中文摘要：

现有的敏感属性多样性模型均没有考虑敏感值间的语义相似性,不能很好地抵制近似攻击.为此,本文在（k,l）-匿名模型的基础上,提出可抵制近似攻击的（k,l,e）-匿名模型,该模型要求匿名数据中的每个等价类都满足k-匿名约束,且等价类中至少有l个互不e-相近的敏感值.实验结果表明,满足（k,l,e）-匿名模型的匿名数据比满足（k,l）-匿名模型的匿名数据具有更高的多样度,能够更有效地保护个体隐私.

英文摘要：

Existing sensitive attributes diversity models do not consider the semantic similarity between sensitive values and can not resist approximate attack. To solve the problem,this paper proposes a（k,l,e）-anonymity model based on（ k,l）-anonymity model.The proposed model requires that each equivalence class satisfy k-anonymity constraints,and at the same time there exist at least l sensitive values in one equivalence class which are not e-similar each other. The paper also proposes a（k,l,e）-KACA algorithm to implement（k,l,e）-anonymity. Experimental results show that the anonymous data satisfying（k,l,e）-anonymity have higher diversity than that satisfying（k,l）-anonymity model,so（ k,l,e）-anonymity model can protect privacy more effective than（ k,l）-anonymity model.