A fractal approach to social network spam detection
Alexandru Catalin Cosoi BitDefender
Carmen Maria Cosoi Nielsen
download slides (PDF)
Web 2.0 is all about social networks. It's all about user interaction, simplicity and usability. Content is submitted,
reviewed and downloaded by users only. For many, this can be considered the definition of freedom - but where few laws
apply, many bad things can happen.
Each social network represents a single or group of entities (networks). Each of these networks have a scale-free property
and their degree distribution follows a power-law, or in other words each of these networks is a social fractal - a
network that scales up and down with equal facility. In this social fractal, information distribution follows different
patterns and new connections are added by using simple mathematical rules.
This paper will try to approach the spam detection problem in an innovative way, especially social network spam, by
using fractal network properties. We will show different methods for finding anomalies in these fractal patterns and ways
to use these features for security purposes. Further on, the developed methods will allow us to identify unwanted profiles,
unsolicited messages, undesired malware and provide phishing protection.
