An Evaluation of Naïve Bayesian Anti-Spam Filtering Techniques

An efficient anti-spam filter that would block all spam, without blocking any legitimate messages is a growing need. To address this problem, this paper examines the effectiveness of statistically-based approaches Naïve Bayesian anti-spam filters, as it is content-based and self-learning (Adaptive) in nature. Additionally, the authors designed a derivative filter based on relative numbers of tokens. The authors train the filters using a large corpus of legitimate messages and spam and test the filter using new incoming personal messages. More specifically, four filtering techniques available for a Naïve Bayesian filter are evaluated.