(스크랩) SpamAssassin rules Automatic rule generation

SpamAssassin rules Automatic rule generation

 

Automatic rule generation

A script apparently associated with the SoughtRules automatic score generation can be downloaded from trunk:

svn checkout http://svn.apache.org/repos/asf/spamassassin/trunk
cd trunk/masses/rule-dev
./seek-phrases-in-corpus ham:dir:~/Maildir/ spam:dir:~/Maildir/.bad.spam-missed/ 

The script contains instructions. You tell it a directory or file containing spam, and a directory or file containing non-spam, and it'll come up with rules that match the spam but not the non-spam.

If you have run MassCheck, you can generate rules for the set of spams that were below the default threshold by doing something like (you'll probably need to change the paths to corpora):

cd trunk/masses
awk '$2 < 5' < spam-*.log > missed_spam.ids
./mboxget < missed_spam.ids > missed_spam.mbox
cd rule-dev
./seek-phrases-in-corpus ham:dir:$HOME/masscheckwork/ham/ spam:detect:~/masscheckwork/nightly_mass_check/masses/missed_spam.mbox  > seek.out

 

https://cwiki.apache.org/confluence/display/spamassassin/WritingRules