Bayes Poisoning - How to combat ?
Dhawal Doshy
dhawal at NETMAGICSOLUTIONS.COM
Thu Mar 17 04:28:58 GMT 2005
[ The following text is in the "ISO-8859-1" character set. ]
[ Your display is set for the "US-ASCII" character set. ]
[ Some characters may be displayed incorrectly. ]
Try using some rules from rulesemporium, one of them is this:
http://www.rulesemporium.com/rules/70_sare_bayes_poison_nxm.cf
You could also setup a spam harvesting site OR some fake email-ids for
collecting spam purely for learning and not rely on autolearn=spam.
A small bumping up of the SURBL scores will also do wonders, since bayes
is no more as effective as it used to be.
- dhawal
Venkata Achanta wrote:
> not spam, SpamAssassin (score=1.891, required 2.5,
> BAYES_40 -1.10, HTML_10_20 0.25, HTML_MESSAGE 0.00,
> MIME_HTML_ONLY 0.18, RAZOR2_CF_RANGE_51_100 0.06, RAZOR2_CHECK 1.51,
> URIBL_SBL 1.00)
>
> I have a bunch of machines running MS and recently the **untagged** spam
> rate has gone up on most of the boxes. After some examination, i found that
> the bayes is getting a negative score in all the spam that is getting
> through, which is a clear symptom of bayes poison.
>
> Apart from
> -deleting the bayesian DBs everytime and starting over again,
> -Turning off bayes in spamassassin
>
> Is there a better fix for a situation like that ?
>
> Thanks much,
> Venkata Achanta
------------------------ MailScanner list ------------------------
To unsubscribe, email jiscmail at jiscmail.ac.uk with the words:
'leave mailscanner' in the body of the email.
Before posting, read the MAQ (http://www.mailscanner.biz/maq/) and
the archives (http://www.jiscmail.ac.uk/lists/mailscanner.html).
Support MailScanner development - buy the book off the website!
More information about the MailScanner
mailing list