|
ARTIFICIAL GENERAL INTELLIGENCE |
||
THE RELATVITY OF KNOWLEDGE - A UNIFIED SYSTEM.Artificial Intelligence is a term which infers that somehow a computer system is smart. If a machine is working on a problem, however, it may get stuck because the database it has, has got too many gaps to produce an answer to the problem. The smart solution is to get the information from some other database - preferably any database - to fill in the gaps - the technical term is isomorphism - or relatedness, between different Domains - or data topics. This can only happen if all the knowledge being used or which could be used is represented in a special way. Tripartite Essentialism Expert System [TREES] has a unique way of doing just that. Basically a System of [eg. Prolog Trees] with [TR] Semantics anchors a Query over a Set of Maps. These Maps are separate Domains of Data with a systematic ontology. The query is answered by pulling relevant data from any suitable Domain or Map The secret is in the Knowledge Representation System It is now possible because of Limited Independent Variety and the closure of recursion, for a computer to be not just smart, but executive and autonomous - also, as an application for this type of process [cf. isomorphism] there need only be the one Software System for every application, without the need for the hundreds of more specific and expensive software and hardware solutions developed specifically for the one application which is what is in store for us at the moment - because of the more ad hoc solutions created for the market place. It is the unique knowledge representation system of [TREES] which produces the one solution to the great diversity of A.I. development problems with isomorphism. Tripartite Essentialism takes Knowledge or Semantics and orientates concepts in the database in such a way that any one set of concepts can be packaged up in a limited and closed way. The importance of this 'closure' is that a database search for a particular or specific set of concepts is not going to take 'forever', which is what can happen at the moment in A.I. and Industry. To be more specific, at the heart of [TREES] is a Concept Representation System, which is a limited and closed description of any given database topic. This concept representation system has an Ontology dependent on a Universal Natural Law [& Empirical] and is Logically Complete. [TREES] is a software for an A.I. system or Robot which would be expected to look for alternative solutions to predicaments which arise in any solution strategy or process. In the terms of Artificial Intelligence, this powerful 'inference engine' and 'shell'. promises a great range of computational powers and applications as a software because the way that it represents knowledge does not tie it to any specific 'domain' or data topic. ie. it is Universal. All knowledge is represented in the format, 'Essentially' of one Procedural Rule with various Empirical consequences such that relativity can be deduced from a rule based deduction system. The domain is formatted in 'Trees' of 'tripartite atoms' whose 'epistemological adequacy' is synthetic a priori valid. This is an Important criterion, for ALL previous system theory applications can only claim a posteriori epistemological status. The knowledge system or tree is constructed from a dynamic process of direct expectation-driven reasoning. The level of practical theory from the compiled knowledge would be 'very deep' and the Tripartite Relativity knowledge Representation System has an 'operationalization value' that would eliminate referencing problems and recursion. Designing a database utilising this tripartite format would still make practical all the conventional software options, but on adding a Tripartite Relativity perspective, - the emphasis being on ergonomics and the relativity of function, there would be added to any database the perspective of another empirical context. The ramifications for 5th Generation A.I.
and Data Warehousing/Mining etc can be demonstrated in more detail at a
consultation.
3. More examples of Tripartite Essentialism Semantics. Food, has as its context the Sun, which powers the green cycle and turns the available Carbon, Nitrogen and Phosphour into Carbohydrate Sugars and oxygen. This series of trophic and vegetational cycles produce a complex ecosystem and food web from which is extracted for the benefits of the gastronomical structure and mechanics of Man, energy packets of variable relative quality and quantity. A kitchen receptacle, when examined holistically with Tripartite Relativity has a macro or function that suits it for the context of the kitchen in design and purpose. The size of the bowl, the energy of the umbrella that it provides to sustain its utility - the macro or utility for the conveyance of fruit. The Meso of the bowl is basically its components and structural constituents, and the Micro of the bowl, basically, its evolutionary assets are its qualitative aspects eg. if the purpose of the bowl was solely general utility, then plastic is an asset to enable continuity of function; if however, its quality is not selecting it to attain conditions of continual use eg. porcelain bowl at a banquet may be more acceptable, then the Micro of the bowl is dependent on context conditions and the amount of psychological pressure its system can 'bring to bear' or weather in the selection process. The cupboard should, like the bowl, be seen in terms of the function that makes it a specific type of cupboard, and not necessarily seen in terms of the energies of its manufacture. (The cupboard is the end product (Micro) of a Sawmill (Meso) that manufactures to differing qualities from forests (Macro). The context of the cupboard per se would be its overall capacity to perform its container function of holding utensils eg. bowl. Its Meso would be its structural arrangements and partitioning and fitments and the evolutionary assets or qualitative aspects - Micro, would be the difference between Oak and Hardboard, plastic and brass. A record player has a Macro of electricity, capital and the context and function of an information transformer. The Meso or structure and mechanics of the hardware would have a qualitative aspect of HiFi. A Book has as its context its topic and a Macro determined by the information formats and ingredients of its source, Its Meso or structure and mechanics are the ideologies or idea formats that collate and relate it to the general purpose and function of the book. Whereas the evolutionary asset of the book, its Micro, would be how effective and intelligent an ideological tool it was for the exploitation of the resources of the context to which its main premise was applied. A lamppost has as its context, the illumination of infrastructure for the purposes of efficient society etc, and as a Macro both the capital of taxation and the power of electricity, its Meso is the structure and components of the object itself, and its qualitative aspects are mainly aesthetic in relation to how much light it provides, how high and the quality of both structure and function. A Home has as its context the housing of a family unit for the purposes of labour maintenance within the context of the infrastructure - the family or Macro will maintain the input of capital, fuel, energies, supermarkets, reservoirs, petrol, and these will be utilised by the structure and mechanics of the social context to be invested in maintaining and sustaining the fabric and quality of the social environment. The Qualitative aspects are related to capital input, class and other types of ideology (aesthetic). A car has as its macro, the social
infrastructure, capital and energy which facilitate its existence.eg.
