-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hello Mike,
Monday, September 8, 2003, 6:39:10 AM, you wrote: MK2> We do site wide here, so I've had to make adjustments, too. Not MK2> ready to do bayes, yet. Not sure how well it would work for a site MK2> wide environment. We do consulting, so most of our FPs used to come MK2> from "Dear Sir, look at my resume", flagged with high importance, a MK2> subject of only hello, and from a yahoo mail account. You do bring MK2> up some good points I'm going to check out, thanks for the help. My understanding that Bayes works /best/ on an individual basis (Sam's spam may be Mary's ham), but my experience indicates it's worth while on a site wide basis also. I run three domains, each of which has its own Bayes database, but the one Bayes database covers all email addresses within that domain. Within each domain there are going to be similarities concerning what is or isn't spam (nobody at the contractorswarehouse.com domain will be receiving weight reducing offers, while they do a lot of business with Asia). It seems to work OK for us. Had a couple of resumes come through today, one in text/html scored 5.8 of 9.0 (Bayes-40), and the other in Word, scoring 0.3 of 9.0 (Bayes-30). Sending both into sa-learn. Bob Menschel -----BEGIN PGP SIGNATURE----- Version: PGP 8.0 iQA/AwUBP10Nf5ebK8E4qh1HEQKaegCeJXGq3xXM4Q6sC45TUn7W0fuoGL8Anjj3 Jp6q1iTip87IO4g3TV6Tn1WR =K2+n -----END PGP SIGNATURE----- ------------------------------------------------------- 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
