Bad air day
: the air conditioning is still off. The ANU
has taken pity on us and given us a bunch of electric fans. At last count, there was 13 fans in the engineering department alone.
Spot The Fake Smile
: fun for all the family... I got 8 wrong.
A couple of nice photos we took on Australia day at the Chinese gardens in Sydney. 1
There seems to be an increase in the use of machine learning techniques for filtering spam
these days. A while ago I got sick of writing procmail rules for new mailing lists and groups so I decided I'd try to adapt one of the Bayesian spam tools to learn how I classified my mail. I use MH
which uses the refile command to move messages around, so I wrapped that command so that every time I moved a message to a new folder, the filter would get this info and use it as training data. This has worked really well, but now I'd like to be able to do the same thing at work
where MS Outlook is mandated. There are spam filters which use Bayesian techniques but these only classify to spam or ham. Writing email rules under Outlook is a slow, painful, point and dribble processes which I hate even more than writing procmail rules. Even if the filter only got 90% of the classifications correct, that'd save me lots of time.
Badger Badger Badger
: for some reason this stupid web site has me hooked...