RE: SUO: RE: CYC event vs. SUMO Process -- really different?

["Chris Partridge" <chris_partridge@csi.com>]

Title: RE: SUO: RE: CYC event vs. SUMO Process -- really different?

Richard,

 

I agree about the (non-location) temporal relations and overspecification points.

 

But isn't there an argument that goes something like this.

 

If we were able to determine that events were such that they did always have a specific temporal extent then we could specify this in our ontology and also say that one can (and usually did) not need to know this. The advantage is that the variablility between different uses of events would be down to the epistemology (what is known) and the underlying ontology would be invariable.

 

If a new application arose where it was vital to always know the specific temporal extent, then this would already be accomodated in our ontology.

 

A minor, picky, point:

Rich>But we can say that the contract ends at a specified point in time.

 

I believe we cannot say this - and this is an important aspect of (intentional) agreements. Even though there may be a contract specifying that settlement should take place at a specified point in time (when the contract would end), there are cases where the contract is disputed and settlement waits until the legals are dealt with - which in modern law courts can be quite a long time. However, it would seem reasonable to say that even in cases like these one can put geological time boundaries around the contect - but not "a specified point in time"

 

Regards,

Chris

 

-----Original Message-----
From: rich@valutech.com [mailto:rich@valutech.com]
Sent: 16 June 2003 19:28
To: 'Chris Partridge'; Patrick Cassidy
Cc: standard-upper-ontology@ieee.org
Subject: RE: SUO: RE: CYC event vs. SUMO Process -- really different?

Chris Partridge wrote:
> Pat,
>
> You say: "it seems to be agreed that Events have a location in Time".
>
> I am not sure this is the case - as my earlier email hinted at.
>
> Consider the hypothetical Mary's birthday party. Assuming
> that it does not
> have to be on the actual birth-day, then it could be on any
> number of days -
> if she is lucky it could extend over a number of days. This
> is not a matter
> of ignorance or indecision - when considering the
> hypothetical party we are
> considering the possibilities for it. The time has not
> crystalised. And if
> it does not crystallise - i.e. there is no party - we need to
> explain what
> kind of event we are talking about.

Yet we can reason that, if Bob is planning Mary's party, the
planning is done before the party.  If Bob is remembering
something about Mary's party, then the party happened before
the thinking occurred. 

I suggest that Events have several temporal relations with other
events, but it isn't necessary to position events at specific
times to perform very useful deductions.  Indeed, specifying
a location in time is overspecification for many problems that
deal with common everyday causality.

> This is not a trivial point - but one that makes its
> appearance (in various
> guises) in commercial systems. Consider a financial option.
> This contracts
> for the exercise of an option during a specified period - so
> there is a
> limit on the period, but not a specific time. And this is
> inherent in the
> intention behind the contract.

But we can say that the contract ends at a specified point
in time.  It must be executed before that moment, and after
the executor has assembled all the relevant documents,
monies, and parties of agreement.  So there is much more
sequence computation going on than specific timestamp
computation in the vast majority of financial applications.
And the sequence constraints are much more complex than
the timestamps indicate.

> Does the option contract refer to an event or not? If not, a
> lot of things
> now need a home - if so, then we have to wave goodbye to
> "Events have a
> location in Time".

Good riddance!  Some events have a location in time, others
float within an encompassing event.  The decomposition of
event space is a mixed one, with timestamps and temporal
constraints. 

Rich

> There is a similar vagueness it seems to me in the boundaries
> of most actual
> events - but that is a different point.
>
> It seems to me that one could sustain the thesis that "Events have a
> location in Time" if one stipulated it as a rule - and was
> prepared to make
> the necessary adjustments elsewhere in the scheme. And this
> may be a good
> experiment to try.
>
> But from a more general SUO perspective, I think it will help if the
> motivation and consequences of decisions like these are exposed and
> discussed. Or, to put it the other way around, one would be
> foolish to rely
> on the SUO if these are not critically reviewed.
>
> I agree with your general point, that we need to try and find
> some process
> for moving forward. But, as Pat Hayes and others said at the
> begining of
> this enterpise, it is not easy. I think this example
> illustrates the point.
>
> I also was under the impression that John Sowa's proposal was
> to allow the
> various different decisions to live together under one roof
> so that one
> could compare the consequences - if not develop some idea of
> the motivation.
>
> Regards,
> Chris
>
> -----Original Message-----
> From: owner-standard-upper-ontology@majordomo.ieee.org
> [mailto:owner-standard-upper-ontology@majordomo.ieee.org]On Behalf Of
> Patrick Cassidy
> Sent: 16 June 2003 17:18
> Cc: standard-upper-ontology@ieee.org
> Subject: Re: SUO: RE: CYC event vs. SUMO Process -- really different?
>
>
>
>
>   I would like try to resolve the specific question of
> whether in our standard ontology a 3D Event will
> necessarily have a location in Space (it seems to be
> agreed that Events have a location in Time).
> In John's note he suggests inclusion of both 3D
> and 4D events in the ontology.  I agree.  But I am
> unclear as to John's position on the definition
> of the 3D Event.
>
> John F. Sowa wrote:
> >
> > There is no need to vote, when there is a much
> > simpler way to accommodate the options.
> >
> >  > What I suggest is that the group forces a vote on
> >  > important definitions, such as that of "Event."
> >
>   . . .
> >
> > To take the example of "event", there is no need to
> > force a decision.  The ontology can simply include
> > three types:  Event4D, Event3D, and a generic Event,
> > which includes Event4D and Event3D as subtypes.
> >
>    OK.  These are important issues that we should
> proceed to soon.  I am only hoping that we can
> resolve what should be a simple question before
> proceeding to those other issues.
>     As to the 3D Event, we have an option: is the
> spatial location relation marked as "necessary" or not?
> If necessary, then every instance of Event in a
> knowledge structure will be marked as having *some*
> location. That location can be unknown, or default,
> or may be specified within the knowledge base.
>
>     Can we have both "necessary" and "optional" locations
> in one logically consistent ontology?  In this case,
> I think so, by creating a parent class of 3D Event which
> has only a time location ("TemporalEvent"), and a subclass
> that has both time and spatial necessary locations
> ("SpatioTemporalEvent").  But the split 3D Event structure
> appears to me to make the translation to 4D more
> problematic and I think it should be adopted only if there
> is more than one person who thinks it desirable.  Also,
> it seems to me to be inconsistent with my understanding
> of reality and the usual meanings of Events to imagine
> that they are not located in space.  Even with
> spatial location marked as necessary in the SUO,
> individual users can still change that to optional
> for their own use if such a representation appears to
> them to be better, without imposing an unnecessarily
> complicated structure on all users.
>
>     What I would like to know is how many participants in this
> discussion think that the SUO we adopt should have 3D Events
> which do not necessarily have some location in Space?  Are
> there enough to warrant making the SUO more complicated by
> addition of an extra "TemporalEvent" class above the more
> common "SpatioTemporalEvent"?
>
>      Pat
>
>
> =============================================
> Patrick Cassidy
>
> MICRA, Inc.                      || (908) 561-3416
> 735 Belvidere Ave.               || (908) 668-5252 (if no answer)
> Plainfield, NJ 07062-2054        || (908) 668-5904 (fax)
>
> internet:   cassidy@micra.com
> =============================================
>
>
>