Knowledge
Technologies and the Asymmetric Threat
Paul
S. Prueitt, PhD
Research
Professor
Cyber
Security Policy & Research Center
George
Washington University
3/16/03
Notes
Note on Artificial
Intelligence
Note on Stratification
Theory
Note on Urgency
Note on Privacy
Issues
Note on Language
and Linguistics
In Summary
Knowledge
Technologies and the Asymmetric Threat
New
methodology will soon allow the continuous measurement of world wide
social discourse. The measurement can be both transparent and have
built-in protection of individual privacy. Web spiders provide the
instrumentation for this measurement. Natural language processing
and machine-ontology construction provides a representation of social
discourse.
Two
different levels of organization to the information facilitate the
built-in protection for privacy. At one level is a stream of data,
most of which is processed into abstractions about invariances in
linguistic variation. At the other level is an archive of invariance
types and compositional rules related to how grammar is used actually
observed in the construction of meaning. Between this grammatical
layer and the data stream can be placed Constitutional restrictions
that require judicial review before individual data elements are investigated.
Put
simply, linguistic and social scientists develop a real time and evolving
model of how languages are used to express intention, but the measurement
of the social discourse cannot be used indiscriminately to investigate
the actions and behavior of individuals, without court supervision.
General
systems properties and social issues involved in the adoption of this
methodology are outlined in this paper. Technical issues related
to logic, mathematics and natural science are touched on briefly.
A full treatment requires an extensive background in mathematics,
logic, computer theory and human factors. This technical treatment
is open and public and is being published by a not for profit science
association, the BCNGroup.org, and by OntologyStream.com.
Community
building and community transformation have always involved complex
processes that are instantiated from the interactions of humans in
the form of social discourse. Given unfortunate cultural conditions,
these processes support the development of asymmetric threats.
Community
building and transformation are involved in responding to asymmetric
threats. But informational transparency is needed to facilitate the
social response to the causes of these threats.
Knowledge
management models involve components that are structured around lessons
learned and lessons encoded into long-term educational processes.
The educational processes must be legitimate if the knowledge of the
structure of social discourse is to have high fidelity. The coupling
between measurement and positive action has to be public and transparent,
not simply as a matter of public policy. Otherwise the fidelity of
the information will be subject to narrow interpretation and, more
often than not, to false sense making.
As
a precursor to our present circumstance, for example, the Business
Process Reengineering (BPR) methodologies provide for AS-IS models
and TO-BE frameworks. But often these methodologies did not work
well because the AS-IS model did not have high fidelity to the nature
and causes of the enterprise. Over the past several decades, additional
various knowledge management disciplines have been developed and taught.
In knowledge management practices there are deficits
in scholarship and methodology that are due to a type of shallowness
and to the intellectual requirements imposed on understanding issues
related to a stratification and encapsulazation of individual and
social intention. The shallowness of the discipline of “knowledge
management” is rooted in economic fundamentalism.
The war calls for a new maturity on issues of viewpoint
and truth finding and the rejection of all forms of fundamentalism.
This maturity is needed because our response to terrorism’s
fundamentalism has had a tendency to engage forms of ideological fundamentalism
within our own society. American has a strong multi-cultural identity,
as well as a treasured political renewal mechanism. When challenged
by fundamentalism we rise to the challenge. In this case, we are
called on to reinforce multi-culturalism.
Figure 1: Experimental
system producing polling like output
One
can make the argument that something is missing from those technologies
that are being acquired for intelligence vetting within the military
defense system. Clearly biodefense information awareness requires
much more that what DARPA’s Total Information Awareness (TIA)
programs contemplates. Because Biodefense must interface with medical
science, Biodefense Information Awareness (BIA) systems also requires
something different, more open to initial scholarly review of what
is planned and more open to commercialization when implemented.
Figure
2: Proposed Biodefense Information Awareness
(PowerPoint)
The
complex interior of the individual human is largely unaccounted for
in first attempts at measuring the thematic structure of social discourse.
But the individual is where demand for social reality has its primary
origin. So what is missing is the understanding of the complexity
of social discourse.
What
is missing is the available science on social expression and individual
experience. Funding is not applied, as yet, to this science because
of the narrowness of the current procurement processes that, not surprisingly,
are focused on near term issues related to continuing funding for
corporate incumbents. There is then, a bootstrap that is needed to
shift the focus from methodology and philosophy that is not sophisticated
and is not accounting for social complexity.
