Im starting to think of how to build a knowledge base. A couple of methods come to mind. The first is absorbing knowledge such as a dictionary by importing it into a database. That would only cover defining words, some relationships and their verbal definitions. I found one, online, that is ascii, broken down into databases of parts of speech (noun, verbs, etc.).
Im also working on a Perl:Tk teach GUI, to teach the knowledge base, not unlike the Open Mind approach. This would be a simple way of filling in the blanks, giving us a quick way of putting some limited knowledge in the knowledge base to teach the baby AI, not unlike pre-birth intuition.
The best way to teach AI would be through conversation, such as Socratic dialog, through a chat interface.
The knowledge base will require a little pixie dust, to turn it into a real knowledge base, one that would be useful to an AI application. That pixie dust is probably the heart of this project.
One of my approaches to problem solving is that when the brute force, direct approach failed me, to work the borders of the problem, eventually working to the core. This is my personal approach to Conditional AI.
One issue to keep in mind is that some words will have multiple association yet must remain unique. If Condai was to converse with two Karls, two of the same first, middle, and last names, age, etc., it must be able to discern that they are unique. Thus in database terms, they must have a unique identifiers associated with them. If there are two Karls that routinely chat with AI, then it must be able to sense clues as to which he/she/it is, and attach that unique identifier for the duration of the verbal exchange. (I wish to think of AI as an it, rather than of male or female gender, you?)
In the mean time Im toying with words, their meaning and classifications on paper, trying to come up with a proper schema for these relation ships.
Anyone remember maloc in C? I envision knowledge associations as those types of chains.
Karl (one of two?)
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Folks,
Im starting to think of how to build a knowledge base. A couple of methods come to mind. The first is absorbing knowledge such as a dictionary by importing it into a database. That would only cover defining words, some relationships and their verbal definitions. I found one, online, that is ascii, broken down into databases of parts of speech (noun, verbs, etc.).
Im also working on a Perl:Tk teach GUI, to teach the knowledge base, not unlike the Open Mind approach. This would be a simple way of filling in the blanks, giving us a quick way of putting some limited knowledge in the knowledge base to teach the baby AI, not unlike pre-birth intuition.
The best way to teach AI would be through conversation, such as Socratic dialog, through a chat interface.
The knowledge base will require a little pixie dust, to turn it into a real knowledge base, one that would be useful to an AI application. That pixie dust is probably the heart of this project.
One of my approaches to problem solving is that when the brute force, direct approach failed me, to work the borders of the problem, eventually working to the core. This is my personal approach to Conditional AI.
One issue to keep in mind is that some words will have multiple association yet must remain unique. If Condai was to converse with two Karls, two of the same first, middle, and last names, age, etc., it must be able to discern that they are unique. Thus in database terms, they must have a unique identifiers associated with them. If there are two Karls that routinely chat with AI, then it must be able to sense clues as to which he/she/it is, and attach that unique identifier for the duration of the verbal exchange. (I wish to think of AI as an it, rather than of male or female gender, you?)
In the mean time Im toying with words, their meaning and classifications on paper, trying to come up with a proper schema for these relation ships.
Anyone remember maloc in C? I envision knowledge associations as those types of chains.
Karl (one of two?)