AntispamLabTM

AntispamLab is a tool for automated testing (and evaluation) of email spam filters, developed in LCA lab at EPFL. It is preferably used on PlanetLab. Keywords: antispam lab, email, spam, filter, tool, testing, evaluation.

Use of the tool is completely automated:

More realistic test as compared to related work:

To learn more about the AntispamLab tool, see our paper [PDF] published at CEAS 2007 (The Fourth Conference on Email and Antispam, Mountain View, California, USA, August 2007).


Usage requirements    Downloads    Usage instructions    Usage examples    More info    Project team    Acknowledgements    Contact information


Usage requirements

Only per-email-server filters are currently supported, whereas per-email-client filters testing would require additional tool development.

To use the tool, you should have a PlanetLab account. An alternative is to create a set of Fedora Core 4 virtual machines locally, for which you need one or more real Linux machines on a local LAN with installed Xen. Another but less realistic alternative is to use many real Fedora Core 4 machines on a local LAN or even distributed over the Internet. For more details, see the README.txt file from the downloads section.


 Downloads

        IMPORTANT NOTES:

        FILES:


Usage instructions

The usage instructions are contained within the README.txt .

Two concrete and detailed examples of use of these instructions to evaluate spam filters are given in the next section.


Usage examples      


More info

To learn more about why did we make this tool, what it offers to you, and how we made it, see this conference paper:

AntispamLab – A Tool for Realistic Evaluation of Email Spam Filters. Slavisa Sarafijanovic, Luis Hernandez, Raphael Naefen, and Jean-Yves Le Boudec. In Proceedings of CEAS 2007, The Fourth Conference on Email and Antispam, Mountain View, California, USA, p. 121-127,  August 2007. [Paper-PDF] [Abstract-HTML]

If you would like to (easier) understand the script files of the tool, see the Sections 3 and 7 of this master thesis report [PDF].

Initial work on providing the main components of the tool and putting them to work together in a local environment created within one machine is documented in [PDF].

The current implementation of the AntispamLab provides basic testing functionalities, and we are working to improve its components such are: spam modeling, user behavior modeling, user contacts modeling.


Project team:

Slavisa Sarafijanovic (slavisa.sarafijanovic@epfl.ch), Luis Hernandez, Naefen Raphael and Jean-Yves Le Boudec


Acknowledgments

Special thanks to Marc Andre Luthi, for his advices regarding some implementation choices and help in solving some weird "installation doesn't work" cases. We would also like to thank MICS, the main source of funding for the AntispamLab project.


Contact information

For any questions regarding AntispamLab, write an email to slavisa.sarafijanovic@epfl.ch .