Why
individual variation in response patterns, for example, is not being
accommodated by commercial information technology is due to many factors.
Some of these are technical issues related to limitations in design
agility. Many of these issues are related to the commercialization
of information production and use. But the deepest problems are related
to the nature of formal systems.
These
problems have not been settled in the academia, and thus there is
no clear guidance coming from the academia. But the issues have been
laid out within a scholarly literature.
Scholarship
informs us that natural language is NOT a formal system. One would
suspect that even children already know this about natural language.
Perhaps the problem is in our cultural understanding of mathematics
and science.
Yes,
abstraction is used in spoken language; but a reliance on pointing
and non-verbal expression helps to bring the interpretation of meaning
within a social discussion, as it occurs and is experienced. So the
abstraction involved in language is grounded in circumstances. These
circumstances are experienced as part of the process of living. The
experience relies on one’s being in the world as a living system
with awareness of self. This experience is not an abstraction.
Again, even children already understand this.
One
can reflect on the fact that natural circumstances and the technical
representations, by computer programs, of context are NOT the same.
Many scholars have come to believe that circumstance is not something
that can be fully formalized. But the confusion within the academia
is profound.
Written
language extends a capability to point at the non-abstract, using
language signs, to what is NOT said but is experienced. Human social
interaction has evolved to support the level of understanding that
is needed for living humans, within culture, to form social constructs.
But computer based information systems have so far failed to fully
represent human tacit knowledge, even though computer networks now
support billons of individual human communicative acts, per day, via
e-mail and collaborative environments.
So,
one observes a mismatch between human social interaction and computers.
How
is the mismatch to be understood?
We
suggest that the problem is properly understood in the light of a
specific form of complexity theory.
An
evolution of natural science is moving in the direction of a stratification
of formal systems using complexity theory. A number of open questions
face this evolution, including the re-resolution of notions of non-finite,
the notion of an axiom, and the development of the understanding of
human induction. Induction, in counter position to deduction, is
seen as a means to "step away from" the formal system and
observe the real world directly.
It
is via this modification of constructs lying within the foundations
of logic and mathematics that an extension of the field of mathematics
opens the way to a new computer science.
Note
on Artificial Intelligence
The
“artificial intelligence” failure can be viewed, and often
is, as simply because humans have not yet understood how to develop
the right types of computer programs. This viewpoint is an important
viewpoint that has lead to interesting work on computer representation
of human and social knowledge.
But
put quite simply, a representation of knowledge is an abstraction
and does not have the physical nature required to be an “experience”
of knowledge.
The
fact that humans experience knowledge so easily may lead us to expect
that knowledge can be experienced by an abstraction. And we may even
forget that the computer program, running on hardware, is doing what
it is doing based on a machine reproduction of abstract states. These
machine states are Markovian, a mere mathematical formalism. By this,
we mean that the states have no dependency on the past or the future;
except as specified in the abstractions that the state is an instantiation
of.
There
is no dependency on the laws of physics either, except as encoded
into other abstractions.
This
fact separates computer science and natural science.
Note
on Stratification Theory
The
tri-level architecture models the relationship between memory of the
past, and awareness of the present, and the anticipation of the future.
However, once this machine architecture is in place, we still will
be working with abstraction and not a physical realization of (human)
memory or anticipation.
Stratification
seems to matter, and may help on issues of consistency and completeness,
the Godel issues in formal foundations to logic. The clean separation
of memory functions and anticipatory functions allows one to bring
the experimental neuroscience and the cognitive science into play.
The
measurement of the physical world results in abstraction. The measurement
of invariance produces a finite class of categorical Abstraction (cA),
which we call cA atoms. cA atoms have relationships that are expressed
together in patterns and these patterns are then expressed in correspondence
to some aspects of the measured events. The cA atoms are the building
blocks of events, or at least the abstract class that can be developed
by looking at many instances of events of various types.
Anticipation
is then regarded as expressed in event Chemistries (eC) and these
chemistries are encoded in a quite different type of abstraction similar
in nature to natural language grammar.
We
are arguing that the development of categorical abstraction and the
viewing of abstract models of social events, called “event chemistry”,
are essential to national security. We are arguing that the current
procurement process is not looking at and is not nurturing the types
of science that is needed in this case.
