Title: When Do Data Mining Results Violate Privacy?

Privacy-preserving data mining has concentrated with obtaining valid results when the input data is private. An extreme example is Secure Multiparty Computation-based methods, where only the results are revealed. However, this still leaves a potential privacy breach: Do the results themselves violate privacy? This talk explores this issue, presenting a framework under which this question can be addressed. Metrics are proposed, along with analysis that those metrics are consistent in the face of apparent problems.

This is joint work with Jaishun Jin and Murat Kantarcioglu at Purdue.

Title: Handling incompatible formats and erroneous data in the context of privacy-preserving data mining

The Research Community is very interested in medical data. Hence, some means to integrate such data is becoming imperative and a necessity. Our research identifies some of the difficulties encountered in medical data integration. Our current focus is on two of the issues related to such difficulties, namely Incompatible Data Formats and Erroneous Data. We provide two methods to address these two issues.

Title: Overview of database privacy research at Stanford

We provide a brief discussion of projects that are being pursued as a part of the Database Privacy group at Stanford by Rajeev Motwani, Hector Garcia-Molina and their students. This talk discusses our work related to individual centric privacy (as against P3P standards for corporate privacy policies), data perturbation techniques for statistical databases, privacy preserving indexing of documents on the network, secure computation of quantiles in the union of two databases and searching on encrypted data.