I discovered I'd written a neat little module a while back, called Wordmap, that lets you create XML lists of word associations. It's not very mature (barely supports spaces in words, can't have symbols in words, etc.) but it's a little nifty. I wonder if CPAN would have any use for it?
Anyway, I'm working on a thing which does Markov analysis of single sentences within a whole document, as prelude to creating the bots I want to make for IRC. (The idea is that you could split a logfile into one log for each personality, store each log as Markov data, and then generate new sentences in the style of the intended personality. The intent is to let people create personalized Op bots, or even chatbots, that seem like someone else.)
The storage program is going to be chaney.pl and the interpreter will be markoff.pl in dedication to Markoff Chaney.