Faculty Candidate: Gang Wang
480 Dreese Labs
2015 Neil Avenue
Columbus, Ohio 43210
Human Factors in the Security of Online Systems
Human factors are playing an increasingly important role in the security of today’s online systems. The successful operation of large systems like online social networks relies on well-behaved users contributing high-quality user content. An improved understanding of user behavior is instrumental to identifying misbehaving users and attacks in such systems, as well as to understand and defend against a new class of malicious crowdsourcing attacks.
In this talk, I will describe our efforts to improve security by characterizing and modeling complex human behaviors. First, I will describe our use of clickstream analysis to build accurate user behavior models that effectively capture anomalous users and previously unknown attacks on online systems. Second, I will describe our recent results on building highly “interpretable” behavior models, which help us to characterize user behavior at multiple levels of granularity and track their behavioral changes over time. Finally, I will briefly summarize my work on understanding and defending against human-based attacks (malicious crowdsourcing). I conclude by highlighting my future plans on using data-driven approaches to building human-centric security systems.
Gang Wang is a PhD candidate in Computer Science at UC Santa Barbara under the supervision of Ben Y. Zhao and Heather Zheng. His research focuses on human (user) aspects of Internet security. His work takes a data-driven approach to address emerging security threats in online social networks, crowdsourcing systems, and mobile networks and applications. He earned a B.E. in Electronic Engineering in 2010 at Tsinghua University. He is the recipient of PhD dissertation fellowship from UC Santa Barbara (2015), and best practical paper award from ACM SIGMETRICS (2013). His research has appeared in a broad set of top-tier venues in Security, Networking, Measurement, and Human-Computer Interaction.