摘要： 为解决外包关联规则挖掘中的隐私保护问题，针对现有基于标准布隆过滤器算法时间效率低、可逆性较差等不足，提出一种基于独立映射空间布隆过滤器的算法。将原始事务数据库转换成布隆过滤器的形式，根据转换后每个事务向量的海明重量进行事务压缩，利用矩阵列向量进行“与”运算，计算候选项集的支持度，从而得出频繁项集。实验结果表明，与原算法相比，该算法在保证误判率的同时，能提高时间效率，具有良好的可逆性和安全性，实用性更强。

关键词: 外包, 关联规则, 频繁项集, 数据挖掘, 隐私保护, 布隆过滤器

Abstract: In order to solve the problem of privacy-preserving in outsourcing association rule mining, this paper proposes an algorithm which is based on independent mapping space Bloom filters to overcome the low time efficiency and poor reversibility. The algorithm translates the original transaction of database into the form of Bloom filter. It compresses the affairs according to the Hamming weight of each transaction vector transformed. It calculates the support of candidate itemsets through “and” every column vector of matrix. It obtains the frequent itemsets. The experimental results show that, compared with the original algorithm, the algorithm improves the time efficiency while ensures the low misdiagnosis rate, furthermore, it has good reversibility, high safety and stronger practicality.

Key words: outsourcing, association rule, frequent itemset, data mining, privacy preserving, Bloom filter

中图分类号: 

• TP309