Dear Ronald, (01)
Well I guess I'll take a shot at this. See below. (02)
Regards
Matthew West
Reference Data Architecture and Standards
Manager
Shell International Petroleum Company Limited
Registered in
England and Wales
Registered number: 621148
Registered office: Shell
Centre, London SE1 7NA, United Kingdom
Tel: +44 20 7934 4490 Mobile:
+44 7796 336538
Email: matthew.west@xxxxxxxxx
http://www.shell.com
http://www.matthew-west.org.uk/ (03)
Dear Colleagues,
ontolog-forum@xxxxxxxxxxxxxxxx [That's me, Ronald Stamper, in the
original note consisting of passages not prefixed by initials MW or
RS. This second version, at Peter's prompting has clearer paragraph
breaks.] (04)
I've been eavesdropping ontolog for a while. Now I'd like to
contribute new thoughts on an ontology-1 (in the philosophical sense).
I hope they are relevant to axioms and ambiguity, among other things. (05)
I hope this message is not too discourteously long. (06)
Our research aimed to find a way of specifying the norms that govern
organized human behavior. The practical results are very promising
but they depend on finding a new metaphysical foundation, in
particular an ontology-1 recognizing, to all intents and purposes,
that (07)
a) there is no knowable reality without an agent experiencing it; and that (08)
MW: So that isn't that there is no reality without agents, only that
what is known must be known by an agent, presumably because only
agents can know things. Sounds to be like you might be more interested
in epistemology than ontology. (09)
RS: Perhaps I should not have inserted the word "knowable". We may
imagine all kinds of things and believe in some of them but the only
ones we can deal with in our practical affairs are those that some
agent or other can deal with. (010)
RS: When one holds an apple, one experiences it as the things one can
do with it. When the apple tastes good, this direct experience has
nothing to do with knowledge, which enters into the picture only when
expressed in words or other signs. We say that knowledge is erroneous
or true, thus acknowledging that it has properties that the experience
of the apple or its taste do not have. Distinguishing the sigh from
the reality is essential for this approach. (011)
b) the agent must discover its own reality in the flux of events
and actions. (012)
MW: Are you saying here that there are actually multiple realities,
which are what agents discover, or that there is one reality, that
different agents have different views on based on what different
agents discover (know again)? (013)
RS: We must begin by considering isolated agents. Each isolated
agent constructs from / discovers in the flux of events and actions
its own reality in which it must survive with as much comfort as it
can devise. (014)
RS: Gibson explains how the agent achieves this. (015)
This builds on James Gibson's Theory of Affordances: no organism opens
its sense 'windows' onto a ready-made reality but, from the flux of
information generated by activity (including its own), the organism
must discover the invariant repertoires of behavior, that the
environment affords it. These 'affordances' are the things it
perceives. (016)
RS: Note that, when an affordance registered by a simple isolated
organism ceases, that feature of the world ceases to exist. Existence
outside the here-and-now presents problems that may be solved by using
signs: memory initially. (017)
We call the ontology-1 "actualism". (018)
MW: I think this name has already been grabbed:
http://plato.stanford.edu/entries/actualism/ (019)
RS: Thanks very much for this citation; it did not appear when I made
a brief search. Our actualism appears to be of an extreme kind that
has no truck with possible worlds. Fictions, histories, plans and so
on are actual signs. (020)
Bishop Berkeley's ontology is the nearest approximation I have found.
He had to explain what happened to the tree in the quad when no one
was around to perceive it. As expressed succinctly in the well-known
limerick: (021)
There was a young man who said "God
Must think it exceedingly odd
If he finds that this tree
Continues to be
When there's no one about in the Quad." (022)
We substitute Society for God, with excellent effect. An isolated
human organism will know little of reality without the cooperation of
the rest of Society. (023)
MW: Again, I notice you are talking about what agents know, not about
what exists. (024)
RS: Knowledge enters here at the stage when isolated agents begin to
interact, using sign, to for a consolidated picture from their many
different realities. Epistemology in this approach concerns the
manner in which individuals can rely on signs as good quality
vicarious experience. Ontology deals with understanding experience as
access to what exists. The ground for any ontology for an isolated
individual must be its own recognition of the affordances discovered
in its own experiences. On that basis we individuals can discover
affordances beyond the reach of our puny physiology+niche by employing
the knowledge (signs) shared by our community. Thus we need
epistemology as a tool for constructing the greater reality that we so
easily take for granted, and, therefore, as partner to the rudimentary
ontology for isolated organisms. (025)
I'm no philosopher, theologian or politician but an engineer looking
for solutions that works. Some have accused me of an anti-individual
political bias but this approach does not diminish the importance of
the individual. Belief in God or in any other Truth, is unaffected
within this paradigm, provided you acknowledge that these are beliefs
for which you bear responsibility. (026)
In practical information systems engineering, ontology-1 leads to a
canonical form of ontology-2 (sophisticated data model). (027)
MW: OK. Data models are one way of representing an ontology, but they
are also the cross over from ontology to epistemology, since that
specify the information we want to hold about certain things of
interest. (028)
RS: The ontology of types/universals that a community constructs
includes the ontologies of instances/particulars of those who
contributed to it.
