Abstract: All privacy-preserving distributed mining algorithms can not deal with collusion problem efficiently. Based on Clifton’s secure multi-party association rule mining algorithm, this paper proposes Advanced Kantarcioglu-Clifton Algorithm(ACKA) algorithm, which adopts secure multi-party sum by collaboratively constructing and solving equation sets. It is proved that the algorithm can not only deal with collusion attack perfectly, and detect malicious substitution.

Key words: privacy-preserving, distributed, association rule mining

摘要：

以往各种分布式数据挖掘隐私保护算法无法有效解决串通问题，从而限制了其大规模应用，针对上述问题，在Clifton分布式关联规则安全挖掘算法的基础上，提出AKCA算法。采用各站点联合建立并求解方程组的安全多方求和方法。结果证明经过改进的算法能够抵御串通攻击，不借助签名验证也能发现恶意篡改。

关键词: 隐私保护, 分布式, 关联规则挖掘

CLC Number: 

• TP311