Dear Ken, please see my comments below,
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
From:
ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Ken Laskey
Sent: Tuesday, January 03, 2012
7:15 PM
To: '[ontolog-forum]
'
Subject: Re: [ontolog-forum] Self Interest Ontology: Emotions in
animals
Rich,
I’m not sure
all situations can be unambiguously decomposed.
That is correct, I agree that not ALL
situations can unambiguously fit to the model. But the model can return a
response as a list of components, and interconnections. The words
describing the KINDS of components and assemblies exist.
If that situation can be cost effectively
subdivided into components and interconnections. In other words, by
keeping a database of descriptions by different Observers, you are actually collecting,
discoverying and measuring the bias, i.e. the self interest ontology, of that
Observer plus a lot of noise in any one situation.
Add more reports from that Observer to the
database, and you will begin to see order in that Observer’s ways of
recording things. That order is the habitual response of the Observer in
the enterprise. Whatever it is, it can be sensed, measured and Observed
with all the equipment commonly available to businesses and governments and
even to occasional citizens.
So identifying the Observer’s bias
is improved as you gain experience, i.e., as you collect more data, discover-process
the data, improve the focus by what has just been learned and added to the
database, and record the distinction between signals and noises as the system
and Observer interpreted them at the time of measurement.
The IDEF0 standard can represent anything
in terms of description as you build models. The AsIs and the ToBe models
are each approximations (not reality signs) of the data they have
discover-processed in the stepwise performance of IDEF0 models as required to
match the sensors, Observers and analysis tools to do the actual learning steps
of induction, deduction and conduction as needed by logicists.
There are certain
classes of problems that have holes in all proposed models and it may well be
self-interest when identifying the one that is most accurate. Given two
choices, one may turn out to be more accurate and that side will feel
vindicated that they did a “better” job when all that happened was
science/history/whatever in that instance better conformed to their prejudice.
Deciding whether economic policy actually had a positive or
negative effect is certainly in this category.
Ken
I certainly agree with that also. Economics
is as difficult to discuss civilly as Politics, and I think that self interest
is the explanation which describes an Observer’s bias, any Observer, to
some degree that varies with each one. That is why we need ways to sense
our actions, test them for validity, subjectivity and other personal facts so
that those factors can be offset from the calculations when required. Measurement
and discovery makes a database with measured detailed history of the model
component and assembly, and the learning curve makes them more efficient as
experience is gained within that discovery step.
After that discovery step A has been
completed, the Before and After situations are known as discovered during the
step. So now we have a decomposition of activity A in terms of a Before
and an After situation. Those we model as TYPES of situation, using
whatever language was the basis of the description. We discover another
situation in the database which is composed of instances of those same TYPES,
as best APPROXIMATED by the discovery process step. Those TYPES are the
components, or key word set which I call Frequent_Words. So at the
beginning of the step, the text is tokenized, Frequent_Words are recognized,
and the remainder of the situation is represented using problem
reduction.
Problem reduction works for Situations
because there is some way to model some aspect of any English word. I
select the set Frequent_Words after sampling a few times to get the words in
patent claims which had a sense of the commonly used structural words, something
like the list below:
(claim
comprising method system apparatus of the a plurality all one or and each every
for …)
I have most recently chosen about 1,300
words to be those frequent words which are being used syntactically and those
which carry semantic meaning – the rare words in the sample. That
seems to work well for analyzing patent claim language.
HTH,
-Rich
From:
ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Rich Cooper
Sent: Tuesday, January 03, 2012
10:00 PM
To: '[ontolog-forum]
'
Subject: Re: [ontolog-forum] Self Interest Ontology: Emotions in
animals
Dear Ken,
My comments are embedded
into your post below,
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT
EnglishLogicKernel DOT com
9 4 9 \ 5 2 5
- 5 7 1 2
Rich,
Your
examples point to cases where occurrences are (at least somewhat) repeatable
vs. one off situations where the debate is often what are similar occurrences
against which we can accumulate corresponding data.
[RGC] Do you mean, by one
off situations, that there is no modularity of components that can be put
together to describe each unique situation? In my experience, there is
not a practical limit on how much you can develop very simple models, refine
and rerefine them with lots of experience, and improve nearly anything you can
define.
The learning curve (see
the book “Bionomics”) consistently shows a gain in anything
measured and used to modularize actions into contexts, decompositions,
activities, objects, as per some widely understood modeling language such as
IDEF0. It doesn’t have to be invented. The consistent
learning curve shows that, for each doubling of experience, there is a gain of
about 20 to 30 percent. It decays exponentially for cost curves at that
rate, or it grows exponentially for value curves. And of course you can
imagine using vectors of values and costs to guide further refinement.
That is the software
implementation of a learning cure based on database discovery. The
database portion can be grown like a forest over the metadata tables, mapping
symbols and terms, whether axioms or Observations. The result is an
automatically generated ontology, in part, because much of it can be tool
based. But the Observer will have to be there, kinda like in
physics.
