Colloquium Details
Mining Behavioral Data: Inference Threats and Privacy Countermeasures
| Author: | Stratis Ioannidis, Research Scientist Yahoo Labs |
|---|---|
| Date: | March 12, 2015 |
| Time: | 15:30 |
| Location: | 220 Deschutes |
Abstract
Online services routinely mine behavioral data to perform tasks crucial to their daily operations, including targeted advertising and personalized recommendations. This practice has raised privacy concerns among consumer advocacy groups, regulatory bodies, and the public at large. We study threats arising from the unfettered access to behavioral data, and show that personal information can be successfully inferred from seemingly innocuous online behavior. We also show that behavioral data by multiple users sharing a single online account can be identified as such and separated, establishing that the above threats are relevant even in the presence of multiple users. Finally, we present methods mitigating privacy risks while preserving the utility of services mining behavioral data, using both statistical and cryptographic techniques.
Biography
Stratis Ioannidis is a research scientist at Yahoo Labs. He received an M.Sc. (2004) and a Ph.D. (2009) in Computer Science from the University of Toronto, Canada, and a B.Sc. (2002) in Electrical and Computer Engineering from the National Technical University of Athens, Greece. Prior to joining Yahoo, he was a researcher at Technicolor's research centers in Paris, France, and Los Altos, CA.