Artificial Intelligence
THE PAMELA PROJECT & VISUAL AIML
This page is best viewed with MS Explorer Res. 800 x 600 

HOME          PAMELA          VlSUAL AIML 2.0

 

 WHO IS PAMELA ?

PAMELA speaks Spanish - her last name is not Anderson -, but you can bet that sex will be part of her life  (among other personal interests). She will probably excel in some professional area (may be as a successful on-line sales woman ?) maintaining her psychological balance through a bit of Zen philosophy (to prevent her from getting too cynic or alienated in our competitive world)
 
She is currently taking a crash course, learning how to behave and think. I'm her teacher (you can call me Botmaster if you like).

 THE GREAT PICTURE

Intrinsic differences between Spanish and English lead me to start PAMELA's AIML education from scratch.

My project relies on a 3 layer model aimed to FORMAT INPUT TEXT - FOCUS ATTENTION ON CURRENT MATTER - PROVIDE INTELLIGENT ANSWERS, more details as follows :

Layer 1 Format input text : Using srai categories to simplify syntax to be passed to the next level. The recursive nature of srais makes them very powerful to filter, replace, and/or reinterpret patterns to provide a clearer understanding.This level could be roughly compared to specialized sensors (like our eyes, ears, etc) feeding its impulses to different areas of the brain in the appropriate "format".

Layer 2 Basic social protocol : Includes categories about PAMELA's personality, how to start basic exchanges involving ME and YOU, a sense of self, ability to categorize inputs using different perspectives (i.e. the position of the speaker, does it relate to being, doing, having, wanting, etc. etc.). The main purpose of this layer is to handle the beginning of a conversation and perhaps more important, to detect "breakpoints" to switch attention (pattern priority) to specific topics dynamically during conversation.

Layer 3 Specialized brain areas : Each "brain area" could be viewed as a cluster of topics closely interrelated, or as a brain area  having some "inputs" and "outputs". Each areas includes srais to further improve the understanding of text in a more specific  context, "reasoning" categories to process the information "intelligently", and categories storing support knowledge (like our memory) to assist in building meaningful responses. A key point to understand, is that all specialized areas have the ability to communicate with all other areas of the brain sharing and/or combining its resources in a virtual network with no pre-defined paths.

A heavy  use of  "topics" acting as subroutines, allows to switch attention from the current subject to a new one, prioritizing otherwise "similar" patterns that could lead to misinterpretation if matched outside the right context.

See Picture with 3 Layer Concept

 

VISUAL AIML : Soon I realized the need for a tool to assist my efforts, I started to look around playing with some ideas, but finally  decided to develop my own utility called Visual AIML. To learn more follow the link at the top menu.

Top

© 2001 ICG

pamela783@yahoo.com

 

Updated 06/02/01