roads, petrol station, funds, tax, society. Its meso is its structure and
mechanics, eg, the metal, engine and components, and its Micro, or
evolutionary assets, are the qualitative difference eg. between a mercedes
and a lada. [TREES; Tripartite Essentialism Expert System]
The volume of world data and its multiplicity of
domains makes it very difficult to extract meaningful relationships that
have integrity.
Thus the science of data warehousing and data
mining are born.
If there were some way of instilling 'common
sense' and scientific values into the classification of knowledge, the
extraction of meaningful data would become a cost effective reality.
Systems thus far have dealt with labels, object
labels, ill-defined
Relationships and sketchy searches. [cf. 'a
posteriori' ] dealing with ideas that are self referential and all the
system design and development hang ups that that entails.
Tripartite Essentialism - a Knowledge
Representation System by Andrew Hennessey - makes possible the Unification
of Data with rational [scientific] empirical labels/values.
From the first sections, it can be seen that
concepts/knowledge can be presented in terms of the way it functions and
relates to and within its domain. [in three ways]
This kind of conceptual, functional, relativity,
which describes some system or event as it functions with and in the real
world, when united with its scientific description eg1. Banana, eg2.
Newspaper
eg.2 newspaper.
Perhaps such an objective system would be the
'Holy Grail' of the world of data mining and warehousing.
To make this ideal come about, I would wish to
make use of all the knowledge/data that is currently in the public domain.
I would wish to devise a search engine that could
identify and pick out data context, subject area, and the parcel of
information to which it is linked. Then in terms of Tripartite
Essentialism, it would seek out and allocate to and within this
information; - the empirical specifications of the context, the macro, the
meso, and the micro.
It might be thought that the heuristics -
criteria used to narrow down to the specifics, would be impossible to
devise for such a task, however, Tripartite Essentialism [Hennessey 1990]
has a meta-language to make this a realistic proposition.
Given that the criteria for syntax analysis of
Grammar already exists in some form in modern Word Processing packages,
then the task critical for the success of this System is in the allocation
of the information to; - Context, Macro, Meso, and Micro.
In terms of Tripartite Essentialism semantics;
Where the Umbrella and Backbone of a concept - is
the intended labelled Object or Noun. [macro]
Where the deployment of processes and activities
of that Object, are Verbs, [meso],
And the Qualitative evaluation of that object and
its processes in relation to some Context and some Time, are Adjectives
[micro]
The problem of repackaging self-referential and
unwieldy data could now have an attainable solution using Tripartite
Essentialism semantics.
Eg. [I propose..] A search engine programmed to
specifically explore, chart and extract information according to a
context-specific glossary and set of empirical standards - this would
comprise the first phase of a sorting process. etc
Further sorting either by 'physical/empirical
values' - as suggested and or by a general domain overview built in to the
original glossary could translate bulky data now in a more time consuming
'linear' form into a more quickly accessible set of functional
relationships.
It comes down to the way we see an object -
Tripartite Essentialism would maintain that all objects, [and all labels]
have an empirical scientifically defined context, a macro, a meso, a
micro: in terms of their perceived function.
A Domain - a specialised area of knowledge, can
be defined as full of objects, labels and measurements in some unit of
measure.
From the earlier sketch of the TRE Knowledge
Representation System where we see a three part atom develop into 9 at
generation 2, then 27 at generation 3, 81 at generation 4 etc -
If we look down on this system topographically,
then Generations 1-3 could be at the core of a map - [its Macro],
generations 4-7 could be around the core [its Meso] - and generations 9+
at the periphery [its Micro].
Each Knowledge Representation System would take
on a morphological character because of its empirical content, and take on
other characteristics according to complexity of semantics/sets of things
associated with the physical data. [eg. relatively simple or complex, list
structures, sub-sets, and internal relations/data structures]
Each domain would have its own characteristic
topography, but would be able to share and swap similar features between
maps - so called 'isomorphism between
domains' .
The analogy is 'Virtual Geography' since the maps
can be expected to depict various aspects of relativity within their
closed and limited series.
Eg. relationships between topography and
complexity, or between magnitude of energies depicted by empirical data
and the quality or suitability of data for isomorphism to be expected.
