Exposing click-fraud using a burst detection algorithm

The explosive growth in the size and use of the World Wide Web continuously creates new great challenges and needs. One such need is dealing with click fraud, which aims at increasing clicks on certain ads and thus the profit of the websites which display them. In this work, we extend the concept of click fraud and redefine it as any pattern of clicks whose goal is to alternate the normal operation of a website in order to produce specific results. An indication of a click fraud may be a burst of clicks that can be simulated by an automated program or script. We deal with the problem of efficient real-time Click Fraud detection utilizing advanced data structures and exploiting their advantages concerning space and time required.