This is something one needs to do with all statistical-based filters,
not just SpamAssassin's Bayesian component. I made the mistake of
overtraining a DSPAM install with spam as compared to ham a few years
back, and it had the same effect. Balancing out the spam overtraining
with a good corpus of ham almost eliminated the spike in false positives.
On 12/14/2010 02:31 AM, Stephen Brooks wrote:
>> From the horses mouth as it where
>http://spamassassin.apache.org/full/3.3.x/doc/sa-learn.html jump down to
> the "Getting Started" section. Basically if you just tell SpamAssassin
> what is spam it'll be biased to marking everything as spam, you also
> need to give it examples of good email, aka ham. Use the tool sa-learn
> as described above but generally :-
> sa-learn --spam /path/to/spam/folder (if you've got samples of spam in a
> junk folder)
> sa-learn --ham /path/to/ham/folder (your inbox for example)
>> Typoed my name sorry :)
>> On 13/12/10 23:12, Gerard Hooton wrote:
>>>> "Did you feed SpamAssassin with Ham as well as Spam?"
>> Can you explain or point me to a howto
>>>>>>> On 13/12/10 15:38, Ger Hooton wrote:
>>>>>>> I have Debian 5 with exim 4.69 and SpamAssassin version 3.3.1
>>>> The problem is that ALL email hav the following in the header:
>>>>>>>> -spam_report: Spam detection software, running on the system
>>>> "xxx.yyy.ie", has identified this incoming email as possible spam
>>>>>>>> Where should I start to debug?
>>>>>>>>>>>>>>> Did you feed SpamAssassin with Ham as well as Spam?
Maintained by the ILUG website team. The aim of Linux.ie is to
support and help commercial and private users of Linux in Ireland. You can
display ILUG news in your own webpages, read backend
information to find out how. Networking services kindly provided by HEAnet, server kindly donated by
Dell. Linux is a trademark of Linus Torvalds,
used with permission. No penguins were harmed in the production or maintenance
of this highly praised website. Looking for the
Indian Linux Users' Group? Try here. If you've read all this and aren't a lawyer: you should be!