Research On Semantic Search Scheme Based On Content-aware Over Encrypted Cloud Data

Nowadays,with the continuous generation of massive data,the pressure on user's local storage is increasing.Using cloud servers to store user data greatly reduces the local storage pressure and operating costs of users.Cloud storage makes data access smooth,efficient,and low-cost,which meeting the user's requirements for data storage.However,there are also many problems of cloud storage need to be solved,and the security issue is the top priority of these problems.Based on the above problem,for protecting the security of the user's personal data and private information,the most widely used solution is encrypting the user data locally,and then uploading the encrypted data.However,this solution brings up a new problem,that is,the efficient search scheme in the plaintext environment cannot be used in the encrypted environment.In addition,traditional encryption techniques are not sufficient to support today's huge data environment.Therefore,under the premise of ensuring data security,how to implement high-performance search effectively is the main problem.There are already many existing solutions at home and abroad,and most of these solutions are used keywords as a document index construction feature.Existing solutions have been able to implement multiple functions such as sorting search and similarity search.At present,there are few multi-keyword search schemes in fuzzy search.In addition,for the search keywords which users input,the existing solution assigned the same weight to each word.This neglects the semantic relationship between keywords,which leads to the lack of semantics and reduced search accuracy.Therefore,based on the above two problems,while ensuring the privacy of data,how to construct the semantics encrypted search is a big challenge now.In order to achieve this aim,this thesis proposes two ciphertext search schemes based on semantic perceptual and central word expansion.The main work of this thesis includes:(1)Semantic-Aware Searching Over Encrypted Data for Cloud ComputingAn efficient searchable encryption scheme based on concept hierarchy under dual-servers is proposed.The concept hierarchy tree is built on knowledge of outsourced data sets.In order to contain more semantic information,we have further expanded the concept hierarchy.With the extended concept hierarchy,document features can be extracted more accurately,and the user's search terms can be effectively semantically extended.For each file in the file set,two vectors are generated when constructing index for semantic concept matching and similarity calculation.Similarly,the trapdoor of search request also includes two vectors.The role of the dual server is to perform a step-by-step search,one for the similarity calculation of the index vector and the trapdoor index,and the other one for ranking the results and returning them to the users.Such a split makes the search more efficient.For two different threat models,this chapter builds two complete search schemes,and both of them are proved in real-world environments.(2)Privacy-preserving Content-aware Search based on Two-level IndexThe central keyword extension semantic search based on two-level index is proposed.Unlike existing multi-keyword search schemes,this chapter constructs a novel two-level index structure.The two-level index structure combines the forward index and reverse index structure.Compared with the general forward index,this scheme improves the search efficiency better.The proposed two-level index scheme uses the central keyword of the query to search,so the complexity of the search only depends on the size of the file set which associated with the keyword.Since the accuracy of the search results mainly depends on the central keywords of the query,in order to further improve the semantic and accuracy of the search results,this thesis expands the central keywords of the query.Our work first gives the basic framework of the central keyword extension semantic search based on two-level index.Then,according to the privacy protection requirements under different threat models,two specific ciphertext search schemes are given,and the scheme is proved in the real environment.