Challenges in Spam Filtering Research

Abstract:

Spam is one of the most ubiquitous and most annoying phenomena on the
Internet; it is also one of the oldest successful e-commerce business
models. The fight against spam, while not as promising from a commercial
point of view, poses a lot of challenging research questions.

In my talk I will highlight several pivotal moments from the history of
spam and the fight against spam. Starting with the appearance of the first
spam messages and of the first rule-based filters, I will turn to the
groundbreaking debut of statistical filters -- and what happened since.

The development of spam and anti-spam filtering techniques is a typical
case of co-evolution. Every change in one of them will be responded, in
some way or other, by the other. Keeping this in mind, I will treat some of
the challenges that the spam filtering community currently faces, such as
image spam, the need for collaboration, and the "leaking" of spam into more
and more areas including the Web, instant messaging, VoIP, and others. I
will present recent approaches for dealing with two of these developments
in greater detail.

Trying a glimpse into the future I will finally pose the question whether
we can ever expect to get rid of spam.