Evolution and spam

classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

Evolution and spam

VETSEL Patrice
Where are we going with spam filtering in evolution ?
Evolution package recommend spamassassin (it's NOT installed)
Evolution package recommend evolution-plugins (it's installed)
Evolution-plugins package recommend bogofilter (it's NOT installed)

End users have icons and options in evolution that seem's to indicate
that there is a working solution of spam filtering. But that's wrong,
and all spam will not be filtering.

Other problem is that in evolution plugins, the 2 plugins (spamassasin
and bogofilter) are activated by default (neither if bogofilter and
spamassassin aren't installed). If end user install bogofilter, he must
de-activate spamassassin plugin because it break bogofilter plugin.

We have to be more friendly and more efficient.

One possible suggestion is to add a first mail in local box in evolution
explaining that problem (like the first mail presentation of evolution),
and to de-activate the 2 spam plugins ?!

Regards

VETSEL Patrice
Aka Kagou
https://launchpad.net/people/vetsel-patrice



--
ubuntu-devel mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
Reply | Threaded
Open this post in threaded view
|

Re: Evolution and spam

Dennis Kaarsemaker
On do, 2006-03-16 at 11:38 +0100, VETSEL Patrice wrote:

> Where are we going with spam filtering in evolution ?

Currently nowhere unfortunately - and dapper is already way past feature
freeze (even if dapper were delayed, feature free would remain intact).
Please write a BetterSpamFiltering spec for dapper+1 with your ideas and
let's hope it will be implemented.

In Dapper I switched from spamassassin to bogofilter but it's not useful
at all since you can't mark messages as non-juk if they are no yet
marked as junk. And thus bogofilter (same goes for spamassassin) cannot
learn good tokens. You have to resort to the command line for this.

I really hope that there will be some improvement for dapper+1, which
only is possible if someone with good ideas about it writes a spec.
--
Dennis K.
 - Linux for human beings - http://www.ubuntu.com
 - Linux voor normale mensen - http://www.ubuntu-nl.org

--
ubuntu-devel mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel

signature.asc (198 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Evolution and spam

Jan Claeys-3
Op do, 16-03-2006 te 11:16 +0100, schreef Dennis Kaarsemaker:
> In Dapper I switched from spamassassin to bogofilter but it's not useful
> at all since you can't mark messages as non-juk if they are no yet
> marked as junk. And thus bogofilter (same goes for spamassassin) cannot
> learn good tokens. You have to resort to the command line for this.

You can mark some ham as spam and then mark it back as ham, no need for
commandlines...


--
Jan Claeys


--
ubuntu-devel mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
Reply | Threaded
Open this post in threaded view
|

Re: Evolution and spam

John Nilsson
In reply to this post by Dennis Kaarsemaker
On Thu, 2006-03-16 at 11:16 +0100, Dennis Kaarsemaker wrote:
> Please write a BetterSpamFiltering spec for dapper+1 with your ideas and
> let's hope it will be implemented.

I took the liberty of creating that page:
https://wiki.ubuntu.com/BetterSpamFiltering

I just added a special case though. Someone with more insight in the
actual problem should flesh out the rest.

Regards,
John


--
ubuntu-devel mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
Reply | Threaded
Open this post in threaded view
|

Re: Evolution and spam

Aigars Mahinovs-2
In reply to this post by Dennis Kaarsemaker
On 3/16/06, Dennis Kaarsemaker <[hidden email]> wrote:
> In Dapper I switched from spamassassin to bogofilter but it's not useful
> at all since you can't mark messages as non-juk if they are no yet
> marked as junk. And thus bogofilter (same goes for spamassassin) cannot
> learn good tokens. You have to resort to the command line for this.

There is not much sence in training the bogofilter (and most other
filters) on what they have corectly identified as spam or non-spam:

http://lwn.net/Articles/172491/

As for initial clasification - if you do inital training, then you
have to resort to either command line or folder manipulation tricks.

