Fraudulent user profiles – bots – are a serious and growing concern on social media. By some estimates, as many as 48 million Twitter accounts and 270 million Facebook accounts are phony, designed for nefarious purposes from ruining reputations to influencing shoppers and voters.
Now, researchers from Israel’s Ben-Gurion University (BGU) of the Negev and from the University of Washington in Seattle say they have developed a generic method to detect fake accounts on most types of social networks, including Facebook and Twitter.
According to their study published in the journal Social Network Analysis and Mining, the new method is based on the assumption that fake accounts tend to establish improbable links to other users in the networks.
“With recent disturbing news about failures to safeguard user privacy, and targeted use of social media by Russia to influence elections, rooting out fake users has never been of greater importance,” said Dima Kagan, lead researcher and a PhD student in BGU’s department of software and information systems engineering.
The algorithm consists of two main iterations based on machine-learning algorithms. The first constructs a link prediction classifier that can estimate, with high accuracy, the probability of a link existing between two users. The second iteration generates a new set of meta-features based on the features created by the link prediction classifier.
These meta-features are used to construct a generic classifier that can detect fake profiles in a variety of online social networks.
“We tested our algorithm on simulated and real-world data sets on 10 different social networks and it performed well on both,” Kagan reported.
“Overall, the results demonstrated that in a real-life friendship scenario we can detect people who have the strongest friendship ties as well as malicious users, even on Twitter. Our method outperforms other anomaly detection methods and we believe that it has considerable potential for a wide range of applications particularly in the cybersecurity arena,” the study authors said.
The algorithm can also be used to reveal the influential people in social networks.
The Israeli researchers involved in this project previously developed the Social Privacy Protector (SPP) to help users evaluate their friends list in seconds to identify which have few or no mutual links and might therefore be phony profiles.
Other researchers who contributed to the present study are former BGU doctoral student) Michael Fire of the University of Washington and Prof. Yuval Elovici, director of the Telekom Innovation Labs@BGU, director of Cyber@BGU and a faculty member of BGU’s department of software and information systems engineering.
The study was supported by the Washington Research Foundation Fund for Innovation in Data-Intensive Discovery and the Moore/Sloan Data Science Environment Project at the University of Washington.