针对海量涉密隐私数据高效安全检索的需求,提出了一种基于密文的全文检索系统---Mimir.Mimir基于B+树构建了一种安全密文全文索引结构,Mimir检索过程完全在密文环境下进行,保证了系统的安全性和存储信息的私密性.与传统的全文检索系统相比,Mimir密文索引中没有存储索引词的位置信息和词频信息,可以有效地抵御已知明文攻击、选择明文攻击和词频统计攻击.对Mimir密文全文检索系统进行了性能测试,实验结果数据表明,Mimir密文全文检索系统在确保高安全性的同时,也具有很好的检索时间和存储空间性能.
For the need of efficient and secure retrieval on the vast privacy data,we propose afull-text retrieval system based on cryptograph,named Mimir.Mimir constructs a secure crypto-graph full-text index based on the B+tree structure.All the retrieval processes of Mimir areabsolutely processed under cryptograph,and it ensures the data security and privacy in Mimir.Comparing with the traditional full-text retrieval systems,Mimir cryptograph index does notstore the word position offset and the frequency of token.Through the security analysis,wedemonstrate that Mimir index structure can effectively resist the known plaintext attack,thechosen plaintext attack,and the statistical attack.We carry out some experiments to evaluate theMimir’s performances.The experimental results showed that Mimir cryptograph full-textretrieval system achieves a good performance on the retrieval time and stored space without leakingprivacy.