I've been using the holidays to explore a wonderful tool developed by researchers at Princeton University called Wordnet. Essentially a team of researches has broken up the English language into some of the sorts of chunks a system like Esoteric is going to need. It understands that certain groups of words (i.e. The Tcl Programming Language) all represent one concept, while at the same time knowing that close, shut and slam all do roughly the same thing to an object like a door, hatch or portal.
The database can be queried online.
Another great project is Erik Mueller's Tought Treasure. ThoughtTreasure's knowledge base consists of concepts, which are linked to one another by assertions. Ideas like handle is part of a door and rain makes things wet.
And now that I have these powerful tools in my hand... I'm finding they doesn't actually solve the problem I'm trying to solve. Well they solve part of the problem, but there is apparently no shortcut to constructing the mental models the actors in my system are going to need.
WordNet, while it does a decent job of relating words to one another, still needs a human to interpret the nuances between words. Thought Treasure, while great at recording commonsense assertions, still needs some source of ideas to relate.
Communication without a mental model just sounds like the old Eliza program.
> Hello, I am Eliza. * Hi Eliza! > How are you today.. What would you like to discuss? * I need to understand the fundamentals of intelligent thought > Can you elaborate on that? * I'm building a game full of intelligent agents. And I need to work out how to represent their world to themselves > Do you believe it is normal to be building a game full of intelligent agents. And you need to work out how to represent their world to themselves? * No, actually that is kind of unique to me > Why no? * Because this is a problem I'm stumbling across because I'm doing original research > Do you believe it is normal to be stumbling across because you're doing original research? * Normal for researchers, yes > I see.
I'm back at square one. I need ideas to communicate before any of these tools is going to be useful.