Pat, (01)
> > ...
> > The reason I asked is that I have worked with Matthew (and David
> > Leal) on
> > and off over the last decade on a general framework for UoMs -
> > sometimes as
> > part of Matthew's ISO 19526 exercise (Matthew and David have been
> > working on
> > the UoM issue far more consistently and for much longer). What we
> > found is
> > that the choice of 3D/4D and extensionalism had a big effect on how we
> > crafted our ontology, including property instances.
>
> I find that interesting, and would like to have been a fly on the wall
> during your discussions. Is there any written trace of them? (02)
Matthew or David may have them. In one sense, the ISO 19526 revisions are a
written trace, but may not be what you are after.
I suggest you ask Matthew or David - it was Matthew's comment that sparked
this off. (03)
>
> > I think this is why
> > Matthew made his comment. For a simple example, see my earlier
> > response to
> > Ingvar, who nicely provided us with an example.
>
> Can you repeat that example? I seem to have missed it. (04)
Ingvar, (05)
Take two spatiotemporal points in a four-dimensional
> manifold,
> ascribe to both exactly the same mass density, then you have (on my
> view)
> two 4D-points, one property, and two property insstances. (06)
For some 4D extensionalists, you only have two 4D-points (also called
property instances) and one property.
So we have two different (mainstream) views to reconcile. (07)
>
> Ingvar J (08)
>
> > I am not aware of any definitive proof that property instances, 3D/
> > 4D and
> > extensionalism are either independent or dependent. It does seem as
> > though
> > 4d + extensionalism (for which Matthew sometimes uses the shorthand
> > '4d')
> > does have implications for property instances. However, these things
> > are so
> > entangled that there is always wriggle room. I interpreted your and
> > John's
> > comment that you had some way of showing that they were independent
> > simpliciter. So, to be clear I have no way of showing these are
> > orthogonal
> > or not and would be interested in any such proof as my experience is
> > that
> > these matters are entangled - so some such proof would be useful.
>
> I will try to sketch the argument, though I would hesitate to call it
> a proof. It amounts to defining a syntactic transformation from one
> ontological formalism style to another, and then observing that one of
> these can reasonably be seen as a formalism of the 4D picture, and the
> other of the 3D one. The basic point is that there are only two ways
> to insert a time parameter into a 'static' description: it either gets
> attached to a relation name as an argument, or to an individual name,
> converting them to a term. One writes either (R a b t) or (R (a t)(b
> t)). The first is naturally read as a continuant-style description,
> using time-dependent fluents to talk of things whose identity is
> considered to be timeless, the latter as a 4D one, in which assertions
> are made timelessly about 'slices' of a 4D entity. (There are other
> ways, but they all involve changing the logic in some way, eg by
> adding temporal contexts or tenses.) Basically, and oversimplifying
> things greatly, one can convert from 3D to 4D and back by 'moving' the
> temporal parameter between these two positions. This is an observation
> about the formal notation, of course, but the philosophy is basically
> irrelevant until it gets cashed out in a formal expression of some
> kind. Whatever your views are on property instances, in particular,
> can (if they are relevant to the final ontology) be ultimately
> expressed in one of these ways, and when it is so expressed, can be
> almost mechanically transliterated into the other, and from there
> 'read back' as being about the other kind of temporal entities. I
> realize that this hardly constitutes a philosophically valid form of
> argumentation, but then of course we are not here doing philosophy,
> but rather ontology engineering: right?
> (09)
Can I make an number of points.
Firstly, can I say that I think Matthew's original comment was not about
whether one can do this translation or not - and I am not sure how having
the translation in itself proves (or disproves) independence - though the
details may be illuminating. I think my comments below will help to make
this point clear.
Secondly, I cannot see what practical relevance whether one has this
translation makes to the current enterprise. So I am puzzled why people are
making such a big deal of it. (010)
Let me quickly state my position wrt the translation. I believe that there
are significant problems in formalising the position with sufficient
accuracy to make it useful. While one may be able to easily translate 80% or
88% or 99%, this is not good enough. I suspect that when/if this has been
done, there will be awkward problems with the translation, which will become
significant when one deals at the industrial level. So if one were to offer
a 3d/4d translation service, one would be expected that it would work with
an accuracy at the (say) five of six nines level on huge quantities of data.
