Beta 4.50.4 released -- faster than 4.49

Randal, Phil prandal at HEREFORDSHIRE.GOV.UK
Thu Jan 5 12:35:33 GMT 2006


Adri Koppes wrote:
>> Jules,
>> 
>> Like some of the other SA settings (value for low/high spam) the
>> expire settings for low/high spam are very site specific.
>> As Raymond tried to explain: e-mail body's are analysed by automated
>> tools. This usually leadd to some url's being added to a URIBL.
>> The URIBL will be updated once every couple of minutes. After that
>> -the same message body- would be tagged higher by SA ..... I suppose
>> the same is valid for bayes. If the same message passes the filter
>> more often, it should get a higher bayes value.
>> 
>> So, having a lower expire timeout on lowspam messages can be
>> 'required' for some setups. But other, who use less dynamic SA lists
>> probably could cope with a higher value ...
> 
> Jules,
> 
> Something else you might consider implementing, is not to
> cache or cache with a low timeout, the results when the
> message is auto-learned by bayes.
> Since bayes has learned from the message, it will probably
> affect the score of the next message, unless the score is
> already sufficiently high enough or low enough for the spam
> result not be affected by a diffirent bayes score.
> 
> Adri.

Not just auto-learning - if I detect a flood of low-scoring spam and
manually sa-learn it then I want any new Bayes score to be effective
ASAP.

With this new infrastructure in place it should be possible to introduce
some spam flooding detection into MailScanner.  You'd then need a
permitted-flooders whitelist too, of course.

As a completely unrelated aside, www.mailscanner.info is inaccessible
here at the moment.

Cheers,

Phil
----
Phil Randal
Network Engineer
Herefordshire Council
Hereford, UK

------------------------ 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 Wiki (http://wiki.mailscanner.info/) 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