Privacy Preserving Data Mining

摘要: In this paper we introduce the concept of privacy preserving data mining. our model, two parties owning confidential databases wish to run a mining algorithm on union their databases, without revealing any unnecessary information. This problem has many practical and important applications, such as in medical research with patient records. Data algorithms are usually complex, especially size input is measured megabytes, if not gigabytes. A generic secure multi-party computation solution, based evaluation circuit computing entire input, therefore no use. We focus decision tree learning use ID3, popular widely used for problem. present solution that considerably more efficient than solutions. It demands very few rounds communication reasonable bandwidth. each party performs by itself same order ID3 its own database. The results then combined using cryptographic protocols, whose overhead only logarithmic number transactions databases. feel result substantial contribution, demonstrating can be made practical, even complex problems large inputs.