Response
mechanisms to these threats must start with proper and clear
intelligence about event structures expressed in the social
world. Human sharing of tacit knowledge must lie at the foundation
of these response mechanisms.
But
transparent human knowledge sharing is not part of the culture within
the intelligence communities. Moreover, the current computer science,
with its artificial intelligence and machine based first order predicate
logic is confused about the nature of natural complexity. Current
computer science talks about a “formal complexity” because
natural complexity has a nature that is not addressable as an abstraction
expressible as first order predicate logic. Formal complexity is
just overly complicated, and it is complicated because of the incorrectness
that is imposed by the tacit assumption that the world is an abstraction.
Putting
artificial intelligence in context is vital if we are to push the
cognitive load back onto human specialists where both cognition and
perception can guide the sense making activity. Computer science
has made many positive contributions within the context of a myth
based on a strong form of scientific reductionism. This myth is that
the real world can be reduced, in every aspect, to the abstraction
that is the formal system that computer science is instantiating as
a computer program. Natural science is clear is rejecting this myth.
Understanding
the difference between computer-mediated knowledge exchanges and human
discourse in the “natural” setting is critically important.
One of our challenges is due to advances in warfare capabilities,
including the existence of weapons of mass destruction. Another obvious
challenge is due to the existence of the Internet and other forms
of communication systems. Economic globalization and the distribution
of goods and services presents yet another set of challenges. If
the world social system is to be healthy, it is necessary that these
security issues be managed.
Note
on Urgency
We
have no choice but to develop a transparency about the environmental,
genetic, economic and social processes. Human technology is now simply
too powerful and too intrusive to allow simple economic and social
processes to exercise fundamentalism in various and separate ways.
Event
models must be derived from the data mining of global social discourse.
But the science to do this has NOT been developed as yet. We must
define a science that has deep roots in legal theory, category theory,
logic, and the natural sciences.
A
secrete government project is not a proper response to challenges
in science. In any case, the American democracy is resilient enough
to conduct proper science and to develop the knowledge technologies
in the public view. The BCNGroup.org is calling for a Manhattan-type
project to establish the academic foundation for the knowledge sciences.
The
current natural security requirements demand that this science be
synthesized quickly from the available scholarship. Within this new
science, stratified logics will compute event abstractions, at one
scale of observation and event atom abstractions at a second scale
of observation. The atom abstractions are themselves to be derived
from polling and data mining processes in order to create the abstractions.
Again,
we stress that the science needed has not been developed.
But there is a wealth of scholarship that can be integrated quickly
if only there was a small effort.
Note
on Privacy Issues
A
stratification of information can be made into two layers of analysis.
The
first layer is the set of individual polling results or the individual
text placed into social discourse. In real time, and as trended over
time, categorical abstraction is developed based on the repeated patterns
within word structure. Polling methodology and machine learning algorithms
are used.
The
second layer is a derived aggregation of patterns that are analyzed
to infer the behavior of social collectives and to represent the thematic
structure of opinions. Drilling down into the specific information
about, or from, specific individuals will require government analysts
to make a conscious step and thus the very act of drilling down from
the abstract layer to the specific informational layer is an enforceable
legal barrier that stands in protection of Constitutional Rights.
New
science/technology is needed to “see” events that lead
to or support terrorism. Data mining is a start, as are the pattern
recognition systems that have been developed. But we also need data
synthesis into information, and a reification process that knows the
importance of human-in-the-loop perception and feedback.
To
control the computer mining and synthesis processes, we need something
that stands in for natural language. Linguistic theory tells us that
language use is not reducible to the algorithms expressed in computer
science. But if “computers” are to be a mediator of social
discourse, must not the type of knowledge representation be more structured
than human language? What can we do?
The
issues of knowledge evocation and encoding of knowledge representation
shape the most critical inquiries.
According
to our viewpoint, the computer does not, and cannot, have tacit knowledge
to disambiguate natural language, in spite of several decades of effort
to create knowledge technologies that have “common sense”.
Based on principled argument a community of natural scientists has
argued that the computer will not have tacit knowledge; ever.
Tacit
knowledge is something experienced by humans. There is no known way
to fully and completely encode tacit knowledge into rigidly structured
standard relational databases.
A
work around for this quandary is suggested in terms of a Differential Ontology
Framework (DOF) that has an open loop architecture
showing critical dependency on human sensory and cognitive acuity.
