Anonymization with Differential Privacy / Secure Data Structures with Intel SGX

Huge amounts of textual data are processed every day using text mining and information retrieval techniques to assist us with analyzing, organizing and retrieving text documents. In many cases, it is desirable that the authors of such documents remain anonymous: They can reveal sensitive information about its authors, and critical news articles or customer feedback could cause retaliation or worsening business relations. However, simply removing personally identifying information from a document is insufficient, since so-called authorship attribution methods can reidentify the author just from the writing style.

In this talk, Ben will discuss the importance of privacy for textual data and present SynTF (published at SIGIR’18), the first text anonymization method based on differential privacy. SynTF protects the identities of the authors while preserving semantics to allow e.g. privacy-friendly topic inference tasks. Differential privacy is a statistical notion of privacy that won the 2017 Gödel prize and works by injecting randomness in the resulting data and thus providing plausible deniability for the authors.

In this talk, Benny Fuhry will mainly present HardIDX, a provably secure database index concept that won a best paper award. It is a hardware-based approach, leveraging Intel's SGX, for search over encrypted data. Only the security critical core, i.e., the search functionality, is implemented in the trusted environment and it resorts to untrusted software for the remainder. HardIDX is logarithmic in the size of the index and searches are performed within a few milliseconds. The implementation has a very small code and memory footprint yet still scales to virtually unlimited search index sizes, i.e., size is limited only by the general - non-secure - hardware resources. He will also talk about other current research projects concerning further secure and fast data structures and database primitives.

Ben Weggenmann is a researcher and Ph.D. candidate at SAP Security Research. His research interests cover computer security and data anonymization with a focus on differential privacy. He has previously worked in the industry on cryptography and embedded security and holds a Diplom degree in computer science from Ulm University and a B.Sc. (Hons.) in mathematics from Monash University.

Benny Fuhry is a permanent researcher and Ph.D. candidate at SAP's Security Research division. He holds a bachelor’s and master’s degree from the Karlsruhe Institute of Technology (KIT), Germany. His main contribution to research were an approach to encrypt analytical web applications and a best paper winning concept for a provably secure database index that uses Intel SGX. He currently researches further data structures and database primitives that are secure and fast to enable secure big data analysis in the cloud. Furthermore, he was SAP's leader for the EU research project TREDISEC (Trust-aware, REliable and Distributed Information SEcurity in the Cloud).