SUO: Re: pitfalls of matrix data
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
Tom,
It was somewhat explicitly recognized in the early years of AI and
computational linguistics -- Newell, Simon, Minsky, Papert, Chomsky,
Schutzenberger, Rescher, etc. -- that context-free parsing was a type
of abductive process, the use of a "nonterminal symbol" representing the
parser's best current guess about the grammatical category of the string
under its nose, and that push-down stacks were a way of systematically
making guesses and then retracting them, backtrack fashion, if needed.
This way of seeing it puts abductive reasoning in the same complexity
ballpark as properly context-free parsing, at the very minimalistic.
Indeed, the neural equivalent of properly context-free parsing was one of the things
that Chomsky and others argued (against the favorite models of Estes, Bower, Skinner
and others) put all higher levels of intelligent functioning beyond the reach of the
strictly finite state, markov, or regular language algorithms, into which complexity
class all 2-adic matrix, stimulus-response type models raced past the barn fell.
Jon Awbrey
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
John F. Sowa wrote:
>
> Tom,
>
TJ: I know John Sowa recommends an automatic ontology category generation
> > mechanism as well. But I don't understand how it would work. If you
> > can send me any references, I'd appreciate it.
>
> The basic idea is as old as Leibniz, who proposed
> several algorithms, which tend to generate many more
> categories than one would need. Newer algorithms
> have been invented, which reduce the number of
> unncessary nodes in the hierarchy.
>
> Following is my brief summary of lattice theory
> with a short discussion of one of Leibniz's
> methods and the FCA method:
>
> http://www.jfsowa.com/logic/math.htm#Lattice
>
> My basic point is as old as Socrates: the distinctions
> are fundamental to determining the categories, and they
> are ultimately more fundamental than the categories.
>
> Re natural kinds: Good question: Are there any?
>
> The original motivation was biological species, but
> with all of the people playing around with DNA these
> days, the notion of a species is far less clear than
> it used to be.
>
> John
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o