1 |
A couple hours ago I committed a new ebuild for a fast spam filter based |
2 |
on Bayesian filters, popular since the publication of Paul Graham's |
3 |
paper A Plan For Spam: http://www.paulgraham.com/spam.html |
4 |
|
5 |
You can find the ebuild on bugzilla: |
6 |
http://bugs.gentoo.org/show_bug.cgi?id=10435 |
7 |
|
8 |
I've only tested it on x86 and it's working pretty well, I much prefer |
9 |
it over all the other spam-filters I've tried, including other |
10 |
Bayesian-based ones. |
11 |
|
12 |
With bogofilter's ebuild there is also one for Judy's libraries, general |
13 |
purpose dynamic array's C libraries which claims to be much faster and |
14 |
efficient than other data structures schemes, like binary trees, |
15 |
B-trees, linked lists, ... |
16 |
This I am interested in people on other architectures to see if it |
17 |
works, since the autoconf version I used on the ebuild seems a work in |
18 |
progress. In case there is any problem on PPC, sparc or whatever I'll |
19 |
try to modify the ebuild :) |
20 |
|
21 |
|
22 |
PS If you are aware of any other spam-filter tool which you consider |
23 |
worth looking at, please tell! |
24 |
I know about Spam-Assassin, which is just too big IMHO, spamprobe, based |
25 |
on Bayesian methods, but unusable to filter each and every mail on a |
26 |
production system, even a small one with only a few active users. |
27 |
|
28 |
|
29 |
-- |
30 |
Javier Marcet <jmarcet@×××××.com> |