Spamassassin usually takes care of my spam problems pretty well, but I've gotten a rash of messages recently that circumvented my bayesian filters. A friend forwarded this SA rule on to me that seems to do the trick:
# cf. http://alt-usage-english.org/excerpts/fxcommon.html body KW__RANDOM_SENTENCE / ((?!(the|of|and|to|a|in|that|is|was|it)+)[a-z']+ ){18,}/ describe KW__RANDOM_SENTENCE random-looking lowercase unpunctuated words score KW__RANDOM_SENTENCE 0 This will detect a sequence of 18 or more words without any commonly expected english words (articles, prepositions, conjunctions, conjugations of "to be", etc.). This will, unfortunately, nuke your foreign language e-mails, so choose your pointage wisely. The arms race continues... Mike .___________________________________________________________________. Michael A. Halcrow Security Software Engineer, IBM Linux Technology Center GnuPG Fingerprint: 05B5 08A8 713A 64C1 D35D 2371 2D3C FDDA 3EB6 601D All in favor of losing your rights, please do nothing.
pgp00000.pgp
Description: PGP signature
____________________ BYU Unix Users Group http://uug.byu.edu/ ___________________________________________________________________ List Info: http://uug.byu.edu/cgi-bin/mailman/listinfo/uug-list
