- Category: Theory
- Published: Wednesday, 01 May 2013 01:58
- Written by Ken Ewell
- Hits: 12909
In order to measure the relevance or meaning of natural expressions found in text, Readware computes transformation functions (selected from a small class of non-recursive functions) from naturals to naturals. The result is the abstract product of mapping a denotative sign of a given semiosis (sign-system) onto the natural attributes and processes of poiesis, whereby both semiosis and poiesis are seen as natural processes. This transformation can also be performed in one's head with a little practice.
Semiosis can be understood simply as the systematic processes used for signifying, representing and denoting meaning. An alphabet and compositional writing system form such a sign system or semiosis, although people rarely think of the alphabet that way. The term "meaning," while seriously ambiguous, is defined for this context as a basis, a conjecture, and as a quantifiable measure of significance, salience and relevance.
Unlike semiosis, poiesis in not about signs; it is wholly about making (things, meaning, relations, etc.). Poiesis can be understood as the elemental dynamism of the production or creation and projection or manifestation of physical and conceptual entities amid their relations to (i.e. in recognition of) salient attributes of the environment. Both the processes of semiosis and of poiesis are seen as necessary to an organism that is making meaning and producing information.
Many computer scientists, AI and language researchers, are familiar with semiotics through the work of Charles Sanders Pierce. Many cognitive scientists, "consciousness" and "mind" researchers, along with a good many biologists, are familiar with semiosis from Jesper. Hoffmeyer's book "Signs of Meaning in the Universe" among other works. AI researchers may be familiar with autopoiesis through the work of Chilean biologists Humberto Maturana and Francisco Varela. In addition to this semiotic and poietic foundation, Readware implements a unique and super-recursive algorithm. In this way, it can be claimed that Readware is more powerful than a Turing machine.
In computability theory, super-recursive algorithms are a generalization of ordinary algorithms that are more powerful, that is, compute more than Turing machines. The term was introduced by Mark Burgin, whose book "Super-recursive algorithms" develops their theory and presents several mathematical models.
According to Wikipedia, "a computational model going beyond Turing machines was introduced by Alan Turing in his 1938 PhD dissertation Systems of Logic Based on Ordinals. This paper investigated mathematical systems in which an oracle was available, which could compute a single arbitrary (non-recursive) function from naturals to naturals. He used this abstract device to prove that even in those more powerful systems, undecidability is still present." Even though Readware is physically implemented and not only abstract, not everything is decidable.
While Turing's oracle machines are strictly mathematical abstractions, Readware was conceived as an experimental instrument needed to test the abstract interpretation processes.
Tom Adi designed an "oracle machine" for use as the "inductive engine" of the software called Readware; it was also called the Readware Analyst because this induction machine recognizes knowledge types by reading and learning (connecting, storing and updating its model) while doing a semantic and structural analysis of a text. Another machine called the query processor accepts free form questions relative to the text; having learned, and "comprehended" the ideas from the knowledge types found in the text using the analyst. A reader/browser highlights the answers to inquires which are displayed within their context.
In complexity theory and computability theory, an oracle machine is an abstract machine that is also used to study decision problems. Judging the relevance of any given expression (in a given context) is a decision problem that normally falls under one's own powers of discretion. We tested our system experimentally and formally in the Text Retrieval Conference (TREC) sponsored by the National Institutes of Standards and Technology and DARPA. The results have been peer-reviewed.
An oracle machine can be conceived as a Turing machine connected to an oracle. The oracle, in this context, is an "entity" capable of solving some problem, which for example may be a decision problem or a function problem. Readware solves both types of problems. The entity, the oracle, (in our case): is an apparatus capable of "reading" or "comprehending" the configuration of the attributes of each description or instance using an inheritance (semantic) rule. Such an organism (O) is fully and technically described by Adi's Theory of Semantics (ATS).
That is, Readware's decision problem is represented as a set A of natural language strings: a natural language description of something. An instance of the input problem is an arbitrary yet denotative natural language string. The solution to this problem is to determine the primary attribute or attributes in the context-- the 'productive reference' that is phonetically denoted in the instantiation (i.e. the poietic conditions of the context of a knowledge K).
It is important to grasp three domains of abstraction; semiotic, poietic and semantic: Readware's function problem is hardly different from that ordinary problem which each of us face. How to read the signs and comprehend what is happening. How to transform the confluence of semiotic and poietic factors into more rational or determinable quantities. Readware's problem is much easier because it is designed only to work on text,-- to recognize text and make decisions about the signs in the text.
Readware's decision problems are cast into solvable formulas selected from a model M (M=K) by mapping the phonemes and signs (the semiotic entitles) of set A (representing physical and conceptual entities) to poietic elements (of poiesis or production and salience). That is, these abstract (mental) processes are activated upon an instance of recognition.
What Readware recognizes are the elementary attributes and processes in the domain and range of a Knowledge K of creation or production (i.e. in the manifest image) of the (semantic -- rule-making) organism or mechanism O. This organism O comprises the intelligence or arrangement informing the rational inductive and deductive computing processes: affording the software, in effect, a capacity to comprehend; to behave as an intelligent system or oracle. Is this the way people read?
Psychologists believe there is a molar schemata or system that correlates sensations and experience in the minds of human beings. But we have no knowledge of this.
Readware grew out of what was initially devised as an experiment; a way to test Adi's theory of semantics. ATS states that each of the 28 Arabic consonants is a sign that refers to one of seven abstract processes—assignment, manifestation, containment, assignment of manifestation, assignment of containment, manifestation of containment, and assignment and manifestation of containment—and one of four abstract polarities —closed-self, open-self, closed-others, and open-others (Adi, 2007, pp. 185, 190-191). This is visualized in Table 1.
This is, in effect, a kind of molar schemata for interpreting Arabic or English sounds. One that determines the elementary processes and attributes to the environment of poietic awareness -- or an awareness of the environment of a Knowledge K of creation (making, producing). Putting it either way is correct, as knowledge is, its essence is as, a production of information.
Using this apparatus, symbolic approximations (inputs) are transformed into a concise and quantifiable metalanguage; a more efficient description of the knowledge K;-- a recognition of the necessary interrelations of the attributes in the environment. In a practical and pragmatic sense the attributes of the environment are conditions that serve to regulate meaning (significance, relevance) by unifying (seeking to harmonize) the connotative and semiotic creations or productions in a given context of the environment --with the poiesis of a self-organized knowledge and awareness. Almost needless to say, this notion is difficult to grok, though it is readily warranted.
In order to not be overcome by the immense ambiguity in modern languages, Adi's semantic study was performed on the classical Arabic language; then it was linked by analogy and derivation to most modern languages with an alphabetic system of symbolic notation. Unlike the mostly borrowed word-roots of modern languages, almost all Arabic words are formed from original word roots that are strings of three consonants each, lending themselves to the aim of avoiding ambiguity.