Elisavet Kozyri (Harvard University)


Friday, December 6, 2019, 3:00pm to 4:30pm


Maxwell-Dworkin Room 323
Title: Expressive information flow labels for reclassification and permissiveness
Abstract: Information flow labels specify restrictions on the associated data and any information derived from this data. However restrictions on derived information may depend on the operations involved in this derivation. We define reactive information flow (RIF) labels that can express such dependencies. We give examples of RIF labels and a static enforcement mechanism. In our attempt to port RIF labels into a dynamic enforcement mechanism, we realized the need for label chains, where each element of a chain represents the sensitivity of its predecessor. We describe enforcers for label chains of arbitrary length and we study whether longer label chains improve permissiveness.
Joint work with Fred B. Schneider, Josée Desharnais, Nadia Tawbi, Andrew Bedford, Owen Arden, and Andrew C. Myers.
Elisavet Kozyri currently a postdoctoral fellow in Computer Science at Harvard University working with Prof. Stephen Chong.
 Her research enhances the expressiveness of information flow policies to allow for sound and more permissive enforcement mechanisms.
She received her Ph.D. in Computer Science from Cornell University under the supervision of Prof. Fred B. Schneider.