An Appendix
to this paper discusses the DOF.
The war presents daunting challenges. Asymmetric threats
are organizing distributed communities to attack the vulnerabilities
of our economic system.
A large number of social organizations have organically
developed around the economic value of one to many communications
systems. Developing agility and fidelity to our information systems
is the strongest defense against these asymmetric threats. The differential
ontology framework may enable processes, which have one to many structural
coupling, to make a transition to a many to many technology.
The asymmetric threat is using many to one activity,
loosely organized by the hijacking of Islam for private hatred and
grief. The defense to this threat is the development of many to many
communication systems.
The many to many technologies allow relief from the
stealth that asymmetric threats are depending on. The relief comes
when machine ontology is used as a means to represent, in the abstract,
the social discourse. This representation can be done via the development
and algorithmic interaction of human structured knowledge artifacts.
The evolution of user structured knowledge artifacts
in knowledge ecosystems must be validated by community perception
of that structure. In this way the interests of communities is established
through a private to public vetting of perception. Without protection
for privacy built in the technology, and protected by law, this validation
cannot be successful and the technology will fail to have fidelity
to what is actually the true structure of social discourse.
Knowledge validation occurs as private tacit knowledge
becomes public. The relief from the asymmetric threat evolves because
community structure is facilitated. The validation of artifacts leads
to structured community knowledge production processes and these processes
differentiate into economic processes. These benefits can be allowed
to accrue to communities who have now, or might have in the future,
a tendency to support terrorism based on a sense of injustice. A
careful dance over the issues of privacy and justice is required.
Global repression of all communities that feel injustice is not consistent
with the strength of the American people. Our strength is in our
multi-culturalism and our Constitution.
Individual humans, small coherent social units, and
business ecosystems are all properly regarded as complex systems embedded
in other complex systems. Understanding how events unfold in this
environment has not been easy. But the war requires that science
devote attention to standing up information production systems that
are transparency and are many to many.
Relational databases and artificial intelligence has
been a good first effort, but more is demanded. Schema independent
data representation is required to capture the salient information
within implicit ontologies.
The current standards often ignores certain difficult
aspects of the complex environment and attempts to:
1)
Navigate between models
and the perceived ideal system state, or
2)
Construct models with
an anticipation of process engineering and change management bridging
the difference between the model and reality.
The new knowledge science changes this dynamic by allowing
individuals to add and subtract from a common knowledge base composed
of topic / question hierarchies supported within the differential
ontology framework.
This technology can become transparent because information
technology has matured and been refined.
We will close by addressing a specific conceptual knot.
We will untie this knot by separating issues related to natural language
use.
Note on Language and Linguistics
Language
and linguistics are relevant to our work for three reasons.
First,
the new knowledge technologies are an extension to natural spoken
languages. The technology reveals itself within a community as a
new form of social communication.
Second,
we are achieving the establishment of knowledge ecosystems using peer-to-peer
ontology streaming. Natural language and the ontologies serve a similar
purpose. However the ontologies are specialized around virtual communities
existing within an Internet culture. Thus ontology streaming represents
an extension of the phenomenon of naturally occurring language.
Third,
the terminology used in various disciplines is often not adequate
for interdisciplinary discussion. Thus we reach into certain schools
of science, into economic theory and into business practices to find
bridges between these disciplines. This work on interdisciplinary
terminology is kept in the background, as there are many difficult
challenges that remain not properly addressed. To assist in understanding
this issue, general systems theory is useful.
These issues are in a context. Within this context,
we make a distinction between computer computation, language systems,
and human knowledge events. The distinction opens the door to certain
deep theories about the nature of human thought.
Within existing scholarly literatures one can ground
a formal notation defining data structures that store and allow the
manipulation of topical taxonomies and related resources existing
within the knowledge base.
In
Summary
The differential ontology framework consists of knowledge
units and auxiliary resources used in report generation and trending
analysis. The new knowledge science specifically recognizes that
the human mind binds together topics of a knowledge unit. The new
knowledge science holds that the computer cannot do this binding
for us. The knowledge science reflects this reality. The rules of
how cognitive binding occurs are not captured into the data structure
of the knowledge unit, as this is regarded as counter to the differential
ontology framework. The human remains central to all knowledge events,
and the relationship that a human has with his or her environment
is taken into account. The individual human matters, always.
Appendix: Differential
Ontology Framework
