[Savannah-hackers] submission of Bayesian Spam Filter - savannah.gnu.org

[crgutier]

A package was submitted to savannah.gnu.org
This mail was sent to address@hidden, address@hidden


Cristian Gutierrez <address@hidden> described the package as follows:
License: gpl
Other License: 

Package: Bayesian Spam Filter
System name: bsf
Type: GNU

Description:
The BSF (Bayesian Spam Filter) project aims to develop an automatic filter for 
SPAM (junk or not solicited email).

The vast common filters for SPAM works with a (big) set of rules with words 
that spammers most use in their mail.

A Bayesian filter works with the statistics from your mailbox. It automatically 
save keywords from your normal inbox and from your spambox so it determines 
whether a new incoming mail is spam or not, using the Bayesian\\\'s rule of 
combining probabilities. Check this url for a simple explanation:
http://www.paulgraham.com/naivebayes.html

This approach means: Faster execution (no big list of rules, they are 
determined statistically) and better accuracy.

Actually, there are some other similar projects. However, our main goals are:
- Good Accuracy (we hope 95% of effectivity).
- Ease of use for the end user.
- Integration with other MUAs (Mail User Agents like mutt, pine, Gnome 
evolution, etc).
- Fast execution.
- Automatic addition/deletion of keywords.

The project is actually in full development in Perl. We are testing differents 
algorithms/approach with differents sources of real mail.

http://www.dcc.uchile.cl/~crgutier/bsf.tar.bz2

Other Software Required:
- perl
- DB_File perl module

Other Comments:
This is the second revision of the contents of the distribution file since the 
first attempt to register the project. It should be all in order now :o)