Abstract

Spam is a given on the internet—it’s estimated that 55% of all email is spam, and each user receives almost 70 such emails per month [1]. It’s used as a vehicle for all sorts of scams: phishing, malware, 419 advance-fee fraud and many other forms of malicious message are all communicated via spam emails. However, real-world prevalence seems much lower due to filtering mechanisms put in place by email providers—but how do they work?

This talk explores how a spam filter is built, how they have changed over time, and various pieces of interesting Computer Science that lurk within their implementation. Many of these techniques can also be generalised beyond spam for application in other domains of classification.

[1] Symantec, https://www.symantec.com/content/dam/symantec/docs/reports/istr-23-2018-en.pdf