CCAT: OBJECT/EVENT/PROCESS: Needs work
fritz@rodin.wustl.edu (Fritz Lehmann)
Date: Fri, 11 Nov 94 10:24:06 CST
From: fritz@rodin.wustl.edu (Fritz Lehmann)
Message-id: <9411111624.AA03015@rodin.wustl.edu>
To: cg@cs.umn.edu, doudna@aol.com, srkb@cs.umbc.edu
Subject: CCAT: OBJECT/EVENT/PROCESS: Needs work
Cc: M.J.Johnson@qmw.ac.uk, anquetil@IRO.UMontreal.CA,
beancar@cucis.cis.columbia.edu, bezivin@unantes.univ-nantes.fr,
bill@violin.att.com, billrich@VNET.IBM.COM,
brayman@zuben.ca.boeing.com, buteau_brandon@prc.com, cassidy@micra.com,
cbwillis@netcom.com, cyre@vtvm1.cc.vt.edu, dick@Glue.umd.edu,
doug@csi.uottawa.ca, fletcher.mcleancsd@xerox.com,
fritz@rodin.wustl.edu, ged@cs.rmit.edu.au, gerbe-o@immedia.ca,
grau@falcon.depaul.edu, guarino@ladseb.pd.cnr.it,
kivs%bgcict.BITNET@umbc2.umbc.edu, kra@yu1.yu.edu,
kremer@cpsc.ucalgary.ca, kschoi@cs.kaist.ac.kr,
lukose@peirce.une.edu.au, moulin@ift.ulaval.ca, oh@vax2.cstp.umkc.edu,
peterman@informatik.uni-hamburg.de, phayes@cs.uiuc.edu,
roger@ci.deere.com, s.griffin@mcs.surrey.ac.uk,
shmyaeng@mailbox.syr.edu, sowa@turing.pacss.binghamton.edu,
thompson@zuben.boeing.com, wei@oucsace.cs.ohiou.edu, willems@cs.vu.nl,
william.woods@east.sun.com
Sender: owner-srkb@cs.umbc.edu
Precedence: bulk
Pat Hayes and Walling Cyre have been debating the distinctions among
EVENT, OBJECT, PROCESS etc., mostly without copying their messages to the CCAT
people or the CG or SRKB lists. Unless a feeling develops that we CCATers are
overloading these lists, I think ontology discussions (interesting ones like
Hayes' and Cyres') should be copied to the CCAT people and to CG and SRKB.
Cyre has a fairly standard engineering notion of EVENT, STATE and
PROCESS, while Hayes feels the distinctions among them are bogus: In an
earlier message Hayes said:
-------begin HAYES quote------
On event/object/process. Trying to read whats been written and decide on
which is right is never going to work, largely because you will never stop
reading. What we say about things is not going to depend much, if at all,
on whether we declare them to be events, objects or processes in any case.
Let me suggest an alternative strategy, which is to have a default
assumption that EVERYTHING IS PRETTY MUCH THE SAME KIND OF THING AS
EVERYTHING ELSE, and only depart from this when there seems to be a real,
pressing NEED to make a distinction of some kind, typically because one
class of things has something true of all of it which are false of the
others. Thus, I am with CYC that people, and indeed many complicated things
such as my car, are both objects and processes and maybe even events as
well. (If we allow some events to take longer than others then I see no
reason to say they arent events, anyway.)
Some linguists have found linguistic reasons to distinguish events from
processes in order to help sort out the kinds of things that people SAY to
one another, but Im not convinved that we always need to import all this
complication into the ways people think.
-----end quote----
Yes, and no. The world may indeed be just a dynamic miasma of moving
substances at the deeper level, but we in perception and thought have to
carve it into manageable entities, and I think we do so "at the joints" which
yields a system of objects and events. The idealizations of Systems Theory,
electronic timing diagrams, Petri nets and ordinary natural language
sentences form (graph-theoretic) systems of discrete entities. Just like a
signal-processing and recognition program that takes a fuzzy satellite
picture and segments, divides, arranges and classifies its content. Even if
"everything is a process" we will still need to differentiate event-processes
and object-processes ("occurrents" and "continuants" to some people) very
early in our ontologizing.
Consider a school of fish. When all the little fish swim together, we
call it a school moving; as some fish depart the school and swim
independently the school degrades, and if enough do that then the school
ceases to exist. For "school" substitute "object". Objects may need to be
discerned from their underlying processes in this way. Doug Skuce and Nicola
Guarino recently told me that Mario Bunge has written on this kind of thing.