My gut feel is that people would sensibly much rather rely on rule of thumb
and experience than theoretical rules. My gut feel is that why it has not
yet been done. But I have no proof at all of this intuition. (011)
I can see the outline of your argument, but for me the flaw is "the
philosophy is basically irrelevant until it gets cashed out in a formal
expression of some kind." I would agree that once one has grasped what needs
to be formalised, then one can work with the formalisation - the problem is
getting to the grasping. (012)
It also seems to me to miss the key problem - and this may be at the heart
of our disagreements - and it may arise because we deal with different
problems in our work.
Let us assume that (a) we have formalised the 3d and 4d view and (b) we have
a formal translation between the two (so these can be discounted as issues). (013)
We approach a domain and we find there are a number of communities and
sub-communities and so already number of different sets of terms. We start
by translating the different sets into formalisms.
We soon realise a number of things.
1) That the different sets capture well different aspect of the world (the
ones that were uppermost in the minds of people using them) - and other
aspects less well.
2) That due to historical and other factors, these sets are internally
heterogeneous with respect to various principles including 3d/4d, identity,
etc. - and this is one reason what there are sets of different terms.
3) That in many cases it is not completely clear what the terms refer to -
and that the experts are not able to easily resolve the issues in a way that
enables them to be harmonised. (014)
So, let's say we decide we would like to arrive at a formalism of these sets
- to agree on a basic set of terms. And we consider two options
Option (a) We take all the terms as they stand
Option (b) we agree on some common framework within which we regiment the
different terms (015)
It is my experience (b) is a much better option, particularly if you are
working in a team. People have a framework to work within.
It is also my experience that the 3d/4d choice is one of the useful elements
of the common framework - and that it is more important to make the choice,
than which choice is made. (016)
So it seems to me you are seriously discounting the effort it takes to
formalise things in such a way that they can be used at an industrial level
- and focusing on what happens when that task is done. Perhaps, this
reflects our different concerns. I just think it is very difficult when one
deals with large industrial domains, to have a sharp enough understanding to
produce a formalism that is useful, that is an improvement on the ad hoc
situation that exists. And if one does not, the attempt at formalism seems
like a theoretical nicety with no real practical benefits.
What the common framework does is to increase the odds of success. (017)
> >
> > I am not sure how to interpret your comments - you say that "The
> > differences
> > between them, I have concluded, are really nothing more than a
> > matter of
> > notational choice." In these terms, the issue becomes whether the
> > notation
> > choice of 3d/4d, extensionalism/intensionalism and property
> > instances are
> > independent. I still see no answer.
> >
> >>> However, you have said that this is obvious, so I presume you have a
> >>> way of
> >>> showing it.
> >>> If you (or John) can explain, I'd be interested.
> >>>
> >>>>
> >>>>>
> >>>>>>>
> >>>>>>>> In other discussions, you have consistently argued for an
> >>>>>>>> ontology
> >>>>>>>> that is strictly extensional. That is also an assumption that
> >> is
> >>>>>>>> orthogonal to the 3-D vs. 4-D viewpoint.
> >>>>>>>>
> >>>>>>>
> >>>>>>> Are you sure orthogonal is the right word here?
> >>>>>>> As far as I can tell (and others say the same) the views are
> >>>>>>> interestingly
> >>>>>>> intertwined. So, for example, it is rather difficult to be an
> >>>>>>> extensionalist
> >>>>>>> and not be 4D. Orthogonal seems to imply that there is no kind
> >>>>>>> of
> >>>>>>> dependence
> >>>>>>> between the two.
> >>>>>>
> >>>>>> Indeed; and there is not. Of course it is possible to be a 3-d
> >>>>>> extensionalist. You believe in continuants, and you say that two
> >>>>>> continuants which share the same 3-d extension at every point of
> >>>>>> their
> >>>>>> common life, are identical.
> >>>>>
> >>>>> Not quite the traditional definition of extensionalism. But
> >>>>> illustrates my
> >>>>> point that it is difficult - but not impossible.
> >>>>
> >>>> There is nothing difficult about this, surely. It is easy to be an
> >>>> extensionalist and not be 4D, is the point. The relevant form of
> >>>> extensionalism has to be stated in 3D terms, of course.
> >>>
> >>> The difficulty is that of sticking to the traditional way of framing
> >>> extensional identity where a single extension is used.
> >>
> >> But that is applied in a strictly presentist framework, where the
> >> universe is conceptualized in three dimensions at a moment in time.
> >> Obviously this must be adapted when we consider temporally extended
> >> universes.
