On the separation of syntax
and semantics and the issue of (real time) pragmatics à
Wednesday, August 18, 2004
Is one of the
differences between the Semantic Web and the Anticipatory Web
Comment by Dr. Richard
Ballard, Founder of Knowledge Foundations Inc
Response by Dr. Paul
Prueitt, Founder of BCNGroup Inc
Still being
edited: 12:01 PM
Footnotes inserted by Prueitt
This <a, r, b> thing of yours has been unflinching
mess that you will not put away or define in any self consistent way [1].
Sometimes r is a relationship, now apparently its a variable [2].
The first use of r is apparently as a first order relationship between 'a' and
'b'. Then r has something else that is about a and b as a function of r?
Declare yourself, are you describing a semantic network or some kind of functional [3]. This looks like a mess because you keep trying to have <a, r, b> become a catch-all for both first and second order relationships. If so, I repeat then [4]
my n-ary objection, conditional realities
(n-aries) are not reducible to an assembly of first order relationships.
I also repeat the one law that unfortunately always seems true,
mathematicians possess no representations for
reality, because they systematically deny its existence.
If "a" and "b" are entities in a
semantic web, then of course they are "locations", but why isn't 'r'
a location too. Is this some database UML thing, where relationships cannot
also be entities too. There are no
right answers to poorly posed questions.
Dick
Richard,
I agree with what you are saying. The n-ary is a more powerful form for encoding localized knowledge facts because the "reduction" of the n-ary < r, a(1), . . . , a(n) > to a set of binaries { < a, r, b > } is complicated and losses information.
The notation for the Orb (Ontology referential base) is like the notation for arithmetic, one can make definitions based on an extension of arithmetic that allows for generalizations of various types. In fact I might believe that the Knowledge Foundation’s work on n-ary based knowledge encoding can be fully expressed in a way that is consistent with the existing notation if specific generalizations are made. I have often thought about how the introduction of the notion that the relationship in an n-ary < r, a(1), . . . , a(n) > might be given a dynamic element. The relationship can itself be thought of NOT as a node, ie as a a or b, but as something that has no location or initial condition. The relationship then is a model of an ecological affordance, and as such is like the valance in chemical compounds. How the valance gets satisfied is then dependant on the environmental conditions when the compound forms.
In some cases, the compounds have multiple metastable states that create radically different functional properties without altering the elements that “compose” the compounds substructural elements. How might we have a type of n-ary cognitive graph, what ever that means, with notational means to express the transitions of compounds between metastable states?
Once one has localized knowledge facts encoded as n-aries then the convolution theory partially developed by the CCM patient (Applied Technical Systems Inc) and deployed at Army Intelligence, can be applied to n-aries. The notion of path in a n-ary “graph” is where some interesting open questions exist for me; but I think you have worked on this so you know how to solve a path over all of the data problem.
If you do, then the Knowledge Foundations Mark 3, your
near future knowledge base engine, can reside on the Hilbert technology. You have other plans I know, and so we will
all have to wait and see.
The work that I am proposing will encode n-aries. The Readware 32 element substructural
letter semantics can be mapped in real time testing to your 18
element semantic primitive framework. (Which perhaps you will explain again for the gathered audience.)
I know that you have an encoding mechanism and some advanced thought on how the get n-aries in the first place. The actions involved in acquiring the n-aries and encoding them into computer memory are two different problems, which I talk about as a separation of syntax from semantics.
I know that you claim that your syntax is the same as semantics, that your semantic primitive framework is correct. But this proof of correctness is in your mind and has not been demonstrated in the marketplace (has it?) I do not reject the claim you make, but I am open to testing.
But your system is a knowledge construction system, that
depends on humans, and that cannot create ontology from text. It might be able to assist in knowledge
management within virtual communities but an automated construction of taxonomy
or ontology from social discourse that is expressed in natural language is what
you have implicitly rejected (if I understand your position correctly).
By
functional, I am not sure you understand why I say that the r is a relational variable. I mean that the relationship between two
referential nodes in an Orb is not constant, the relationship between two
things varies and so the referential token that represents the reality, as is
known or can be represented, must "take" on referential values that
differentially indicate the specific relationship that is deemed representative
between the two referential nodes at that time and in that context.
Real time pragmatics and categorization issues are
involved, as you know. This is a
discussion that those not trained in the concepts will soon become
unhappy.
The 2-ary is a poor first order construction that is useful only up to the point where real knowledge of things in the world (that are outside the descriptive capacity of the numeric model) can be agreed on.
If one is almost completely blind, even a rough piece of
unpolished glass might be useful.
We are taking the first step, and allowing you to make
comments on this because we know that your work is also important.
[1] The notation is developed in a consistent fashion and published at http://www.bcngroup.org/area2/KSF/Notation/notation.htm
[2] The reason why I treat this as a variable is discussed in my reply
[3] I have to remind myself here that my training is in the foundations of mathematics and Richard Ballard’s training is in Physics. The variable is not a functional, at least not in pure mathematics. I need a variable to fix specific constants in response to pragmatic aspects of a changing world. As the world changes, there is a need to say that a relationship that was is no longer, or that a new relationship has appeared when before there was not relationship.
[4] we agree on both points here, the problem is in where we are in our effort to deploy technology. The Ballard technology already has n-aries worked out but has not become a market standard as yet. The 2-aries are being used in a simpler technology that can be refined to include exactly the types of n-aries that Ballard and I have talked about for a number of years. He forgets that I agree with these two issues.