It seems that my previous spam measurements were quite good, but also that the spammers have stepped up to the next level : comment spam. I already had to delete 7 spam comments last week, so I've decided to install Drupal's spam module. It contains a trainable Bayesian filter, manually entered custom filters, counting the number of URLs, and detection of content posted from open email relays.


The Bayesian filter does statistical analysis on spam content, learning from spam and non-spam that it sees to determine the liklihood that new content is or is not spam. Seems a bit like the spam filters included in Thunderbird.