| Cloud computing is being used by more and more users(companies and individuals)because of its powerful computing capabilities.Cloud storage has also emerged.By uploading local data to the cloud,users don't have to worry about local consumption such as the infrastructure needed to store large amounts of data.However,when the data is outsourced to the cloud server,the data owner and the cloud can perform related operations on the data,which is prone to security problems such as data embezzlement by internal personnel.At the same time,third-party attacks cause data to be hidden and even maliciously falsified,making data security and user privacy unsafe.In order to deal with these security risks,it is often necessary to securely encrypt the outsourced data.But the encryption scheme has two seemingly contradictory goals.First,encryption should have to satisfy its powerful security algorithms to keep it hidden from the cloud server.Second,the solution needs to meet the user's daily retrieval needs when the computational cost and storage cost are acceptable.These two contradictory goals have led to the failure of traditional plaintext retrieval techniques.Therefore,the encryption retrieval scheme came into being.By researching encryption and retrieval,it is possible to achieve secure use of data while achieving efficient data storage.There are many existing keyword-based ciphertext retrieval schemes.These schemes can support different search needs,including single keyword and multi-keyword search,sort search,fuzzy search,semantic search and so on.These groundbreaking work makes the research of encrypted cloud data retrieval more and more intensive,but the existing scheme mainly achieves the effect of user privacy protection by adding random numbers to realize the query unrelated.The existence of these random words,while preventing the statistical analysis of the cloud server,also reduces the accuracy of the retrieval.On the other hand,in the existing keyword-based retrieval scheme,the keywords retrieved are mainly generated based on a dictionary.The length of the dictionary directly affects the dimension of the vector.When the data set is large,the matching calculation time is also longer.At the same time,the 0-1 vector generated from the dictionary does not reflect the semantic information very well.Based on the above two points,this paper designs an outsourced data security retrieval scheme based on confusing keywords and an outsourced data security retrieval scheme based on Word2 vec.The main work of the two schemes are as follows:(1)An outsourcing data security search scheme based on confusing keywords is proposed.In this scheme,the selection and embedding of the confusing keywords can be selectively implemented based on the user history retrieval data.Different from the random value added at the end of the vector generated in other schemes,this scheme designs a selection algorithm for confusing keywords,and embeds it before generating the vector,thus realizing the "random" effect and achieving the purpose of protecting user privacy.In order to eliminate the irrelevant documents caused by confusing keyword matching,an improved secondary search scheme based on confusing keywords is proposed.By constructing two different trapdoors,two screenings are performed to ensure the accuracy of the data.Experiments demonstrate the effectiveness of these two schemes.(2)An outsourcing data security search scheme based on Word2 vec is proposed.In this scheme,we combine machine learning to improve the original scheme.Using the Word2 vec training method to build a semantic model,we will generate low-dimensional vectors from the keywords extracted from the data document,rather than the traditional 0-1 long vectors generated from the length of the dictionary.After that,we outsource the encrypted data to the public cloud.The cloud server will return documents in the order of similarity scores based on the search request.Experiments show that it is more efficient based on real data. |
PDF Full Text Request