> >
> > I do not understand your comment. As has been pointed out a number
> > of times,
> > 3d/4d is shorthand for positions on the reality of change.
>
> Um... that is not how I am understanding it. The *reality* of change
> seems to me to be beyond rational dispute. I see the distinction as
> being about how best to formally describe change. (018)
I think we have discussed this before.
There are lots of reference to the unreality of change, series A and B, etc.
which I raised before - so I will try a different tack. (019)
Consider a situation where PatH is in his office at time t1, and in the
garden at time t2.
We can all agree that, in one sense, there has been a change, but .... (020)
A 4d-ist (or whatever) says that the time-slice of PatH at t1 is inside his
(4d) office - and the time-slice of PatH at t2 is inside his (4d) garden.
Ok, there is no object here that is changing, the same object that is
different in some respect - and no object that is a change. We have what
Russell called an at-at description of change (at-t1-at-t2). (021)
A 3d-ist (or whatever) says at time t3, that at t1 PatH was inside his
office - and at t2 PatH was inside his garden - and that between t1 and t2
PatH's position changed. Change consists of (exactly) the same object being
different at different times. And one can talk of this change. And these
changes are re-ified as occurents. (022)
Now what it seems to me you do is to give a 4d interpretation of the 3d view
- typically based upon time-slices. Nothing wrong with that (I think it is
in many ways sensible), unless you deny there is such a thing as the 3d view
that the 3d-ist (presentist) holds. (023)
If this does not work, shall we just agree to differ? (024)
>
>
> > For an
> > extensionalist the matter is solved. What is odd (for me) about your
> > position is that you do not seem to see any need to a criteria of
> > identity,
> > while arbitrary applying your own intuitive judgements on this.
>
> I do not think it is useful to engage here in a philosophical debate
> about the merits or otherwise of identity criteria. One thing is
> clear, that different people and communities will use many different
> such criteria for many different, but to them compelling, reasons. (025)
My experience is that they do not have an systematic criteria. That is a
common problem I face. (026)
It
> is therefore not appropriate or useful to set out to impose any one
> such discipline on the entire world of ontology users, no matter how
> convincing one may feel that one's philosophical credentials are. (027)
Agreed, and I do not believe I have done that. And I am not sure where
philosophical credentials come in.
Similar points apply to the use of notation, logic, etc. (028)
IMO,
> the responsible ontology engineering position should be to make these
> explicit and try to find accurate ways to map between them or make
> them at least mutually comprehensible, but not to make judgements
> about their correctness or otherwise, since those decisions are
> determined by criteria which lie outside the purview of ontology
> engineering, (029)
I am not sure you can say a priori that just because a criteria is studied
in philosophy it is not relevant to ontological engineering. While academics
may wish to regard their departmental divides as insurmountable, the world
is not divided in the same way. How can one exclude the possibility that,
for the purposes of ontological engineering, some criteria studied in the
philosophy department may (or may not) be useful? (BTW in the history of
information technology, one theme is that as various technologies have
developed the notion of identity has refined - so there is some historical
evidence that tightening up the notion of identity was linked to advances in
information technology). (030)
such as in the history of a science or the practice of a
> community. Hence my deliberate avoidance of this topic in these email
> threads. What I think we can usefully do, perhaps, is to help people
> state their identity criteria more clearly and explicitly; but even
> there, my experience has been that they often have quite a clear idea,
> and the apparent confusion arises from it not fitting into some
> philosophical category. (031)
Interesting, my experience is that even before one starts digging, fault
lines appear in the identity criteria. I suppose you would regard any
attempt to analyse the identity criteria as pointless, but my experience is
that where any criteria exists it is extremely heterogeneous. But I suppose
we just have to note our differing experience. (032)
As I hoped I made plain above, my desire here is not to impose any
particular view on a global basis. But to suggest that it makes sense to
agree what view one is using in a particular project and stick to it, and if
a one view has practical advantages over another to take account of this.
Much as one might agree to use Java to code a particular system, or
intuitionistic logic to formalise something. Or, as John keeps suggesting CL
for this ontology. (033)
>
> Pat
> (034)
Chris (035)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/uom-ontology-std/
Subscribe: mailto:uom-ontology-std-join@xxxxxxxxxxxxxxxx
Config/Unsubscribe: http://ontolog.cim3.net/mailman/listinfo/uom-ontology-std/
Shared Files: http://ontolog.cim3.net/file/work/UoM/
Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?UoM_Ontology_Standard (036)
|