Home  Blog  Travel  Party  Free software  Writings  About me  Contact

Similar Email Search

(This page used to be in a wiki of the Helsinki Institute of Physics. As the server is not online any more, I've moved it here. The software is not being maintained and not very usable in its current form.)

This paper presents a prototype implementation of a mail server featuring a content-based message search function. The user can choose an interesting email and ask the server to retrieve similar ones. See the paper below for a description of the project.

DocumentDateAvailable formats
Implementing a Similar Messages Search Function in an IMAP Mail Server 30 Sep 2002 pdf

The system was developed on top of the DBMAIL mail server, the Bow statistical language modeling toolkit and the Mutt email client. You can download the modified source code below. DBMAIL and Mutt are licensed under the GNU General Public License and Bow under the GNU Library General Public License (version 2).

dbmail-cvs-similarsearch-2002-09-30.tar.gz 30 Sep 2002
bow-20020213-similarsearch-2002-09-30.tar.gz 30 Sep 2002
mutt-1.4i-similarsearch-2002-09-30.tar.gz 30 Sep 2002

Apologies for missing documentation and some variables referring to my local installation in the code. Just as I finished the work I moved to Japan and didn't have time to do a proper cleanup and write installation instructions. However, the document should give you a relatively good picture what the project was about.

This page has been mainly written in 2002. A few minor modifications were made when moving the page here.

Arto Teräs <ajt@iki.fi> — last update 2016-07-03.