Social bots are becoming far more sophisticated and threatening in today's online social networks (OSNs). There have been reports of various attacks, abuses, and manipulations based on social bots, such as infiltrating Facebook or Twitter, launching spam campaigns, and performing financial fraud. Unfortunately, state-of-the-art approaches to detecting social bots are less than satisfactory: they are only executable by OSN providers (such as Facebook or Twitter), need to utilize users' private account data, and are fairly resource-consuming.
Researchers from the University of Oregon, including first-year Ph.D. student Yebo Feng (pictured on the left), Profs. Jun Li and Lei Jiao, and Prof. Xintao Wu from the University of Arkansas have invented a new learning-based approach that only needs content-agnostic flow-level network traffic data (e.g. NetFlow) to detect social bot activities in OSNs. This innovative approach has many significant advantages: it is privacy-preserving as it does not rely on users' private account information, it is scalable since it only analyzes highly summarized flow-level data, and it is easy to deploy because both OSN providers and network providers have the required information to apply this approach.
The paper describing this work, "BotFlowMon: Learning-Based, Content-Agnostic Identification of Social Bot Traffic Flows," won the Best Paper Award this June in the IEEE Conference on Communications and Network Security (CNS), a well-known conference focused on network security. This paper is expected to be available in the IEEE Xplore digital library soon.
Yebo Feng is a Ph.D. student in the UO's Center for Cyber Security and Privacy. His research interests include data analysis and network security. Prof. Lei Jiao is an Assistant Professor at the Department of Computer and Information Science, University of Oregon who is broadly interested in the mathematical and algorithmic sciences for systems and networks. Prof. Jun Li is the director of the Network & Security Research Laboratory in UO's Department of Computer and Information Science and serves as the Founding Director of the interdisciplinary Center for Cyber Security and Privacy. His research interests include computer networking, Internet of things, and cyber security and privacy. Prof. Xintao Wu is a professor in the Computer Science and Computer Engineering Department at the University of Arkansas whose major research interests include data mining and privacy, fraud detection in social networks, fairness aware learning, and big data analysis.
This research is supported by a grant from the National Science Foundation.