An assessment of spam email filtering approaches : a dissertation submitted in partial fulfilment of the requirements for the degree of Bachelor of Applied Computing (Honours)

Abstract

With the increasing emergence of electronic communication on the internet, spam, which is defined as unsolicited and unwanted email, usually in the form of advertisements for products and services, has become a major concern for both individuals and organisations. Although numerous articles or research projects have discussed anti-spam approaches, few of them report a systematic analysis of the effectiveness of technical approaches. The purpose of this dissertation is to measure the effectiveness of various anti-spam approaches. Six types of anti-spam approaches were evaluated: the Blacklist approach, the Whitelist approach, the Keyword Phrase approach, the Scoring approach, the Bayesian Approach and a combination approach. The results from the test suggest that the Bayesian approach is the most effective at both detecting spam messages and not mistakenly identifying legitimate emails as spam. It is one of the most efficient approaches from the point of view of the time to set it up and maintain it.