Ask Jeeves, actually has several different types of search capabilities.
1. Direct Hit > Popularity technologies, utilised by MSN, DH, and so.
2. Teoma > Machine denoted clustering (communities).
I think a common fallacy of search technologies, is that their are major differences in the maths: as in algos against heuristics this is not so. Its the application of both and the statistical intrepretation, hence weighting applied to all the factors measured. And most importantly what the Factors are, how many off them.
Obviously from here there is a lot that can be worked out, but these are rule of thumb observations, such as page size, site structure, linking strategies, and so. Somewhat akin to the observance of wildlife, you can observe and workout patterns, that the studied follow, but you don't know precisely why!
Thats the way it is with the technologies behind the big SE's including ask jeeves, its the most closely gaurded secret in this game.
Ask Jeeves and the rest will be using RBR which in thousands will create case's, like using machine code, and a compiler.
This post by Brett, the systems operator, is very illustrative of the teoma technologies and how it evolved out of Rutgers Uni, note advanced reading topics at the bottom, in here is some excellant information to read up on teoma hits the ground running [webmasterworld.com]
what you are actually studying (subject,level) ? may be able to find you some other material from the world of AI.