spamassassin-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Reindl Harald <h.rei...@thelounge.net>
Subject Re: non-English sender and body
Date Sun, 12 Jul 2015 08:48:28 GMT


Am 12.07.2015 um 04:59 schrieb James:
> On 07/11/15 19:55, Reindl Harald wrote:
>> Am 11.07.2015 um 17:36 schrieb James:
>>>>     Does Bayes catch them?
>>>>
>>>>     =======================================================
>>>>
>>>>     I don't know since I don't get any non-english emails. But, you
>>>>     could collect and feed them to "sa-learn spam" and see it that
>>>> helps.
>>>>
>>>>     Warmest regards,
>>>>     Mark Chino
>>>>
>>> It doesn't help (I think I feed sa-learn to much spam).
>>> They are always close.
>>> score=3.9 required=4.5
>>
>> and where is the problem to rasie the bayesc scores in loal.cf if you
>> *really* trained it?
>
> Mmm.
>
> X-Spam-Status: No, score=3.9 required=4.5 tests=HTML_MESSAGE,
>      HTML_OBFUSCATE_05_10,RCVD_IN_PBL,RDNS_DYNAMIC autolearn=no
> version=3.3.2
>
> Looks like my bayes is not working again.
> I had a lot of problems getting it right

then you found the real problem, any spamfilter without bayes has a real 
problem but i don't understand "lot of problems getting it right"

* you need some hundret ham *and* spam samples
* always call "sa-learn" as exactly the user SA runs later
* two cli calls to each sample folder, done

the root of all evil is when people start calling tools as root


Mime
View raw message