Kevin Vance - Also, I have to say that the SpamAssassin + SpamBouncer combination…

Entries | Archive | Friends | Friends' Friends | User Info

02:32 pm

Friday, June 6th, 2003
Previous Entry Share Next Entry
Also, I have to say that the SpamAssassin + SpamBouncer combination is working very well. A few spams are still squeaking by, but it's a managable number, and that number continues to decrease as the bayesian algorithm is trained. I'm also moving my threshold lower and lower. It's now 4.0. I've never gotten a false positive yet. In fact, the only messages that got a positive score (but low enough to get into my inbox) were from companies that I've dealt with in the past whose mailings I don't particularly remember requesting. And thus, my inbox is safe again... but for how long?
Link )Reply )

Comments
[User Picture]From: suppafly
2003-06-06 11:38 am (UTC)

...but for how long..

da da duh!!!
(Reply) (Thread)
[User Picture]From: teferi
2003-06-06 11:56 am (UTC)
SpamAssassin's learned from about 2,000 spams I've received (whether by autolearn or by running sa-learn by hand), and probably close to 1k not-spams, and I seem to have finalkly hit some kind of critical threshold, because I haven't seen a spam in about a week
(Reply) (Thread)
[User Picture]From: kvance
2003-06-06 01:35 pm (UTC)
Cooool. What's your threshold?

I trained initially with only 1000 spams, but many thousands of non-spams (I have over a year of email archived). So hopefully after more come in, I won't be seeing any either.
(Reply) (Parent) (Thread)
[User Picture]From: teferi
2003-06-06 02:06 pm (UTC)
still default 5.0.
(Reply) (Parent) (Thread)