> >> I'm writing tools to read my mailing list traffic for me and tell me
> >> what's interesting.  I get between 500 and 1000 mailing list mails
> >> per day.
> >
> > Sound exciting. What is the principle, on which your tool selects
> > "interesting" postings? By keywords?
>
> I've written parts one and two:
>
> http://seattlerb.rubyforge.org/IMAPCleanse/
>
> I'm planning on using per-mailbox Bayesian analysis for phase three.
> What's interesting to me on this mailing list is not the same as
> what's interesting on FreeBSD-current.

Bayesian filtering on a topic by topic basis!! Awesome!!!

Integrate that into the mailreader in a transparent way and even Gmail
will look clunky!!

(Sorry, I'll stop salivating now...)

-- 
Giles Bowkett
http://www.gilesgoatboy.org