Bayesian Filtering of Spam in MT

James Seng [1] told us [2] commenting my comment – o my god 😉 – that he just recently added to MT exactly what I intended to do with the BayesianClassification for cafelog. Well done!

Well, it is worth reading that post and those comments on James site, as he e.g. states:

I have customized the Bayesian algorithm to handle the blogs comments and pings, which is normally shorter then emails by via special handling of URLs, emails, hosts and IP addresses.

So James introduced an interface in MT to train the algorithm. Once the filter is trained it needs hardly any further maintenance. That is what I was looking for.

[1] http://james.seng.cc/archives/000152.html
[2] joi.ito.com

Leave a Reply