Spam filtering software
The classification engine in
uses the Maximum Entropy principle.
At the time of writing, there are several other free
Bayesian junk email filters. Some are specialized for junk mail only, others
do more general classification. Some use the "Naive Bayes" assumption of
Learning Theory, others use more ad-hoc methods. The following list is
neither complete nor authoritative. Suggestions and additions are welcome.
The following classifiers are based on an
influential essay by Paul Graham (or at least started out that way). Strictly speaking, they don't appear to actally implement Bayesian Statistics, but can still perform quite well.
Naive Bayesian classifiers
The following classifiers treat tokens (generally single words) independently according to an actual Bayesian model.
- ifile - (multiple categories)
- POPFile - (multiple categories)
- Select - (multiple categories, multiple algorithms)