Intervention
or nonintervention in another country is certainly full of self-interest (both
by those who intervene and those who don’t), but matching two historical
situations is often nontrivial, especially when those doing the matching are
also exhibiting self-interest.
I agree. I
don’t mean that the task is small, just that it can be specified in terms
of data organization plus supporting procedures, constraints, dictionaries and
search processes. That is just the first step in a problem reduction
process. The next step is to identify what data is available for
analysis. I consider us in this step at this time. Next we have to
generate a decomposition (my words) of the context. That is, how can we
specify general linguistic components that can be configured to solve or
simplify many kinds of corpus analysis discovery problems. One of those
problem targets is to specify a cost effective method to automatically generate
ontologies.
I described a couple
posts ago how that automated method can be implemented with the database
analysis methods I described a couple posts before that. That may have
not been obvious then, so I am trying to make up by explaining it that
way.
So
some models are easier to build than others, and the valid use of models
depends on us remembering that.
Ken
That is certainly wise
advice. Perhaps we could also go too far in the sense of Isaiah Berlin. There is a
time to stop building models and to try a couple of discovery steps to make
progress. That is what I am hoping to stimulate.
You are right; all models
have to be cost justified if possible. But small ones might slide under a
regulatory limit or two. Then just one-offs of implementations seems to
make sense to try.
Then, we will have
processes in place to analyze the data that is actually available within that
step. The bigger resources will be needed when we want to scale up and
employ thousands of computational linquists to tend to the databases. But
that will take time.
-Rich
Dear
Ken, Doug and John, et al,
Ken
wrote:
…
we could come up with numerous other examples where action was needed and what
we believe would have been a worse situation was averted by those
actions. Some are fairly clear cut; others are the grist for historians.
Agreed;
it isn’t a simple model for most situations of interest. But
knowing how deeply self interest is layered on emotions, we could develop
models.
The many
explanations of observed behavior differ by the Observer, who is not able to
reliably match the next Observer’s model. It is our own projections
we build into our models. But by finding ways to develop and test models,
we become able to distinguish among those that work within some validatable
boundaries, and those that simply don’t work.
While
doing R&D at Hughes, I sometimes worked in electronic warfare. That
is, EW is the art of listening electronically to emissions. From those
emissions, EW designers are able to model the behavior of the emitters.
But the emitters belong to somebody who is very interested in who is watching
them. A jeep with a radio bouncing down a jungle trail, compared with a
map of the trails in the area, can be used to predict where the jeep is
going. Tracking multiple vehicles with emitters let the situation be more
clearly modeled, and let the predictions be analyzed, improved, eventually
validated. So jeep drivers have an interest in minimizing their
emissions.
Submarines
no longer use active sonar, not EVER. They now listen to outside noises,
and use algorithms and databases to model the observed emissions. You may
remember about ten years ago that a submarine surfaced right into a commercial
vessel, drowning dozens. The commercial vessel must not have been running
(no mechanical sound emitted). It must also not have been radiating radio
frequency emissions. The vessel must have been completely stopped in the
water and quiet on every available emission band.
Radar
was once active, sending pings at regular intervals and measuring distance that
way. With the EW advances, major power militaries learned to detect
emissions without their radars, sonars, and other sensors emitting recognizable
pulses that the emitter could recognize. Sending out pings gives away the
position of the radar, the jeep or the submarine, because the pings are emissions
themselves, which give information about the sensor to the observed
targets.
Stealth
technology was developed to improve the defense situation. By not
emitting or even reflecting emissions, stealthy devices can remain
unobserved.
The
point is this. We can use the many refined EW techniques to model self
interest.
This
process can start with things we can reliably sense, and then construct models
of the situation we believe is indicated by those sensible things. Models
that match the evidence detected by the Observer can be kept, but those models
which don’t validate can be revised, improved and tried again in various
situations. Ultimately, with a lot of work, we can determine which models
are valid for which situations.
The
usual discovery process is shown below as adapted to reflect the linguistic
mining of messages. The process is described in much greater detail in my
patent, the 7,209,923. Said patent is attached to this email if you are
interested:
To
repeat your statement:
…
we could come up with numerous other examples where action was needed and what
we believe would have been a worse situation was averted by those
actions. Some are fairly clear cut; others are the grist for historians.
Yes, we
could. So recording those actions and objects as sensed by listening to
the emissions is the way to do it. The most likely source of those
actions and objects are human Observers. The Observer needs tools to
develop the models, to organize the data, to fully discover whatever is
observable, generating a theory that validly models the actions, the situations
before the actions, and the situations that occur after the actions.
That is
the way I see a Self Interest Ontology being developed. The ‘923
describes how I am doing so with the USPTO patent database. Even very
mechanical documents such as patents exhibit subjective judgments, by each
Observer (e.g., inventor, examiner, attorney, agent, litigator). The file
history of a patent is a document that is also public, easily accessible, and
available on the web.