I'm wedded to the idea that our ontology needs EVENTS and their
PARTICIPANTS, and that the so-called THEMATIC or DEEP-CASE RELATIONS (the
"roles" or semantic relations) between the event and its distinct
participants will turn out to be crucial for correct analysis. This does not
refer to the grammatic cases and prepositions in English, but to deeper,
purely conceptual relations which occur in an elaborate hierarchy of
relations, on which the natural-language cases are based, like AGENT,
INSTRUMENT, etc. These relations presuppose a distinction between an EVENT
and its participants. The EVENT/OBJECT distinction is imperfectly reflected
in the verb/noun distinction in natural language.
-------begin HAYES quote-----
I thus suggest the following
strategy: 'event' just means 'completed process', so 'process' is 'ongoing
event'; and objects are processes, and so once dead or destroyed are
considered events. Processes are objects, of course, just like every other
physical thing. Thus, Julius Caesar is now an event, but you and I are
processes. Thats it: no more need be said about events/objects/processes
(unless one is interested in analysing the wierd subtleties of the English
tense system), and we neednt bother much about the terminology, which is
completely dependednt on the point of view of the speaker in any case.
------end quote------
This suggestion not only does violence to the quoted English words
(despite their excessive lattitude) -- it also has no discernible use. What
useful qualities or constraints would subconcepts of these notions inherit?
And why should the subjective "now" of completion be an important distinction
between EVENT and PROCESS? The "weird subtleties of the English tense
system" are not fundamental to my objections, although it would be good if
the ontology could support natural language tenses. This
OBJECT/EVENT/PROCESS stuff is pretty basic to any real-world ontology and,
just because thinkers disagree about it, that doesn't mean it can be swept
under the rug. Although we may not please everybody, let us at least try to
erect a standard to which honest men may repair (to paraphrase Washington).
------begin HAYES quote from another message to Cyre------
Let me withdraw my earlier suggestion and make a more radical one.
Considered as categories, the concepts EVENT PROCESS STATE and OBJECT are
useless: their boundaries are too arbitrary and confused. They all just
mean 'something with a temporal extent' (as opposed to, say, the integers).
Any attempt to classify them more exactly is bound to lead to confusion,
examples which dont fit in any category (or fit into several), etc. And
they serve no useful purpose. There is no need to make these distinctions
in formulating axioms about the world. (In attempting to state general
properties of the conceptual system apparently underlying English tenses
such distinctions may be essential; but thats a different task.)
------end quote------
Alas, the fact that the boundaries of these concepts are "too arbitary
and confused" doesn't mean they are useless, unimportant or can be dispensed
with. They do NOT "just mean 'something with a temporal extent'", although
each of them is, among other things, something with a temporal extent.
Again, it's not just English, but the pure conceptual analysis of happenings
in the real world, that requires these concepts.
Hayes wrote earlier (more or less) that STATE is something basic that we
all should be able to agree on at the start. Evidently that's not so easy,
but I think we can come to some agreement. So long as a particular STATE
remains in force, not much happens of any _importance_. Doug Skuce's
ontology has distinguished STATE from EVENT only by the speed of change; I
don't quite agree, but I don't have a formal definition of a STATE.
I have to agree with Hayes that people use Event, State and Process in
lots of different ways, especially Process; here are a few examples.
Terence Parsons ("Events in the Semantics of English" ) writes: "States.
The sentence "The dress is pink" reports a state. States hold for varying
amounts of time. It does not make sense to ask how long a state took (though
one can ask how long it lasted), nor does it make sense to ask whether it
culminated (finished)." States, Processes, Accomplishment-Events and
Achievement-Events are all kinds of "Eventualities" in Parsons. "At a given
time, a state simply holds or it does not." It appears to amount to (an
equivalence class of) propositions which remain true without change for a
period of time. Those propositions of course must be temporally restricted
rather than eternal.
In the Cyc Project, PROCESS is to EVENT what STUFF is to INDIVIDUAL-
OBJECT, that is, if a PROCESS is divided up, the pieces are still PROCESSES,
whereas an EVENT, like an INDIVIDUAL-OBJECT, loses its identity upon being
broken up. All properties of a PROCESS are static, i.e. invariant over the
life of the PROCESS, unlike the case of EVENTS.
