Research On Ciphertext Retrieval And Sorting Method Supporting Fuzzy Multi-keyword Search

Cloud storage,as an important application area of cloud computing,is being used for data outsourcing by more and more enterprises and individual users.In order to protect the privacy of outsourced data,data owners should encrypt these data before outsourcing data.However,the retrievability of ciphertext data is relatively poor,so Searchable Encryption(SE)came into being.At present,there are a lot of researches on the functions of fuzzy query,multi-keyword efficient query and correlation ranking,but there are few SE schemes that can integrate the above three functions at the same time.Firstly,this paper deeply studies the methods of fuzzy query,multi-keyword efficient query and correlation ranking in the current SE field.After analyzing the reasons why the existing ciphertext retrieval schemes are difficult to expand to achieve the above three functions at the same time,a new correlation measurement algorithm CL-TFIDF(Cross Layer TFIDF)is proposed to realize the fuzzy multi-keywords.The ciphertext ranking is proved to be correct by theory.Unlike the existing correlation measures such as the classical correlation score TFIDF,which assigns to the whole word such as "fine",CL-TFIDF assigns to two-letter binary elements such as "fi","in" and "ne".Then,by combining CL-TFIDF with VSM(Vector Space Model)and secure KNN technology,this paper proposes and implements a multi-keyword ciphertext fuzzy ranking search scheme(MFRSE),which can simultaneously complete three functions of fuzzy query,multi-keyword query and correlation calculation in the calculation of a vector inner product.The Top-k files with the highest correlation can be obtained by the ranking algorithm with linear time complexity,and the high accuracy of MFRSE scheme is proved by the experimental comparison of MFSE(Multi-Keywords Fuzzy Searchable Encryption)which is an existing multi-keyword ciphertext fuzzy search scheme.Finally,on the basis of the MFRESE scheme,by transforming the vectorized object from "single vocabulary" to "document index",the advantages of VSM model are fully exploited,which greatly improves the computational efficiency of the scheme.The scheme is named MFRESE(Multi-Keywords Fuzzy Ranked Efficient Searchable Encryption).Compared with MFRSE and MFSE schemes,the high accuracy and high performance of MFRESE schemes are proved.