Privacy-Preserving Data Processing  Techniques Based on Cryptography

The rapid development of Internet provides us with tremendous opportunities for collaborative data computations. Many organizations and companies want to do joint work to get mutual benefit of their individual data. At the same time, with the powerful data processing techniques, it also can bring some privacy violation problems. In the real world, many organizations and individuals may have concern on their own privacy, and may be reluctant to share their own data for data mining unless their privacy will not be violated or misused by other parties. For this reason, a privacy-preserving system is needed to execute the collaborative data processing.

Our research is based on cryptographic secure multi-party computation techniques. It allows a set of n players to securely compute any agreed function on their private inputs and the corrupted players do not learn any information about the other players' inputs. That is, the only information learned by one party participating in the computation is the one that can be learned from the final output results of the collaborative data processing.

Related Papers:
1. Chunhua Su, Jianying Zhou, Feng Bao, Tsuyoshi Takagi, Kouichi Sakurai, "Two Party Privacy-Preserving Agglomerative Document Clustering", 3rd Information Security Practice and Experience Conference,  LNCS6477, pp. 193-208, HongKong, May, 2007.

2. Chunhua Su, Feng Bao, Jianying Zhou, Tsuyoshi Takagi, Kouichi Sakurai, "Privacy-Preserving Two-Party K-Means Clustering Via Secure Approximation", The 2007 IEEE International Symposium on Data Mining and Information Retrieval, pp.385-391, Niagara Falls, Canada. May, 2007.

3. Chunhua Su, Feng Bao, Jianying Zhou, Tsuyoshi Takagi, and Kouichi Sakurai. "A New Scheme for Distributed Density Estimation based Privacy-Preserving Clustering". (AReS'08), Proceedings of 2008 International Conference on Availability, Reliability and Security, pp. 48-57, IEEE Computer Society Press, Barcelona, Spain, March 2008.