Aaron Swartz said:
> Thanks for SpamAssassin, WebMake, SiteScooper, and the zillion other useful > things you've made. cheers! > I plan to install SpamAssassin, so I was peeking around the site and found: > """ > >>From 1.3 on, they are assigned using a genetic algorithm, to optimise their > efficiency and minimise false positives and false negatives. More > information can be found here. Note that you can help this system by > providing statistics on your mail spool. > """ - http://spamassassin.taint.org/faq.html > But you don't say how the algorithm is fed or how we can contribute. It'd be > nice if you could provide this information. Well, check the "masses" directory in the distro, the README covers this. As does the FAQ on the website ;) In short, you can run "mass-check" on your set of messages and provide the output, and I'll incorporate it in the GA run. Now though, I'm beginning to think that the *actual messages* are more helpful, esp. false-positives -- as the "mass-check" output has to be thrown out after a month or 2 as the tests improve and new ones are added, to avoid skewing the GA. > Also, is there a list of major features coming in 2.0? Just sort of curious > about the version leap. We haven't yet coerced the CVS changelog into a nice feature list, but the main ones are: - new spam-phrase analysis module, which uses analysis of spam mails to determine common phrases, providing another way to get spam points from the text. - "rawbody" config item added, which allows a more sensible way to scan text bodies. "body" now cleans up whitespace and removes NLs for plain-text matching. - config file split up into a directory of files. Also, config file locations are now legal by the FHS. (hence the major number change) - some spamc/spamd command line arg semantics changed (hence the major # change) - Mail::Audit, Mail::Internet, Net::SMTP dependencies removed. also provides a nice speedup ;) - new GA scoring, of course. ;) OT: BTW I like swhack, http://blogspace.com/swhack/weblog/ . ghod only knows how I never found that before. Is there a way to see more than 1 day's bloggage? --j. _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
