On Thu, Dec 05, 2002 at 05:04:28PM -0700, Michael Moncur wrote: > Actually 6 out of the 14 messages in your corpus are from these people. They > frequently change URLs so it's hard to recognize them. But they all have the > same sparse format and similar URLs that match some of the rules I posted. > They tend to be affiliate spam - the URLs redirect you to a legitimate > business (such as Columbia House) using an affiliate code that makes them > money. > > They also tend to match the CTYPE_JUST_HTML and WEB_BUGS rules, both of > which have near-zero scores in 2.43. I raise the score on both of those > locally because they don't cause false positives for me. > > These folks are serious spammers. They even register each of the domain > names to look legit. Look up findhsm-list-cluster-182-643.com or any of the > others, they're all registered to Daily Promotions. (It's a shame a WHOIS > lookup would be expensive, or this would make an excellent eval test.) > > From: MyChoice <[EMAIL PROTECTED]> > Subject: Keep your partner coming back for more.
Would a "domain contains many numbers" test help? header DOMAIN_NUMS From =~ /(\d\S?){5}/ -- Duncan Findlay ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk