Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: SUO: -- Technical Methodology




Dear John,

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.west@shell.com
Internet: http://www.shell.com


> -----Original Message-----
> From: John F. Sowa [mailto:sowa@bestweb.net]
> Sent: 17 June 2003 17:43
> To: West, Matthew R SITI-ITPSIE
> Cc: Patrick Cassidy; standard-upper-ontology@ieee.org
> Subject: Re: SUO: -- Technical Methodology
> 
> 
> Dear Matthew,
> 
> I mostly agree with your recent notes.
> 
> I also believe that a 4D ontology is better for most
> purposes, but I am not willing to say that people who
> find a 3D option important for their application should
> be prevented from choosing it.

MW: I agree. This is why I think your proposal is the best way
forward.
> 
>  > MW: Has an example been given of an activity/event/process
>  > has been demonstrated to be atemporal? not yet, though we
>  > do have a couple of challenges that should be explored and
>  > chased down to determine the answer. Anything else just says
>  > we don't know and we don't care.
> 
> I use Petri nets as an example of a formalism that represents
> events and causal dependencies without assuming any temporal
> representation.  

MW: I am familiar with Petri nets, and find that they are a 
powerful formalism for representing many classes of activity.

> In effect, they take causal dependencies to
> be more fundamental than temporal precedence, and they treat
> a metric for assigning numbers to time points as irrelevant.

MW: Petri nets deal with classes of activity/process rather
than particular ones, so they deal with valid patterns of activity
which will be atemporal.

MW: I also agree that in general indexing some activity to
some temporal frame is quite secondary, and is about measuring
time rather than being a necessary condition for an activity.
The necessary condition is that a particular activity happens
in space time, and therefor has some spatio-temporal location.
This does not have to be measured against some scale.
> 
> Petri nets are especially valuable for representing events
> where "we don't know and we don't care" about a metric time
> or a metric space.  An example is computer programming.
> When we write code that will run on many different machines,
> we have no idea how fast it will run.  On a multiprocessing
> system, we have no idea when or where a process might be
> interrupted or moved to another CPU -- and we don't care.
> (And if the program is running on a laptop on an airplane,
> the movement of the processes through space should be
> totally irrelevant to the results.)
> 
> Therefore, events and processes represented by Petri nets
> are an important example of an "agnostic" approach to
> time and space.  Strictly speaking, those events are not
> atemporal -- they still exist somewhere in time and space,
> but time and space are less relevant than the causal
> connections.
> 
> Bottom line:  A representation of events at an agnostic
> level that assumes causality, but not a metric space or
> time is a common and an important supertype of both a
> 3D and a 4D representation.

MW: I agree. This is certainly how our representations in 
EPISTLE work too.
> 
> John
> 
>