Re: Low-scoring discount ED spam
--On Tuesday, May 04, 2010 4:22 AM +0100 RW rwmailli...@googlemail.com wrote: Are you training BAYES? A lot of these are hitting BAYES_50 or even BAYES_00. On 03.05.10 20:06, Kenneth Porter wrote: I've been copying them into my Uncaught folder which is run with sa-learn --spam --mbox each night. I just noticed that my Uncaught folder is huge and has lots of ancient messages, so the learning takes a long time. I'll strip it down to a month's messages and archive the old folder and see if that improves things. do you wipe bayes database often? If not, it's not needed to retrain on all messages, since they are not forgotten. You can save your spam to archive for later use (reconstruction of bayes if you loose it) after learning. -- Matus UHLAR - fantomas, uh...@fantomas.sk ; http://www.fantomas.sk/ Warning: I wish NOT to receive e-mail advertising to this address. Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu. How does cat play with mouse? cat /dev/mouse
Re: Low-scoring discount ED spam
--On Wednesday, May 05, 2010 11:29 AM +0200 Matus UHLAR - fantomas uh...@fantomas.sk wrote: do you wipe bayes database often? If not, it's not needed to retrain on all messages, since they are not forgotten. I don't recall ever deleting the DB. It's my understanding that sa-learn remembers which messages it's learned before, but it makes sense to remove them periodically to an archive to reduce the load figuring out that they've been seen before.
Re: Low-scoring discount ED spam
On Mon, 03 May 2010 18:15:51 -0700 Kenneth Porter sh...@sewingwitch.com wrote: I've been getting regular spam that advertises a percentage discount for ED in the subject line, and names the ED in the From line. It consistently fails to breach the 5.0 score line and keeps showing up in my regular Inbox. I think I have the latest code and rules. Am I suffering from the current blockage of the sa-update sources? Here's a sample, in Dovecot mbox format: http://sewingwitch.com/ken/Stuff/foo.txt Are you training BAYES? A lot of these are hitting BAYES_50 or even BAYES_00.
Re: Low-scoring discount ED spam
--On Tuesday, May 04, 2010 4:22 AM +0100 RW rwmailli...@googlemail.com wrote: Are you training BAYES? A lot of these are hitting BAYES_50 or even BAYES_00. I've been copying them into my Uncaught folder which is run with sa-learn --spam --mbox each night. I just noticed that my Uncaught folder is huge and has lots of ancient messages, so the learning takes a long time. I'll strip it down to a month's messages and archive the old folder and see if that improves things.