A Fuzzy Clustering Approach to Filter Spam E-Mail

Spam email, is the practice of frequently sending unwanted email messages, usually with commercial content, in large quantities to a set of indiscriminate email accounts. However, since spammers continuously improve their techniques in order to compromise the spam filters, building a spam filter that can be incrementally learned and adapted became an active research field. Researches employed machine learning techniques which have been widely used in solving similar problems like document classification and pattern recognition, such as Naive Bayesian, and Support Vector Machine. In this Paper, we examine the use of the fuzzy clustering algorithm (Fuzzy C-Means) to build a spam filter. The proposed use of the Fuzzy has been tested on different data set sizes collected from Spam assassin corpora by real user's emails. After testing Fuzzy C-Means using Heterogeneous Value Difference Metric with variable percentages of spam and using a standard model of assessment for the spam problem, we demonstrate the potential value of our approach.