This is the code for the hash server based spam detector. It uses traffic analysis in a privacy-preserving way to robustly detect the number of people recieving similar emails.
A presentation.
Hash client (written in perl)
Most people will be interested in the hash-client perl script. This script is setup to work with a mail delivery agent such as 'maildelivery' or 'procmail'. It annotates the subject of emails which fit the spam filter with 'dup-X' where 'X' is the median count returned. Directions for setup are in the README file.
Hash server (written in ocaml)
The hash server code works with a dbm database for persisent storage.