The Construction And Optimization Of The Index For Ciphertext Retrieval In Cloud Computing Environment

In recent years,cloud computing has made great development in research and application,because it provides economic and convenient service.Recently,more and more companies and users plan to outsource their data to the public cloud.However,the data in the cloud may be maliciously used by the cloud service provider and malicious attackers,since the data owner is no longer control of the data in direct.Therefore,in order to guaranty the privacy and security of data,it is recommended that data owners encrypt the data before outsourcing data to the cloud.While this approach protects data from unauthorized using by untrusted cloud service providers and external users,it makes also the use of data much more difficult because many of the techniques based on plaintext are no longer suitable for ciphertext.Therefore,it is imperative to explore effective search techniques for encrypted data.It is also an application premise to use the data effectively.Face to the challenge in privacy-preserving of data,this thesis investigates search techniques over encrypted data in cloud enviroment.It analyzes first the existing searchable encryption schemes over encrypted data,and then proposes a multi-keyword ranked search method based on B~+ tree.By using the TF-IDF vector model in the index construction and query trapdoor generation,the retrieval results are ranked in descending order according to the relevance score and the first k results are returned.Moreover,based on the proposed scheme,this thesis presents a multi-keyword retrieval scheme for semantic expansion of search keywords according to the user 's expression of query requirements,which helps users find more relevant query results.It can find the keywords associated with query keywords through semantic analysis,and add these keywords to the final query.In addition,the keyword matching number is added to the search,and the relevance score and the keyword matching number are simultaneously used as the retrieval conditions.It shows that the proposed scheme has a high retrieval efficiency by testing on a real data set,and the retrieval speed could be further improved by using multi-threading.