基于可逆方阵的隐私保护关联规则挖掘

Invertible Matrix-based Privacy-preserving Association Rules Mining

数据隐私问题引起人们的广泛关注,如何在分布式数据库的环境下挖掘关联规则成为研究的热点。该文探讨在垂直划分数据库中,如何在保护各方隐私数据的前提下挖掘全局频繁项集。各分布式数据库包含全局数据库的一部分属性,共同参与全局挖掘,同时各方不向外泄漏隐私数据。在商品服务器模型的研究基础上,提出一种基于可逆方阵的加密协议,对于垂直划分的分布式数据库,该协议具有较好的隐蔽性、高效性和准确性。

With the growing concern over dat- privacy-preserving problem, how to discover association rules from distributed databases becomes one of the hot topics of this field. This paper is devoted to study the problem of discovering global frequent itemsets from distributed vertically partitioned databases with the goal of preserving the confidentiality of each database. All sites are worked together to find global frequent itemsets without revealing private data, each one holds some attributes of global database. The paper presents an invertible matrix-based encryption protocol based on the research of commodity server model, which protocol is of great confidentiality, effectiveness and correctness for distributed vertically partitioned databases.