Re: sa-learn problems

2013-11-11 Thread Matus UHLAR - fantomas
On 12.11.13 05:48, Benny Pedersen wrote: why does it not always listen to commands ? eg i have one mail i like to --forget digest from and it just ignore it as not listen to it ? how do you know it does not listen to it? sa-learn --spam does not change anything, sa-learn --ham not either, s

Re: SA not correctly classifying spam

2013-11-11 Thread Karsten Bräckelmann
On Tue, 2013-11-12 at 05:57 +0100, Benny Pedersen wrote: > Karsten Bräckelmann skrev den 2013-11-12 05:44: > > > In other words: If all you're being taught as a child is good, how will > > you ever know what is bad? > > this is also my question, when to use bayes ignore I have no fscking clue, h

Re: SA not correctly classifying spam

2013-11-11 Thread Benny Pedersen
Karsten Bräckelmann skrev den 2013-11-12 05:44: In other words: If all you're being taught as a child is good, how will you ever know what is bad? this is also my question, when to use bayes ignore will it just ignore the bayes ignore header or whole mail if its found ?

sa-learn problems

2013-11-11 Thread Benny Pedersen
why does it not always listen to commands ? eg i have one mail i like to --forget digest from and it just ignore it as not listen to it ? sa-learn --spam does not change anything, sa-learn --ham not either, so should one of them not work ?

Re: SA not correctly classifying spam

2013-11-11 Thread Karsten Bräckelmann
On Tue, 2013-11-12 at 01:57 -0200, Sergio Durigan Junior wrote: > On Tuesday, November 12 2013, Benny Pedersen wrote: > > Karsten Bräckelmann skrev den 2013-11-12 03:20: > > > > > [1] Also, just as shown in this thread, properly handling list posts is > > > not trivial. > > > > maillist is good

Re: SA not correctly classifying spam

2013-11-11 Thread Sergio Durigan Junior
On Tuesday, November 12 2013, Benny Pedersen wrote: > Karsten Bräckelmann skrev den 2013-11-12 03:20: > >> [1] Also, just as shown in this thread, properly handling list posts is >> not trivial. > > maillist is good ham learning spams :) Yeah, that's a good reason to keep scanning mailing lis

Re: SA not correctly classifying spam

2013-11-11 Thread Sergio Durigan Junior
On Monday, November 11 2013, John Hardin wrote: > It's a very good idea to retain your training corpora. It makes it a > lot easier to review if Bayes goes off the rails, and to wipe and > retrain from scratch if problems occur. That's a good reason for keeping them around. >> Currently, in my .

Re: SA not correctly classifying spam

2013-11-11 Thread Benny Pedersen
Karsten Bräckelmann skrev den 2013-11-12 03:20: [1] Also, just as shown in this thread, properly handling list posts is not trivial. maillist is good ham learning spams :) yes i know, but could not resists, users that dont use shortcircuit rules will make more unneded cpu curcles but its

Re: SA not correctly classifying spam

2013-11-11 Thread Karsten Bräckelmann
On Mon, 2013-11-11 at 17:24 -0800, John Hardin wrote: > On Mon, 11 Nov 2013, Sergio Durigan Junior wrote: > > Currently, in my .procmailrc, the first thing I do is scan for spam. > > However, as you have noticed, this scanning also covers mailing lists > > (and everything else). I think I will tw

Re: SA not correctly classifying spam

2013-11-11 Thread Benny Pedersen
John Hardin skrev den 2013-11-12 02:24: That's not what Karsten was suggesting. If you get spam via that mailing list you should complain to the list admin that they need to do a better job of filtering their inbound. olso that spf test would work on initial sender domain if you add maillist

Re: SA not correctly classifying spam

2013-11-11 Thread John Hardin
On Mon, 11 Nov 2013, Sergio Durigan Junior wrote: On Monday, November 11 2013, John Hardin wrote: On Mon, 11 Nov 2013, Sergio Durigan Junior wrote: Here is an example of a misclassified spam message: There's not a lot there for SA to work with.

Re: SA not correctly classifying spam

2013-11-11 Thread John Hardin
On Mon, 11 Nov 2013, Sergio Durigan Junior wrote: However, I had some direct spams before, but as I explined in the other thread, I deleted them after I fed sa-learn. It's a very good idea to retain your training corpora. It makes it a lot easier to review if Bayes goes off the rails, and to

Re: SA not correctly classifying spam

2013-11-11 Thread Sergio Durigan Junior
On Monday, November 11 2013, John Hardin wrote: > On Mon, 11 Nov 2013, Sergio Durigan Junior wrote: > >> Here is an example of a misclassified spam message: >> >> > > There's not a lot there for SA to work with. Indeed. Sorry, that was the only spam I

Re: SA not correctly classifying spam

2013-11-11 Thread Sergio Durigan Junior
On Monday, November 11 2013, Karsten Bräckelmann wrote: > On Mon, 2013-11-11 at 20:26 -0200, Sergio Durigan Junior wrote: >> Here is an example of a misclassified spam message: >> >> >> (This spam message was sent to a mailing list, not directly to my

Re: SA not correctly classifying spam

2013-11-11 Thread John Hardin
On Mon, 11 Nov 2013, Sergio Durigan Junior wrote: Here is an example of a misclassified spam message: There's not a lot there for SA to work with. The biggest issue is URIBL_BLOCKED. Your URIBL queries are being blocked, likely because you're usin

Re: SA not correctly classifying spam

2013-11-11 Thread Karsten Bräckelmann
On Mon, 2013-11-11 at 20:26 -0200, Sergio Durigan Junior wrote: > Here is an example of a misclassified spam message: > > > (This spam message was sent to a mailing list, not directly to my > address, as can be seen. I still don't have spams that were

SA not correctly classifying spam

2013-11-11 Thread Sergio Durigan Junior
Hi there, As requested by Karsten here (I took the liberty to include him in the Cc list): I am starting this new thread in order to try to solve/identify what's going on with my SA instance (*if* there's anyt