Back to Knowledge
Technologies and the Asymetric Treat
Differential Ontology Framework
Appendix to : Knowledge Technologies and the Asymetric Treat
Paul S.
Prueitt, PhD
3/16/03
Computers
are not now, and likely will never be cognitive in the same way as humans
are. However, the notion that
cognitive-like processes might be developed has brought some interesting
results. At the core of the solutions
that are been, or are likely to soon be, made available is a class of the
machine-encoded dictionaries, taxonomies and what are often called
ontologies. While encoded dictionaries,
taxonomies and ontologies are often referred to as knowledge representation,
one has to be careful to point out that knowledge is “experienced” and that
these computer data structures are experienced only when observed by a
human. Some experts in the field
express the notion that computers do or will experience, and this is the
so-called strong AI position.
The
rejection of strong AI opens up an understanding that knowledge definition,
experience and propagation requires a greater degree of agility and
understanding by users of the processes that have been developed by the strong
AI community over the past 50 years.
There has been a failure to communicate.
The
strong AI community, on the other hand, should eventually come to support the
principled arguments made in the quantum neuroscience community. Quantum neuroscience has an extensive
literature that reports on issues related to human memory, awareness and
anticipation. This literature is
referenced in separate technology volumes, simply because the purpose of this
paper is to develop concepts not to provide what should be a complete citation
of the literature (as opposed to a very partial citation that might be
misleading.) However, we point out that
the quantum neuroscience literature addresses the question of how things happen
and in doing so this literature challenges classical Newtonian physics as a
model of natural complexity.
The cause of events has a
demand and a supply component.
Demand-side knowledge process management can be used to create
machine-encoded ontologies. These
formal constructs, that are the states of computer programs, are constructions
that exist in two forms, implicit and explicit. Implicit constructs are defined as proper continuum
mathematics. The constructs of
continuum mathematics are represented on the computer in a discrete form. The discrete form of implicit ontology has
the benefit of the precision of the embedded formal continuum mathematics, seen
in mathematical topology, mathematical analysis. Demand is a holonomic, e.g., a distributed, constraint like
gravity and so a distributed representation is needed. The continuum mathematics provides the
distributed representation that the computer “cannot” provide by itself.
An explicit ontology, on the
other hand, is in the form of a dictionary or perhaps a graph structure. The constructs of explicit ontology are
expressed as discrete mathematics, graph theory, number theory, and predicate
logics. The interface between discrete
mathematics and continuum mathematics has never been easy, so one should not
expect that the relationship between implicit and explicit ontology be easy
either. However, a formal theory of
by-pass has been developed, by Prueitt, which shows a relationship between
number base conversions, data schema conversions, and quasi-axiomatic theory
[1].
First-order predicate logics
are often developed over the set of tokens in some of the explicit ontology
that exists, for example by using a standard resource description framework
(RDF). Value is derived, and yet these
ontologies with their logics suffer from the limitations of explicit enumeration
and relational logics. This limitation
is expressed in the Gödel theorems on completeness and consistency in formal
logics [2] as well as in other literatures.
By-pass theory is designed to manage this problem.
The bottom line is that high
quality knowledge experience and propagation within communities needs the human
to make certain types of judgments. But
the methodology for human interaction with these structures is largely missing,
and is certainly not known by most of those who need now to use these
structures for national intelligence vetting.
We introduced the term
“Differential Ontology” to talk about a human mediated process of acquiring
implicit ontology from the analysis of data.
Figure 1: Differential Ontology Framework
By the expression “Differential
Ontology” we choose to mean the interchange of structural information
between Implicit (machine-based) Ontology and Explicit (machine-based)
Ontology.
•
By
Implicit Ontology we mean an attractor neural network system or one of
the variations of latent semantic indexing.
These are continuum mathematics and have an infinite storage capability.
•
By
Explicit Ontology we mean an bag of ordered triples { < a , r, b > }, where a and b are
locations and r is a relational type, organized into a graph structure, and
perhaps accompanied by first order predicate logic.
The differential ontology
framework uses implicit ontologies now found in stochastic and latent semantic
indexing “spaces” and derives a more structured form of dictionary type
ontology. The class of process
transformations between implicit and explicit forms of machine ontology is to
be found in various places. For
example, we have seen market anticipation in the large number of automated
taxonomy products that are appearing in the marketplace. However, we claim that the notion of
deriving explicit ontology from implicit ontology is an original
contribution.
Within the explicit ontology
there are localized topics. These
localized topic representations “sit’ by themselves. In the implicit ontology
the information is distributed like gravitational wells are in physical
space. Move anything and everything
changes; sometimes only minutely and sometimes catastrophically. The perturbations of representation are then
formally seen as an example of a deterministically chaotic system. Natural systems may or may not be
deterministic, and thus the argument is that formal chaos is not the same as
the process seen in the emergence of natural systems in the setting of real physical
systems. Neural network type attractor
manifolds is an early form of the implicit ontology. But now we have also genetic algorithms, the generalization of
genetic algorithms in the form of evolutionary programming and other
mathematical constructs. The tuning of
these systems to the physical reality is thus of major consideration. This tuning is not a done-once for-all-time
task. As the world changes, the tuning
of implicit ontology must change also.
How is this to be done?
Moving between these
implicit and explicit forms of machine ontology "state gesture
mechanisms" can be attached to topic constructs, whether distributed or
localized, using new type of “stratified” architecture [4-6] that is grounded
in cognitive neuroscience, specifically in the experimental work on memory and
selective attention [7,8]. The
stratified architecture has multiple levels of organization, expressed as an
implicit or explicit, continuum or finite, state machine, in which within each
level certain rules and processes are defined.
Cross level interaction
often MUST involve non-algorithmic [9] movement within the state spaces. Thus a human, who CAN perform
non-algorithmic inference, is necessary if the over all differential ontology
system is to stay in tune with the external complex world. This means that implicit and explicit
formalism is NOT sufficient without real time human involvement. A state gesture mechanism drives the information
around that part of the system that lives in the computer, but the demand for
this supply comes from within private personal introspection, perception and
decision-making by humans. Again, our
viewpoint seeks to return responsibility, for control actions in the world, to
the human and push away the notion that the crisp and precise states within
computer information systems can reasonably be managed outside of this
responsibility.
State
gesture mechanisms allow the machine ontologies to be “assistant-to” human
decision making. The mechanism is not a
closed formalism or an information technology, but rather an intellectual
framework with stratified theory and cognitive neuroscience as the framework’s
grounding [10].
Following
architecture, and design of a related applied semiotics theory developed by
Pospelov and his colleagues [11], the state gesture mechanism itself is deemed
largely subjective. This means that the
mechanism is not reducible to algorithms.
A ‘second order’ cybernetic system is required that is primarily
controlled by direct human intervention.
So the machine ontologies are required to be sub-servant to a human
knowledge experience process. The
details of the machine ontologies are visible as reminders or signed informants
within a sign system or natural language.
It is for this reason that one can call this a knowledge operating
system.
Elements of
micro-ontologies, that are formative within the moment, act as signs about the
knowledge process that are regularly occurring in social networks. The differential ontology framework requires
a human interaction with these signs.
That is the Demand side.
Ontologies are streamed using formatted micro-ontologies from point to
point in the knowledge ecosystem resident to the community of practice. This is the Supply side.
A computational mirror of the
state gesture mechanism is also “assistant-to” the location where information
is moved within the enterprise. So the
mechanism acts as an automated reporting and assessment technology. This assistance has great value to social
and economic processes. Productivity
goes up. Social value goes up. The fidelity of knowledge representation
goes up. Return on investment goes
up.
Responsibility can, and
should, be assigned exactly where responsibility is situated. This is true, in particular, where Constitutional
rights to privacy are involved. A
knowledge operating system, such as the Knowledge Sharing Foundation, can
provide complete transparence to all instances where a human examines private
knowledge.
