A multi-layer model to detect spam email at client side

A solution to spam emails remains elusive despite over a decade long research efforts on spam filtering. Among different spam detection mechanisms that have been proposed, Naive Bayesian Content Filtering has been very popular and has attained a reasonable level of success. SpamBayes is one such content filtering spam detection tool based on Naive Bayesian classification using textual features. It is easy to deceive the learning techniques focusing only on textual attributes. Hence, in this paper we propose a multi-layer model that imposes, on top of SpamBayes, a second layer of non-textual filtering that exploits alternative machine learning techniques. This multi-layer model improves the accuracy of classification and eliminates the grey email into spam and ham emails. The experimental results of this model are quite encouraging.