This feature in a working form is much sought
after in the world of artificial intelligence - only made possible in a
universal application at this time by Tripartite Essentialism,
[Hennessey 1991]
Recent work by Dr Peter Plichta. isbn
1-86204-014-1 pub element 97 has proven that threeness is a matter of a
priori mathematical fact, and renders Tripartite Essentialism a system
with non-arbitrary structure.
In view of this there is a strong case to suggest
that the maps of each domain would be ontologically sound and archetypal.
These maps would be able to support extensive
scientific testing because of their Logically sound ontology. TRIPARTITE RELATIVITY KNOWLEDGE REPRESENTATION SYSTEMCONTENTS. [a] abstract 1. Introduction 2. The Relativity of Knowledge
Abstract: The TRE semantics developed by
Hennessey 1991 allow a completely new kind of artificial intelligence and
philosophy of mind to come about. The semantic system can take the human
knowledge database to the very limits of language and semantics - not only
allowing fully functional and executive A.I. but creating new frontiers in
epistemology and the philosophy of language.
1. Introduction.
At the heart of the Semantics of TRE is a simple division of a system into a description of three aspects of function. A division which can be Universally applied to all systems, items and descriptions of objects. A brief description of the Semantic model. [fuller one follows with examples] System1 is denoted 'Macro', it is the umbrella of basic constituents and energy source from which emerges an energy transfer. Eg. the Sun this energy is then processed by the intervening system, denoted 'Meso' which has Structure and Mechanics eg. grass, that acts on system2 eg. cow, the 'Micro', which is the Evolutionary Asset or product of this chain of relativity. The cow is considered an evolutionary asset in some 'Context', eg. human/sociological. This chain of relativity has given a general evolutionary opportunity for a herbivore. Here are a few more examples of
Tripartite Relativity .(ie. Three parts and a context ie. Context and
Macro: CMacro, Meso, Micro.)
The relationships are relationships between aspects of function. The ramifications of this relativity are that a developed system of many domains would have a limited set of events - for the possibilities produced between and amongst the activities of TRE systems are finite and closed. Since we are dealing more with the Limited Independent variety of ‘archetypal’ states rather than the infinite variety of labels there are in the Cosmos. This has beneficial consequences for heuristics. Also, by relating all labels to their empirical function, we can introduce physical scientific data as another kind of heuristics which can be tagged on to the concepts and domains that we are dealing with. Eg. Millivolts, Gigajoules, Solar Year,
half life - [scalar and vector components etc]
When the Tripartite semantic algorithm is continuously applied to a concept ie. Macro1, Meso1, Micro1, then Macro2 of Macro1, Eg. the macro of an ant is the forest, the macro of the forest is the mountains/landmass, the macro of the landmass is planetary tectonics/planetary energy, the macro of planetary energy is stellar energy etc, Then ……….. Meso2 of Macro1 Micro2 of Macro1
Macro2 of Meso1 Meso2 of Meso1 Micro2 of Meso1
Macro2 of Micro1 Meso2 of Micro1 Micro2 of Micro1
then J .... Macro3 of Macro2 of Macro1 Meso3 of Macro2 of Macro1 Micro3 of Macro2 of Macro1
Macro3 of Meso2 of Macro1 etc etc
It looks like a very big Knowledge Tree. By the time the concepts are instantiated, developed by the semantic algorithm - round about generation 4-6 - the conceptual details are very complex, but at generation 10 or so, the concepts reach extinction or unity or monotony - and the heuristics have reached a limited ceiling. I have some examples that I charted out on A1 which can demonstrate this to you. A brief description of what happens on this chart follows. The Context must first be chosen, for it
shapes the development of the concepts within the Tripartite framework as
the 'Tripartite semantic algorithm' is applied - used to produce - other
'generations/instantiations/orders within the 'tree or database'.
In the case of a context for a social paradigm, for example; First order macro [1], meso [2], and micro [3] produced ; 2nd order [1] information [2]
substance of social infrastructure and mechanics [3] Qualitative aspects
of social processes.
Finding the macro/umbrella, meso and micro of each set in turn, produced a 3rd order pertaining to; [1] the umbrella of information in a social context, and the physical processes which facilitate its value, development and production. [2] the umbrella of 'infrastructure' in geography - both physical and psychological systems [3] the enabling aspects of social
mechanisms, infrastructure and product eg. market/skills/social product/artefact.
4th order produced; [1] the umbrella of time/psychological and intellectual product [2] physical parameters of physical/social/intellectual fabric [3] relative efficiency/ qualitative
aspects of empirical experience
5th order; [1] Biological and existential aspects of Being [2] sub-atomic physics/metaphysics [3] aspects of the Philosophy of Mind
………….. and by the 6th generation, the TRE System is addressing DNA and the physics of Chaos - and by the 8th - the extrapolation/generation we are dealing with has come up against First Cause, Divinity and the stranger aspects of Philosophy of Mind …. There are absolutely no limits to the kind of domain that this system can tackle. I believe that confronted with a vast
mountain of data, these semantics/heuristics can reduce access time and
increase the relevance of results as they will always halt at a fixed
point on a search chain of fixed and limited length.
|