NoCeM, pl.* and false positives

[Adam W.]

Hi!

I finally made the solution to repost cancelled articles to my local
groups (chmurka.spam.*, available on news.chmurka.net). I process notices
from three sources:

- Eternal September
- usenet.ovh
- i2pn2

I saw no false positives from Eternal September (great! But it might be
also that they were already cancelled by usenet.ovh or i2pn2 and that's
why I didn't catch them in Eternal September's notices), but considerable
amount of them from usenet.ovh and i2pn2. Worth to take a look:

i2pn2:

<2a256fc7-b040-456c-b8f4-***@googlegroups.com>
<c94561f5-0358-420a-b2ac-***@googlegroups.com>
<6147a53c-a78d-4cd6-9fe2-***@googlegroups.com>
<6651fcd4-3121-4a5d-84d2-***@googlegroups.com>
<fbd9c9d6-c1ef-41d6-ac68-***@googlegroups.com>
<417faff2-162d-4dcd-847b-***@googlegroups.com>
<fd36bd85-686d-4d2a-a81f-***@googlegroups.com>
<d3bc4b28-fe63-4571-93a8-***@googlegroups.com>
<3bdc286e-2290-4835-b511-***@googlegroups.com>
<4d764c69-b289-411c-8189-***@googlegroups.com>
<a6320cb8-b1db-436a-8d4f-***@googlegroups.com>
<448ccf45-25fd-4b75-a2b9-***@googlegroups.com>

usenet.ovh:

<4c8681af-ec20-4975-840c-***@googlegroups.com>
<7df23659-c788-484b-8d59-***@googlegroups.com>
<4c8681af-ec20-4975-840c-***@googlegroups.com>
<368bc529-8331-49fb-a248-***@googlegroups.com>
<122d2020-9458-453f-9bd4-***@googlegroups.com>
<122d2020-9458-453f-9bd4-***@googlegroups.com>
<7a84fa49-8b68-40ab-a079-***@googlegroups.com>
<ff738b23-8e03-4725-973d-***@googlegroups.com>
<368bc529-8331-49fb-a248-***@googlegroups.com>
<ff738b23-8e03-4725-973d-***@googlegroups.com>
<6038abe5-c021-4c71-801c-***@googlegroups.com>
<997d2ef3-7d6b-4193-be6a-***@googlegroups.com>
<6038abe5-c021-4c71-801c-***@googlegroups.com>
<6038abe5-c021-4c71-801c-***@googlegroups.com>
<997d2ef3-7d6b-4193-be6a-***@googlegroups.com>
<f6d7a945-fa96-4396-acb1-***@googlegroups.com>
<f6d7a945-fa96-4396-acb1-***@googlegroups.com>
<5b9f5ea3-c9e9-4e5e-87fd-***@googlegroups.com>
<fb48efcd-64ac-40a4-85fd-***@googlegroups.com>
<f99f1471-0de9-4e85-9f21-***@googlegroups.com>
<3ee983ce-a373-4c56-8d41-***@googlegroups.com>
<cfa4021c-0f28-4368-9725-***@googlegroups.com>
<55b75539-123c-464b-afb3-***@googlegroups.com>
<a86e33a1-448c-499c-971c-***@googlegroups.com>

[llp]

Hi,

Post by Adam W.
Hi!
I finally made the solution to repost cancelled articles to my local
groups (chmurka.spam.*, available on news.chmurka.net). I process notices
- Eternal September
- usenet.ovh
- i2pn2
I saw no false positives from Eternal September (great! But it might be
also that they were already cancelled by usenet.ovh or i2pn2 and that's
why I didn't catch them in Eternal September's notices), but considerable

Since a few days, nocembot detect spam in all hierarchies.
But i found a tipo in utf-8 spam détection messages.
Please, tell me if you found new false positive.
The goal is to reach 0% false positive like in fr and big8 hierarchies.

Admin of news.usenet.ovh

[Retro Guy]

Post by Adam W.
Hi!
I finally made the solution to repost cancelled articles to my local
groups (chmurka.spam.*, available on news.chmurka.net). I process notices

Very nice! I save to a .mbox file that I scroll through, which can be
monotonous but somewhat helpful.

Post by Adam W.
I saw no false positives from Eternal September (great! But it might be
also that they were already cancelled by usenet.ovh or i2pn2 and that's
why I didn't catch them in Eternal September's notices), but considerable

It looks like the rule I had set in spamassassin detecting empty posts
probably accounts for most of these. (See discussion in e-s.support 'False
positives in spam filter?'. I removed that rule yesterday.

Considering we're sending out NoCeMs for close to 50,000 posts per day
(yesterday was 50,746 for i2pn2), there could be a few false positives that
I don't notice in the .mbox.

It is VERY helpful that someone let's us (NoCeM sources) know if/when they
see a few false positives. I've been pretty much dependent on only myself
to try to make my filtering work well, so thanks for posting!