Conclusion

Bayesian spam filtering is an incredibly powerful statistical technique—with acceptable computational complexity—for identifying spam messages. Bayesian techniques address many weaknesses of other methodologies:

  • The entire message can be examined, not just special parts.

  • All words are significant, not just special keywords or addresses.

  • Updating is, in practice, infrequent (never more than one or two email messages per week through the training program; often none).

  • So far, spam attacks on Bayesian filters have been relatively unsuccessful.

  • When combined with other techniques, Bayesian filters can be a very strong component of an institution’s global spam system.

The CRM114 system is just one of many available Bayesian spam filters; commercial ...

Get Slamming Spam: A Guide for System Administrators now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.