Abstract: A technique is presented to extend the cross two-table frequent itemsets mining method to the cross three-table frequent itemsets mining. The common attributes count set for three-table frequent itemsets is treated as the parameter for secure three-party protocol, a cross three-table frequent itemsets mining algorithm with privacy preserving is proposed. This algorithm not only can mine the cross-three table frequent itemsets but also can preserve the private data. The analysis and experimental results show that the presented algorithm is secure, efficient and scalable.

Key words: privacy preserving, cross-table mining, frequent itemsets, secure three-party

摘要： 给出将跨两表频繁项集挖掘方法扩展到跨三表频繁项集挖掘方法的技术，以三表频繁项集的公共属性记数集作为三方安全协议的参数，设计一个跨三表频繁项集挖掘的隐私保护算法，以便在挖掘求出跨三表频繁项集的同时保护三表中的隐私信息。理论分析和实验结果表明，算法安全、高效，具有可扩展性。

关键词: 隐私保护, 跨表挖掘, 频繁项集, 三方安全