We claim only an empirical basis for this Semantic Normal Form. Could
this have an axiomatic potential? (029)
MW: Well, data models generally contain relatively few axioms (the
cardinality constraints would qualify). But these can be added. (030)
Please give me your comments. Is our ontology-1 really new? I'm sure
you will disabuse me soon enough if I'm mistaken. (031)
MW: My question would be whether it is an ontology or epistemology. (032)
RS: Individuals' experiences provide the groundwork for an ontology-2
that a community or society can elaborate into a consensual version.
What to accept into this consolidated ontology-2 is achieved with the
aid of the various tacit epistemologies people bring to judge what to
accept or reject. Our approach incorporates an epistemology that
requires one to check what agents are responsible for the additions or
amendments. (033)
I gave a paper at the 2007 ICCS (the series that, I believe, John Sowa
instigated); sadly he was not there. But that now encourages me to
introduce the ideas to this group. (034)
RS: By the way, you will not find this paper in the proceedings,
despite its being an invited keynote contribution. Springer insisted
on owning the copyright; I offered them a license to using as they
wish but that was not enough. I hate bullies, including, perhaps
especially, large corporations, so I persisted with my refusal. I'd
be happy to supply a copy to anyone. (035)
Our research on the formalization of social (and legal) norms, which
started at the London School of Economics, forced us to handle
semantics rigorously while never losing sight of the human connection
between sign and reality. (036)
MW: Have you read John Searle, The construction of social reality ? (037)
RS: Indeed! This approach goes much further; it assumes the social
construction of physical reality too, at least beyond the private,
individual experiences of it. (038)
Talk of meaning obliges one to start from the fact that signs have
meanings because they stand for real things. Hence, to have a clear
position on meaning, one must make a commitment about the nature of
reality. (039)
MW: Right. (040)
None of the conventional ontology-1s seemed to work. (041)
MW: Which ones did you try? Realism is the one I favour (along
probably with the majority of others here). What was it that did not
work about realism? (042)
RS: Platonic Realism – not open to empirical examination; simple
Objectivism – social reality such as that dealt with by the law (our
source of empirical material) is constructed and modified by social
processes; Formalism (?) – meanings based on a sign standing for other
signs (computational semantics, some mathematics, EDI Semantic
Repositories or other dictionary meanings . . .) just confine one to a
world of signs; Conceptualism – signs stand for concepts or percepts
in the minds of people (Sausurian semiology, much of linguistic
semantics, IT conceptual schemas, IFIP 8.1 report on a Framework for
Information Systems Concepts etc) does not allow empirical study;
Platonic Realism again – mathematical functions mapping signs onto
set-theoretic structures (some kinds of logic, Montague Semantics)
makes the introduction of possible world essential, I think, if you
need to handle time. What have I missed? So we preferred the
judgment of the proverbial (in UK) 'man on the Clapham omnibus', with
the proviso that you need to identify him and demand justifications. (043)
Finally we adopted Gibson's Affordances and added social norms, which
are invariants that afford repertoires of behavior in the social
sphere. Gibson's Theory of perception of the material world is
insufficient to explain an individual human's perception except,
perhaps, in cases of enfants sauvages. The rich perceptions of those
raised as members of society incorporate many extra perceptions that
have been supplied by other people and form the community's shared
perceptual norms. Moreover the reality of the social world also
depends on the many other behavioral, cognitive and evaluative norms
that we also derive from the community that nurtures us. Therefore we
have taken Society as the root agent in our model of perception. (044)
Berkeley explained a person's perception of a tree as the result of an
idea in His mind, (045)
MW: Now that is beginning to sound like conceptualism. (046)
RS: Angelic Celestialism, we might say. Even more inaccessible to
empirical study! (047)
so that we can accept that the tree continues to exist in the Quad
when "nobody" can see it, because God is always there. (048)
Without the mysterious intervention of God, Society provides an
empirically testable explanation for continuing to accept the tree's
existence. (049)
MW: How? I would have thought the tree was sufficient for continuing
to accept the tree's existence. (050)
RS: No, not at all. Your remark suggests a simple objectivist
position: the tree and all the other things, including the quad, have
their independent existences; they just continue to be there
regardless of anyone. In our sever actualism, you can believe that
they continue to exist – it is a very reasonable position! I have the
same sort of belief but, as an extreme actualist, I have to admit
that, having left the quad, the tree ceases to exist for me, in here
and now, as a realized affordance. But my memory of and my written
note about the tree does exist here and now with me outside the quad.
