The blogspam plugin is just great.
However, it lacks support in the web interface to train comments as SPAM, when they were erroneously identified as ham. It would be great to have such support, also in the spirit of helping blogspam.net to get better and better.
What would consist the most appropriate user interface is not entirely clear to me in the general case (wiki page editing). The case of blog comments look easier to: when the admin user is logged in (and if the blogspam plugin is enabled), each comment can have an extra link "mark as SPAM" which would both delete/revert the comment and submit it to the configured blogspam server for training.
Comments can't have an extra link when the admin user is logged in, because the admin user sees the same static pages as everyone else (non-admins still see the "remove" link provided by the remove plugin, too). Perhaps a better UI would be that the action of that link was overridden by the blogspam plugin to go to a form with a checkbox for "also submit as spam"? --smcv
Similarly, ham training can be plugged directly into the current comment moderation interface. Each comment that gets approved by the admin, can be sent to blogspam.net as ham. If this is considered too "aggressive", this behaviour can need to be explicitly enabled by turning on a configuration option.