Hello,
I began to write a new thread about spam control, then remembered this
recent one on a similar topic.
Integrating spam control more deeply into all of our tools and
services - including particularly MediaWiki - is important for many
audiences.
Is there an overview of current anti-spam tech (for MW in particular,
but related: for our preferred Ticket-handling and Mailing-list
toolchains), and projected roadmaps? Comparisons with the best known
proprietary tools, to see what remains to be built? This strikes me
as something that we and the FSF and other groups could collaborate
on.
Sam
On Fri, Apr 18, 2014 at 3:26 PM, Tilman Bayer <tbayer(a)wikimedia.org> wrote:
Hi Cristian,
On Fri, Apr 18, 2014 at 11:25 AM, Cristian Consonni
<kikkocristian(a)gmail.com> wrote:
2014-04-18 0:46 GMT+02:00 Lodewijk
<lodewijk(a)effeietsanders.org>rg>:
* There is a specific need, a conference we could
support, a developer
event or something we could help out with. There is a clear goal, and it is
one-time. We have a clear benefit. For example: helping OTRS to become less
messy.
[OT]
As a Wikimedia list moderator (just a couple, the Wiki Loves Monuments
ones) I am ready to pay gold for anything that would reduce the amount
of spam in the queues be it a better spam filter, a system to delete
spam from multiple lists at once, magic or whatever!
Cristian
p.s.: "pay gold" is a figure of speech, but I would definitely
personally support such a project!
yeah, as moderator on some other lists, I share your feelings about
spam. Note though that SpamAssassin is already running in our Mailman
installation [...]
The current blocker is that it is generating some false positives,
i.e. someone would need to spend time to tweak and test the settings:
https://bugzilla.wikimedia.org/show_bug.cgi?id=56525
So it looks we have to some of our own homework first here. But if it
turns out that there are limitations in SpamAssassin which we could
help them overcome to make it more effective for our purposes, that
might be a very good use of donor money.
(As an example concerning your suggestion for "a system to delete spam
from multiple lists at once": I'm not sure if SpamAssassin is
currently integrated with Mailman in a way that enables it to learn
from list moderator actions immediately. See e.g.:
https://en.wikipedia.org/wiki/SpamAssassin#Bayesian_filtering
http://www.jamesh.id.au/articles/mailman-spamassassin/ --> "The Future" )