I can use these records to support my belief and also to help me
persuade others that the tree is still there. After a little while,
the tree may no longer be there, having been felled preparatory to
some building work. I can phone a friend with rooms overlooking the
quad, or ask a person who has just left the college to check the
validity of my belief. If questioned about it, I can justify it
because of a number of shared cognitive norms concerning the longevity
of oak trees, the conservative practice of the college fellows in the
management of their property, especially their desire to keep any
two-hundred-year old tree if at all possible. Thus, through the
skilful use of signs that Society has evolved, the community, as the
responsible agent can hold the tree in existence as a realized
affordance and allow me to share in that aggregate perception. That's
why Society replaces God for good engineering reasons. The Root Agent
in all our business and legal schemas/ontology-1s is Society. (051)
An analogy might help: you directly experience your pet cat over an
extended period (during which it is available to you as an affordance
or invariant repertoire of behaviour – all the things you can do to or
with a cat that differentiate a cat from things in other categories).
However, during that time, sometimes a few cells in your hand have
direct experience of the cat, at other times cells in your legs,
sometimes cells in you ears register the cat, you also see it but not
always . . . and so on. Through the communications via your nervous
system, the direct experiences of thousands of individual cells in
your body become aggregated into the perception made by the whole of
you. I'm saying that the whole of Society perceives things beyond the
reach of any one of us by aggregating our reports of our direct
experiences. Unlike the cells of our bodies, we, as individuals, can
obtain and use pictures of things in the world assembled by Society as
the perceiving agent. (052)
Many people think that this is a very laborious way of explaining the
continued existence of the tree compared with the normal assumption of
realism. Of course, for everyday tasks, we all adopt realism as a
practical ontology-1. Certainly I do. But as an engineer of
information systems, actualism does compel one to examine exactly how
we use information to arrive at the apparent perception of a reality
far beyond anything within the scope of an individual's experience. (053)
Based the cognitive norms shared across Society and using the reports
of responsible observers, checked by critical discourse with others,
we can justify believing that the tree still stands in the quad. (054)
MW: This sounds like something based on our collective knowledge
again, tied perhaps to beliefs. (055)
RS: Exactly! (056)
The rich world we ontolog participants believe in lies beyond the
perceptual reach of any isolated human being. (057)
MW: So here at least you seem to acknowledge a world independent of
our perceptions. (058)
RS: No! I believe in such a world as a valuable heuristic for
everyday decision-making but I acknowledge it as a belief that ceases
to be correct when I'm trying to understand the details of the
semiological mechanisms in a community or even in an information
system. Actualism then becomes an instrument helping to make visible
the complex details of observation, reporting, forming beliefs and
norms, checking what is said etc. (059)
Society enables us to aggregate our puny individual experiences
through the information we exchange and the norms we share. We test
them until we arrive at the familiar picture of an apparently
objective reality. (060)
MW: Sounds like epistemology again. (061)
RS: Epistemology serves as a toolkit for building an ontology-2 that
exploits the very simple experiences of many individuals. The
ontology-1 of extreme actualism enables us to build, from the world
views that unaided individuals can construct, to the sophisticated
world views that Society can develop. (062)
To represent fully something that we believe exists, we must say who
perceives it and what affordance the agent realizes for the period of
its existence. So we need sentences of the form: (063)
Agent affordance (064)
The root Agent must be a particular (convention: uppercase capital).
During the Agent's realization of this affordance, it is, in effect a
modified kind of Agent able to experience some other affordance. (065)
(Agent affordance) affordance (066)
When an affordance ceases to be realized, as far as the root Agent is
concerned that thing totally ceases to exist. (067)
MW: But only as far as the agent is concerned, not necessarily in
reality, right? (068)
RS: No! If the agent is a simple organism with no memory (not
counting evolved, instinctive patterns of behaviour) then, when that
affordance ceases to be realized that particular ceases to exist. If
the agent is more sophisticated, with a memory and a role in Society,
then the thing may continue to exist, for practical purposes, until
every memory or record of it is expunged. Then the thing does cease
to exist for all practical purposes.
The environment may afford the Agent two of these repertoires of
behaviour simultaneously: (069)
Agent (affordance while affordance) (070)
again becoming the modified agent (071)
(Agent (affordance while affordance)) affordance (072)
Thus a) the effective existence of anything (073)
MW: As opposed to the actual existence of anything. (074)
RS: Effective / actual? Does the thing exist in a way that allows one
to act upon it, effect some change in it, or even merely observe it?
A particular thing that some person realizes as an affordance exists
effectively in that sense, because that agent can act upon the thing
on our behalf. There is no 'real reality' beyond the effective
reality unless it belongs in the imagination or books or hypotheses,
which are all signs that certainly are effectively real and available
in the here-and-now. (075)
depends fundamentally on the agent who takes responsibility for
his/her/its choices. (076)
MW: This sounds like beliefs again. (077)
It also b) depends on the coexistence of some other, 'ontological
antecedents'. Between the Agent and the affordance in question one may
draw a lattice where each affordance has one or a maximum of two
antecedents. If this lattice includes the antecedent affordances that
are necessary and sufficient for the existence of the one being
analysed, one will have a schema in Semantic Normal Form. (078)
A little example (I'll leave you to draw a graphical version) is a marriage: (079)
marriage (person-1 (Society), person-2 (Society)) (080)
If person-1 proposes such a marriage, he (usually) must use a sign
that stands for that marriage, which does not exist: (081)
proposal (person-1 (Society), "marriage ( . . . .)") (082)
As the ontology-1 only allows us to talk about things existing here and now, (083)
MW: Oh, so you are a presentist too then are you? (084)
RS: Absolutely! It's a marvelous discipline making one see things as
they are. (085)
everything in the past or future is available only in this semiological form. (086)
SNF-compliant schemas are very stable and systems built on them can
accommodate changes of requirements with remarkable economy. Looked at
from another angle, the schema in SNF has a valid generality across
cultures, while able to accommodate any differences through the
variations in the authorities that determine when things start and
finish their existence. This supports the accretion of semantic
information without imposing any artificial uniformity. Presumably,
these properties are highly relevant to the semantic web. (087)
SNF-compliant schemas can be aggregated. We have built some quite
large schemas (088)
MW: Large I discover is a relative term. Is large for you 100 entity
types, 1000 entity types, 10000 entity types 100,000 entity types or
1,000,000 entity types? (089)
RS: You are quite right to ask. Here, in our remote French hamlet, I
have none of the data. However I can offer guestimates for some
systems: Higher Education admin, c.600; Packaging and Marketing for a
large brewery, c.1000; Marketing and Distribution for Electrical White
Ware, c. 400; Raw Material Quality Control for a Food Manufacturer,
c.200; Medical Insurance Policy, c.200; etc. These are large enough
for the specification of computer applications in those areas. (090)
but need far more experience to discover where the limits lie. (091)
Building a schema in SNForm is not a trivial task, by the way. If you
wish to try, don't fall into the trap of treating an ontological
dependency as a cause-effect relationship. Your schema can represent
only what exists here and now, as defined by the Agent and its
activities. (092)
MW: I prefer realism and 4-dimensionalism, so I think I'll pass on this. (093)
RS: The agent is the origin of its experienced spatial domain.
Topological affordances enable to agent to experience the
dimensionality of the space. (094)
Times have to be constructed semiologically. An agent, during the
realization of an affordance, may assume it has a start and will have
a finish; provided the agent is sophisticated enough, it can refer to
those events as "The start of x." and "The finish of x." and use these
signs as rudimentary time references. Introducing chronometers that
allow us to attach numbers to them leads to the notions of time we now
take for granted. Start and finish events always lie in the past or
future, not in the here-and-now. (095)
Every node on the schema should be accompanied by a list of attributes
that includes the identifier of the thing it stands for, the universal
it instantiates, when its starts and finishes its existence and the
authorities that determine these start and finish events/times. (096)
MW: So when does the number two start and finish, or the colour green? (097)
RS: For an isolated agent everything starts (if at all possible) not
before the agent comes into existence and finishes no later than the
agent's death. For most animals that is or is nearly the case but they
also perceive certain affordances by evolved instinct. For social
agents with varied and sophisticated semiological skills, we can build
theoretical models of reality going back to the Big Bang and
reasonably believe that many different shades of green began to exist
in the spectra of various elements, but these are beliefs. Labeled
shades of green that members of society actually experienced and about
which they can communicate probably came into existence (my guess)
100,000 years ago. But colour categories are culturally dependent and
probably change over time. (098)
We know that a few animals and birds can recognize the affordances of
two or three things, so those integers may have been parts of the
realities of some isolated creatures. There may be some traces in the
archeological and historical records concerning the starts of the
existence of some integers in Society. The infinite and the eternal
do not exist in an actualist's world except as signs, in particular as
rules (signs for norms) governing processes that have no necessary
termination. (099)
All our cherished beliefs and theories are acceptable in this
actualism provided that one acknowledges they are beliefs and,
therefore, subject to critical appraisals recommended by whatever
epistemology anyone chooses to judge them – the value of the chosen
epistemology can the be judged by its ability to eliminate damaging
practical consequences in our actions. (0100)
Our Legally Orientated Language for manipulating data held under
SNF-compliant schemas allows us to represent social norms precisely
and in a form that 'naïve' users find easy to understand. (0101)
We have designed many and built some systems (but not yet enough)
using these concepts with marked success. (0102)
MW: Can you say what the applications you have built do? (0103)
RS: See the list above, first: most of these ontology-2s, if they were
used to implement systems, were taken over by people employing
conventional methods. Our interest has been to add to the ontology-2
the norms that govern the starts and finishes of things in order to
complete the specification of the human system roughly in the manner
of drafting legislation governing actions by people in the domain. We
then have the further interest in automating the implementation of
helpful computer applications from such system specifications. We
have implemented one administrative system this way with truly marked
success: even with a first attempt, development costs were cut by a
factor of three, support and maintenance costs by a factor of seven in
comparison with a similar system implemented using an adjustable
package that had already evolved over 200 different implementations.
Subsequently we found that nearly all our maintenance costs had been
caused by the initial adoption for a short period of orthodox
requirement specification methods that introduced 'malignant
structures' (Ades). Perhaps more importantly, our methods were able
to suit the exact needs and all the needs of the organization whereas
the organization using the packaged solution had to be altered to fit
the package and some of their functions had to be met by marginal
annotation of printouts. Because the SNF schema components are
reusable, and because we can perform the analysis much more quickly
now, we are confident that development costs can be reduced by another
factor of three. We'd welcome the opportunity to test the methods on
other applications. (0104)
Note that although we have been funded by Research Councils in the UK
and Netherlands, we have never had any amount of funding approaching
that of CYC. (0105)
The formal aspects need far more work. (0106)
RS: How better to explain actualism, this weird way of thinking, more
effectively also calls for much more work. I'm just completing a book
on how to do it in a practical way. (0107)
Awaiting some feedback, with my regards, (0108)
RS: For this first feedback from Matthew West, my warmest thanks. It
is so healthy to be forced to respond to such perceptive comments. (0109)
Ronald Stamper (0110)
PS: It has taken some time to be able to send this message and now
I'll be away from my desk for a few days. - Further apologies! (0111)
RS: I'll post this straight away to the forum although I see there is
another posting from John Sowa for my attention. Having the narrowest
of narrow band connections to the net, makes for hard work. (0112)
Forgive me if I've made a pig's ear of the layout of the text I'm now posting.
[Second version – I hope this pig's ear is slightly more readable.] (0113)
Ronald (0114)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (0115)
|