emc, the anti-spam tool for IMAP users

• Introduction
• Quick start
• Download
• Does it work?
• Acknowledgements

Introduction

Welcome to the home page for emc. emc is built to be an easy to use solution for people with an IMAP server. It is email client independent. Whether you use Gnus, Outlook or Mozilla, this tool helps you to combat spam.

It works by scanning your INBOX folder and move the spam to a different location. It recognizes spam by first building a database from messages you have archived. But everything is explained clearly in the manual.

Quick start

Pressed for time? Follow these steps:

1. Install it by unzipping the distribution in some new directory.
2. Edit emc.config, just follow the comments
3. Run emc_scan. It will complain if your configuration file is incorrect. It will produce probabilities.txt.
4. Now you're ready to run emc_classify to classify your email.
5. Optionally you can move the binaries to a more convenient location. In that case move your emc.config and probabilities.txt files to the .emc directory in your home directory. Create the .emc directory in your home directory if it doesn't exist.

Download

The last release was made on 2006 May 2. The changes are described in the ChangeLog. You can download the source code, but compilation might be impossible as it depends on not yet released libraries.

emc binary downloads

Platform	File name	Version
Linux (compiled on Linux 2.4.29 with glibc-2.2.93-5).	emc-release-0.7.4-Linux.zip	0.7.4
Linux (compiled on Linux 2.4.29 with glibc-2.2.93-5).	emc-debug-0.7.4-Linux.zip	Debug version of above
Windows (compiled on Windows 2000 SP2 with ISE 5.5 and Microsoft C 6.0)	emc-release-0.7.4-Windows.zip	0.7.4
Windows (compiled on Windows 2000 SP2 with ISE 5.5 and Microsoft C 6.0)	emc-debug-0.7.4-Windows.zip	Debug version of above

Does it work?

Good question. It works for me, somewhat. I never, even with other tools, could approach the phantastic results reported by Paul Graham or William S. Yerazunis.

So I still have mail that is not classified as spam, for example mail that just contains some attachment. And sometimes mail is misclassified as spam when it isn't. Let me know about your experiences. Suggestions welcome!

Acknowledgements

emc is mainly based on the improvements discussed by Gary Robinson in Spam Detection. It uses Greg Louis formula's.

Try running emc_classify with the --algorithms option to get an idea what spamminess would be reported by various algorithms:

emc_classify --algorithms