In the Penman Upper Model, a PROCESS is distinguished from an OBJECT or
QUALITY, and has participants (like ACTOR) and circumstances (like LOCATION).
PROCESS includes most of the ontology -- including fairly abstract things
under RELATIONAL-PROCESS like OWNERSHIP and ASCRIPTION, as well as MATERIAL-
PROCESSES like ACTIONS. It doesn't have "EVENT", although the Pangloss
ontology (which includes the Penman Upper Model) might have an EVENT concept.
Myles Brand ("Intending and Acting") calls EVENTS "spatiotemporal
particulars"; an event _occupies_ a spatiotemporal region. Identity of
events is considered. Some say identity of occupied regions suffices; others
require identity of causes and effects. Unlike OBJECTS, EVENTS can co-occur
in the same region.
R. M. Martin ("Semiotics and Linguistic Structure") attaches an event-
variable to a primitive (noncomposite) relation. Speak(Pat) becomes
Speak(Pat)e, an event of Pat Speaking. Events are decomposable into unions
of events. A state is a kind of event e such that if it is decomposed into
e_1, e_2, these have the same properties as e has; similar to Cyc's 'Process'
notion. For Martin, a process involves change of state. Several axioms are
provided.
Richard Montague treated events as predications of moments in time
(INSTANTANEOUS GENERIC EVENTS) or of intervals in time (PROTRACTED GENERIC
EVENTS).
To Whitehead ("Process and Reality"), an EVENT is a NEXUS of ACTUAL-
OCCASIONS, interrelated in some determinate fashion in one extensive QUANTUM.
AN ACTUAL-OCCASION is an EVENT with only one member. A moving atom is an
EVENT consistinig of many ACTUAL-OCCASIONS, each of which have spatial
extent.
In Kathleen Dahlgren's Naive Semantics, an EVENT is a kind of REAL,
TEMPORAL, RELATIONAL, ENTITY. An EVENT can be GOAL or NONGOAL, and can be
(following Vendler) ACTIVITY, ACCOMPLISHMENT or ACHIEVEMENT, and can be
NATURAL or SOCIAL. EVENT verbs are distinguished from STATIVE verbs like
"own".
Webster's Third New International: Event. 1. Something that happens...4.
An occurence, phenomenon or complex of processes occupying a restricted
portion of four-dimensional space-time...
Process. 1a. A progressive forward movement from one point to another on the
way to completion; the action of passing through continuing development from
a beginning to a contemplated end; ...d (1) A natural progressively
continuing operation or development marked by a series of gradual changes
that succed one another in a relatively fixed way and lead toward a
particular result or end...(3) A set of facts ... that can be observed and
described throughout each of a series of changes continuously succeeding each
other... (4) A succession of related changes by which one thing gradually
becomes something else.
State. 1 A mode or condition of being... 2. A condition or stage in the
physical constitution of something...(compare Condition, Mode, Situation,
Posture, Status).
Roget classifies Event under ABSTRACT RELATIONS: CHANGE: COMPLEX CHANGE:
PRESENT EVENTS: EVENTUALITY. State is under ABSTRACT RELATION: EXISTENCE:
MODAL EXISTENCE: ABSOLUTE (versus RELATIVE): STATE.
I don't know how the medical hierarchies treat these matters, but surely
they have concepts subsumed by these, such as conditions, acute changes,
episodes, electrolyte processes, syndromes, etc.
I suspect that we will regard everything real as some nebulous low-
level thing ("eventuality"?) selected from the miasmic background, and then
somehow differentiate OBJECT, STATE and CHANGE-OF-STATE, treating the latter
as an EVENT. A PROCESS may be a SYSTEM of STATES and EVENTS. (This
conforms to most modelling practice, I think.) A good scheme should be
general enough to cover narrowly formalized worlds like Petri nets and
timing diagrams, as well as events described in ordinary discourse. We
should consider what concepts we want, and why, rather than getting too hung
up on the words. I hope other people will add to this discussion.
Yours truly, Fritz Lehmann
GRANDAI Software, 4282 Sandburg Way, Irvine, CA 92715, U.S.A.
Tel:(714)-733-0566 Fax:(714)-733-0506 fritz@rodin.wustl.edu
=============================================================
PS: Does anybody know the email address of Gian Piero Zarri in France?
(Author of the RESEDA ontology of human events).