A brief summary of the semantic db project:
1) a general scheme to represent knowledge (using some notation borrowed, and tweaked, from quantum mechanics)
2) a programming language to reason with that knowledge (the BKO language).
3) a console where you can use the BKO language.

So, the main idea is to converge all knowledge into (what I call) superpositions.
And then write relatively general functions that can work on superpositions, mapping superpositions to superpositions.


There are lots of pieces in the full project, but some key pieces are:
1) similar -- compares superpositions and returns a result in [0,1] depending on how similar (this is a basic version of pattern recognition)
2) find-topic -- looks up frequency lists and returns degree of membership
3) categorize-code -- with respect to some metric it puts items into categories
(based on my maths idea of bridging sets)
4) the maths idea of "function matrices"
5) active-buffer -- first (very early) step in the direction of a computer understanding what it is reading.