On Sun, 30 Dec 2012 16:43:53 +0200 Jari Fredriksson wrote: > > I receive tons of spam currently on one mailing list. The list post > has following features > > - sender gets negative points because it is listed in DNSWL_MED
It's Redhat that's in DNSWL_MED - consider putting them in your trusted network to prevent that. > - the email is in html and has lots of tokens similar in spam and ham. > Bayes training is slow and seems futile In most unix mailing lists anything in HTML is very likely spam anyway. Assigning a few points to a meta of HTML_MESSAGE and the list-id might take you most of the way to catching this. > The spam has very common patterns, and contains lots of words that I > would like to ban from that one List-Id. The very common patterns sound more useful than the words. > So far I have created many meta rules containing those words, but the > list is endless. The words like mostly U.S. town names and U.S. sports > team names. On the face of it that doesn't sound too difficult. > In addition to that the spam contains words "game", > "basketball", "football", "live", "vs.", "at", "tv" and "pc". That sounds really error prone - particularly the last five words. > The list maintainer seems not to get them out, spam flow is > continuous. > > The rules would be much cleaner if the banned words would be in a flat > file or such, maintaining the lists would be easier. > > Is there such plugin already? It doesn't seem like a very good idea, but if you really want to do that you could use the flat-file to autogenerate an "in list" rule in its own .cf file.
