Burfiks: bayesian filter for the web
An elementary theorem of probability theory known as Bayes' Theorem has been made popular by Paul Graham to identfy automatically spam messages.
Burfiks is a Firefox extension that applies a bayesian filter to sort the links found in a web page according to their interest. While browsing the net, the user can classify a document as interesting or not interesting. Burfiks uses these informations to build a probability distribution that has good properties to predict if a page will be classified by the user as interesting or not. All the links of the current web page currently are sorted and displayed according to this probability distribution.
Disclaimer: huge network traffic
Burfiks silently retrieves all the
You have been warned.
Burfiks authors are not responsible for the consequences of using Burfiks, not even in any case.
Download
The Burfiks package is compatible with any version of Firefox newer than 1.5, and it is available for Linux, Mac (ppc and x86) and Windows. Currently, for testing purposes, we distribute x86 GNU/Linux binaries only.
The current version is 1.1469
Advanced users might be interested in the packages below. Follow these links only if you know what you are doing. And if you know Burfiks manual by heart.
- GNU/Linux - x86
Burfiks sources are available too. OCaml is needed to compile the Burfiks server.
Documentation
- Online manual
License
Burfiks is distributed under the terms of the GPL (GNU Public License) version 2, available from http://www.gnu.org/licenses/gpl-2.0.txt
Burfiks includes code from Xavier Leroy's SpamOracle and Juliusz Chroboczeck's Polipo.
Contact
- Francesco Zappa Nardelli
INRIA, Rocquencourt
francesco.zappa_nardelli [at] inria.fr - Patala Arun Kumar, Vidyut Ghosal
Indian Institute of Technology Kanpur, India
{parun,vidyut} [at] cse.iitk.ac.in