As an example, an evocative
question can be answered in various ways.
The answers made are part of a natural language generation capability
that involves the emergence of word structure from a mental experience. This emergence must involve a human
awareness, and the action of requesting the computer supporting processes can
be a trigger to reporting mechanisms about the inspection of private
information. This the barrier to
inspection is transparent.
The structure of
differential ontology mediated communication allows a mechanical roll-up of the
information into one of several technologies for natural language generation
from semi-structured information.
However, the process is formally underdetermined, and some constraints
must be imposed during the construction of the natural language response. One sees the same situation as the
mathematics of the wave functions in quantum mechanics. The formal mechanism for “collapsing the
wave” and realizing locality cannot be derived from the classical notions, and
perhaps cannot be derived at all in a single formal system (having both
completeness and consistency). Many
post-modern scholars of physics have equated this problem with the notion of
finding a single unified theory of everything [14].
Differential
micro-ontologies represent the first of a new generation of more complete
(supply side plus demand side) knowledge process management paradigms. Machine based ontologies generate natural
language that communicates to users in a way that is familiar. This drives knowledge sharing in a new way.
What is surprising about
this vision is that it is NOT the popular Artificial Intelligence vision of the
future. The “real” future brings the
human more fully into contact with other humans within what is essentially a
Many to Many (M2M) communication device.
Advanced notation
for Formative Ontology and Differential Ontology
References:
[1]
Finn, Victor (1991). Plausible Inferences and Reliable Reasoning. Journal of
Soviet Mathematics, Plenum Publ. Cor. Vol. 56, N1 pp. 2201-2248
[2]
Nagel, W and Newman, J. (1958). Godel’s Proof. New York University
Press.
[3]
Prueitt, P.S. (1995) A Theory of Process Compartments in Biological and
Ecological Systems. In the Proceedings of IEEE Workshop on Architectures for
Semiotic Modeling and Situation Analysis in Large Complex Systems; August
27-29, Monterey, Ca, USA; Organizers: J. Albus, A. Meystel, D. Pospelov, T.
Reader
[4]
Edelman, G. M. (1987). Neural Darwinism. New York: Basic Books.
[5]
Prueitt, Paul S. (1996c). Structural Activity Relationship analysis with
application to Artificial Life Systems, presented at the QAT Teleconference,
New Mexico State University and the Army Research Office, December 13, 1996.
[6]
Prueitt, P. (1998). An Interpretation of the Logic of J. S. Mill, in IEEE
Joint Conference on the Science and Technology of Intelligent Systems, Sept.
1998.
[7] Levine, D. & Prueitt,
P.S. (1989.) Modeling Some Effects of Frontal Lobe Damage - Novelty and
Preservation, Neural Networks, 2, 103-116.
[8] Levine D; Parks, R.;
& Prueitt, P. S. (1993.) Methodological and Theoretical Issues in Neural
Network Models of Frontal Cognitive Functions. International Journal of
Neuroscience 72 209-233.
[9]
Prueitt, P. (1997b). Grounding Applied Semiotics in Neuropsychology and Open
Logic, in IEEE Systems Man and Cybernetics Oct. 1997.
[10]
Penrose, Roger (1994). Shadows of the
Mind, Oxford Press, London.
[11]
Pospelov, D. (1986). Situational Control: Theory and Practice. Published
in Russian by Nauka, Moscow.
[12]
Park, S; Hunting, S; and Engelbart S; (Editors) (2002). XML Topic Maps: Creating and Using Topic Maps for the Web. Addison
Wesley
[13] Maturana and Varela; (1989) The Tree of
Knowledge
[14]
Nadeau, R, Kafatos, M; (1999). The
Non-local Universe, the new physics and matters of the mind. Oxford University Press.
[15] Albright, M; Kohut, A (2002). What the world
thinks in 2002, The Pew Global Attitudes Project, The Pew Research Center
for the People & the Press.
[16] Prueitt, Paul S.; (2002). “Transformation of Knowledge Ecology to a
Knowledge Economy”, KMPro Journal.