One
example source of information which is, unfortunately, only in voice emissions,
not easily converted to text. That source is a TV series call “In
Treatment” which is also available on the web. It details stories
in handy thirty minute chunks with only two people 90% of the time. The
dialog is direct, honest psychological treatment of unusually conflicted
individuals. The vocabulary they use, the syntax they use, is of limited
scope, and the characters are deeply developed by the plot.
Also,
there are many, many shows available, for example:
http://btjunkie.org/torrent/In-Treatment-Season-1/4358f915d28d17f71174da3167fb318d1505c96e0f87
That URL
will download forty three shows in one season.
The
point is that there are available ways to gather empirical data, to construct
models that describe it in FOL, and to have Observers discover new linguistic
realities that can data mine human self interest based on the emissions.
That would of course have to include the Observers doing the modeling, because
their idiosyncratic actions and objects have to be nulled out of the
model.
JMHO,
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT
EnglishLogicKernel DOT com
9 4 9 \ 5 2 5
- 5 7 1 2
It
is easy to point to examples where intervention seems to cause more problems
than we suspect would have happened by leaving the system to sort itself
out. However, I’m sure we could come up with numerous other
examples where action was needed and what we believe would have been a worse
situation was averted by those actions. Some are fairly clear cut; others
are the grist for historians.
Unfortunately,
our what-if scenarios don’t really tell us whether what in hindsight
looks like success or a disaster would, in fact, have a better outcome if
different action/inaction occurred.
Ken
---------------------------------------------------------------------------
Dr.
Kenneth Laskey
MITRE
Corporation, M/S
H305
phone: 703-983-7934
7515 Colshire Drive
fax: 703-983-1379
McLean VA 22102-7508
Dear John,
You wrote:
Things
would have been far better for the Afghan people, Pakistan,
the US,
the Russian people, and the entire world if Reagan and the CIA had done
nothing.
Yes, in generalized form, the
conclusion I draw is that organizations err on the side of doing too much,
especially as suggested by the ideas of Isaiah Berlin as documented by Curtis, and as
supported by the numerous examples which he shows.
So more generally, the consistently
human error is in doing too much
when we think we are in the right. That has held true for so many
examples in history that it can assumed that every organized plurality of
people with a common self interest will eventually go too far if not
stopped.
The notion of checks and balances is
sometimes thought to limit just how far the organization can go. Jefferson was the architect of the American system of
checks and balances. In Brittain, Cromwell hanged the then king for
treason. There was a period of time when England did without a king, but the
upper classes, I am told, wanted to cement their roles as ruling class, and
reinstated the royal line after Cromwell's death. Britain's
political structure of parliament and elections were intended to provide checks
and balances there, I am told by historians.
Dictatorships of all persuasions seem
to appeal to the self interest of the dictator and those few forces that keep
him in power. The word "dictate" from Latin simply means to state,
much like dictation machines in the old technologies of the fifties. The
connotation is that the dictator has the power to make his statements become
real. The rest of the citizens can dictate until the llamas and camels
come home, but there won’t be a reality that corresponds to their
dictations. Syria
is the most contemporary example I can think of.
Democracies spread the base of power
somewhat by letting citizens express their choice through voting within a
limited set of options. That means the self interest of the electorate
has a greater voice. But it doesn’t mean democracies are any less
subject to Isaiah Berlin's
warning. Athens
warred on other city states, forcing their own self interest to be
realized. The North invaded the South in the American civil war to
enforce their economic interests.
So the only concept of which I am
aware that can limit the power of any organization is some kind of well
constructed set of checks and balances, but even that is not sufficient.
It is only a step in the right direction until we can come up with a better way
to limit organizations more effectively.
But there will always be zealous
advocates who persuade organizations to do too much. Sad, but true.
I don’t see a way to stop said organizations from doing too much.
But by modeling self interest, we may be able to learn how to detect, perhaps
even automate the detection, of when the organizations are going too far.
JMHO,
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
-----Original Message-----
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx]
On Behalf Of John F. Sowa
Sent: Tuesday, January 03, 2012 1:38 PM
To: ontolog-forum@xxxxxxxxxxxxxxxx
Subject: Re: [ontolog-forum] Self
Interest Ontology: Emotions in animals
On 1/3/2012 1:58 PM, Rich Cooper wrote:
> But the very righteousness that
drove the revolutionaries,
> they felt, justified taking
inhumane steps to force people
> to be in line with their plans,
since they felt their plans
> would bring good. Instead,
their convictions turned out
> to be the cause of their
downfall.
Fundamental principle: never
trust anybody who claims
to know the will of God or anything
else that is too
complex for anybody else to
understand.
> The Sandinistas, for example,
which even Reagan supported.
Reagan also funneled money through
the CIA to support
Osama bin Laden in the fight against
the Soviet Union
in Afghanistan. He even sent
money to the Taliban to
recruit and train more fighters against
the USSR.
That was another example of people
who thought that they
were doing what was right.
Things would have been far
better for the Afghan people, Pakistan, the US, the
Russian people, and the entire world
if Reagan and the
CIA had done nothing.
John
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: 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 join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J