > >
> > > > > > >MW: Existant has the concept of existing at one or more
> > > > > > points in time. It
> > > > > > >is
> > > > > > >fundamental to a 4D view that existance is for a period of
> > > > > > time. This is one
> > > > > > >of the
> > > > > > >ways in which it differs significantly from a continuant
> > > > > > based view which
> > > > > > >the SUMO
> > > > > > >has built in at a high level.
> > > > > > >
> > > > > > >MW: Also, a spatio-temporal extent can be quite arbitrary,
> > > > > > rather as a set
> > > > > > >can have
> > > > > > >arbitrary members.
> > > > > > >
> > > > > > >MW: Another problem is that many of the things that you have
> > > > > > as abstract,
> > > > > > >like roles,
> > > > > > >are temporal parts of things in a 4D ontology. I would
> > > > > > therefore think it
> > > > > > >was safer
> > > > > > >to start above physical.
> > > > > > >
> > > > > > >MW: There are also problems with the abstract concepts. Our
> > > > > > classes would be
> > > > > > >extensional and
> > > > > > >unchanging. Properties and qualitities would be classes, and
> > > > > > attributes
> > > > > > >might be either
> > > > > > >classes or spatio-temporal extents.
> > > > > > >
> > > > > > >MW: I.e. all abstract concepts would be classes/sets.
> > > > > > >
> > > > > > >MW: The best bet is to start with Chris M's Basic Set Theory
> > > > > > and try to
> > > > > > >build something
> > > > > > >from there. However, it would be important to have a
> > > > > > partition and probably
> > > > > > >a naming
> > > > > > >convetion so that 4D and 3D concepts can be distinguished.
> > > > > > >
> > > > > > >snip
> > > > > >
> > > > > > Adam Pease
> > > > > > Teknowledge
> > > > > > (650) 424-0500 x571
> > > > > >
> > > >
> > > > Adam Pease
> > > > Teknowledge
> > > > (650) 424-0500 x571
> > > >
> >
> > Adam Pease
> > Teknowledge
> > (650) 424-0500 x571
> >
Adam Pease
Teknowledge
(650) 424-0500 x571