In today’s information-driven world, the Internet has become an integral part of our lives; we use it to gain and share knowledge and experience. In particular, one delay-tolerant application of the Internet, email, has become a powerful, essential, and inexpensive way to exchange information. As a consequence, this application has become a potential avenue to abuse the Internet: unsolicited email messages can be sent in bulk--known as spam. One of the big challenges for researchers is developing techniques to filter these spam emails.
In this paper, Islam et al. develop an “email classification technique by adopting a grey list (GL) analyzer through an integrated classification system.” This is achieved “by using different aspects of learning-based anti-spam filtering” techniques. The main objective of this analyzer is to reduce the problems with false positives (FPs). The paper discusses in detail related works and their drawbacks in terms of FP problems.
The classification of emails to identify the GL is carried out elegantly, through mathematical analysis. Once the GL emails are identified, two techniques are used to analyze them: user feedback and sender verification.
The effectiveness of the analyzer is illustrated through numerical results, by comparing its performance with three other algorithms, using a public email dataset. Without providing any illustrations, the authors admit that the proposed analyzer has some drawbacks, such as additional complexity, cost, and time overhead.
In summary, this paper is an important contribution to combating spam emails.