You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 8 Next »

A spam message got through! Can I send it to you so you can update the filters?

No. We really have more than enough spam already, so we don't really need you to forward spam to the spamassassin-talk mailing list or submit bugs for missed spam.

Nevertheless, there are two ways in which you may be able to help.

First, if there are existing tests which nearly bring the message over the threshold, then yes, we could do with that data in our rescoring corpus of messages so that the GA (genetic algorithm) will know to bring those scores a little higher. Before every major release, the SpamAssassin developers send out a call for advanced users to submit data for use in the GA process. If you follow the [http://www.mirror.ac.uk/sites/spamassassin.taint.org/spamassassin.org/lists.html SpamAssassin-talk] mailing list, maintain spam and ham corpuses, are familiar with Perl, rsync, and CVS, then this may be a way in which you can contribute.

Second, if the mail was entirely missed by SpamAssassin and you have an idea for how it might be detected (in a way that doesn't also catch ham) with a new rule, see ContributingNewRules.

Third, you may want to install additional rules. [http://www.rulesemporium.com/rules.htm RulesEmporium] is a great resource for additional rules and scripts that are not included with SA by default.

And when you do need to forward or attach spams, be sure to note that we need the original message including all the message headers; without headers, it's very likely to be ignored, as it's almost useless for filter development. Never ever cut-and-paste spams since that will almost always change the spacing and line-breaks of the message.

Also note, if you have a large collection of these, it's better to zip or tar.gz them up, instead of sending them one by one. A gzipped mbox is probably the most convenient way to ship collections of one to one hundred messages (as long as they're all examples for the same thing).

  • No labels