Privacy protection for statistical analysis distributed database

摘要

PROBLEM TO BE SOLVED: To perform statistic analysis in which privacy such as the identification information of a data source and the detailed contents of live data can be safeguarded for private dispersion database.;SOLUTION: In compaction statistic, randomized data (hat) X and (hat) Y are acquired by separately subjecting data X and Y to first randomization. Subsequently, the randomized data (hat) X and (hat) Y are subjected to second randomization, thereby acquiring randomized data (tilde) X and (tilde) Y for a server and helper information Tx|x and Ty|y for a client. Here, T represents an experience distribution, and second randomization safeguards the privacy of the compaction statistic of the data X and Y. Next, the server acquires Tx, y. Lastly, a client acquires estimated (dots) Tx, y, by applying the helper information Tx|x and Ty|y in Tx, y. Here, "|" and "," between X and Y represent a conditioned distribution and a joined distribution respectively.;COPYRIGHT: (C)2014,JPO&INPIT