After testing their method on randomly generated and real-world networks, they discovered it outperformed many other methods in a range of settings
Malicious or fictitious users on internet networks have become the bane of the internet's existence. While many bemoan their increasing frequency, few have developed methods to track and expose them. A Ben-Gurion University of the Negev researcher has developed a new method to detect groups of anomalous users.
Their findings were just published in the peer-reviewed journal Neural Processing Letters.
"The advantage of this study is that we can detect anomalous groups of users, such as groups of fake profiles, rather than single users. Uncovering groups of fake profiles is a challenging and less explored task," says Dr. Michael Fire, head of the Data4Good Lab and a member of the Department of Software and Information Systems Engineering.
Image - Reuters