--
Best regards,
    Aigars Mahinovs        mailto:[hidden email]
 #--------------------------------------------------------------#
 |     .''`.         Debian GNU/Linux              LAKA         |
 |    : :' :      http://www.debian.org  &  http://www.laka.lv  |
 |    `. `'                                                     |
 |      `-                                                       |
 #--------------------------------------------------------------#

--
ubuntu-devel mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
Reply | Threaded
Open this post in threaded view
|

Re: Evolution and spam

Lionel Dricot (aka Ploum)
My 0.02€ :

To train bogofilter, I simply toke the first message of my inbox and
learned it as ham with the cli. It was a short message.

Since then, I had very few error. Perhaps one or two the first day,
nothing else.

So, I suggest to package an already trained bogofilter on one simple
ham (ubuntu announce ?) that will be enough.


On 16/03/06, Aigars Mahinovs <[hidden email]> wrote:

> On 3/16/06, Dennis Kaarsemaker <[hidden email]> wrote:
> > In Dapper I switched from spamassassin to bogofilter but it's not useful
> > at all since you can't mark messages as non-juk if they are no yet
> > marked as junk. And thus bogofilter (same goes for spamassassin) cannot
> > learn good tokens. You have to resort to the command line for this.
>
> There is not much sence in training the bogofilter (and most other
> filters) on what they have corectly identified as spam or non-spam:
>
> http://lwn.net/Articles/172491/
>
> As for initial clasification - if you do inital training, then you
> have to resort to either command line or folder manipulation tricks.
>
> --
> Best regards,
>     Aigars Mahinovs        mailto:[hidden email]
>  #--------------------------------------------------------------#
>  |     .''`.         Debian GNU/Linux              LAKA         |
>  |    : :' :      http://www.debian.org  &  http://www.laka.lv  |
>  |    `. `'                                                     |
>  |      `-                                                       |
>  #--------------------------------------------------------------#
>
> --
> ubuntu-devel mailing list
> [hidden email]
> https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
>
>
>

--
ubuntu-devel mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
Reply | Threaded
Open this post in threaded view
|

Re: Evolution and spam

Martijn van de Streek
On Thu, 16 Mar 2006, Lionel Dricot (aka Ploum) wrote:

> To train bogofilter, I simply toke the first message of my inbox and
> learned it as ham with the cli. It was a short message.

But can your mum do that? :)

The spam filters should auto-train for 'ham' on non-spam folders, and
auto-train for spam on 'spam' folders (and auto-re-train on messages moved
from one place to the other)

Martijn
--
Sorry isn't an excuse when you do something stupid on purpose.

--
ubuntu-devel mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
Reply | Threaded
Open this post in threaded view
|

Re: Evolution and spam

Richard Kleeman
In reply to this post by Dennis Kaarsemaker
Dennis Kaarsemaker wrote:

> On do, 2006-03-16 at 11:38 +0100, VETSEL Patrice wrote:
>
>> Where are we going with spam filtering in evolution ?
>
> Currently nowhere unfortunately - and dapper is already way past feature
> freeze (even if dapper were delayed, feature free would remain intact).
> Please write a BetterSpamFiltering spec for dapper+1 with your ideas and
> let's hope it will be implemented.
>
> In Dapper I switched from spamassassin to bogofilter but it's not useful
> at all since you can't mark messages as non-juk if they are no yet
> marked as junk. And thus bogofilter (same goes for spamassassin) cannot
> learn good tokens. You have to resort to the command line for this.
>
> I really hope that there will be some improvement for dapper+1, which
> only is possible if someone with good ideas about it writes a spec.
>

A short term fix I have used is to shunt my email through a large ISP
mailbox (optimum online) which gets rid of about 80% of the spam and
then use thunderbird which has a mediocre spam filter to kill the rest.
Works quite well and I would estimate 80% of my raw mail is spam.


--
ubuntu-devel mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel