A review of spam email detection: analysis of spammer strategies and the dataset shift problem

Abstract Spam emails have been traditionally seen as just annoying and unsolicited containing advertisements, but they increasingly include scams, malware or phishing. In order to ensure the security integrity for users, organisations researchers aim develop robust filters spam email detection. Recently, most based on machine learning algorithms published in academic journals report very high performance, users are still reporting a rising number of frauds attacks via emails. Two main challenges can be found this field: (a) it is dynamic environment prone dataset shift problem (b) suffers from presence an adversarial figure, i.e. spammer. Unlike classical reviews, one particularly focused problems that constantly changing poses. Moreover, we analyse different spammer strategies used contaminating emails, review state-of-the-art techniques learning. Finally, empirically evaluate present consequences ignoring matter practical field. Experimental results show may lead severe degradation estimated generalisation with error rates reaching values up $$48.81\%$$ 48.81 % .