Research On Search Method Over Encrypted Cloud Data Based On User Demand

In recent years, cloud computing becomes increasingly popular. In order to reduce the cost on storage and management of data, more and more users and companies are planning to upload their large-scale data onto public clouds. For protecting data privacy and alerting unauthorized access, it is a common practice for data owners to encrypt sensitive information before outsourcing. A popular way to search over encrypted data is searchable encryption, which is no doubt a subject worthy of study. However, most of searchable encryption schemes support only keyword-based exact search and only a few schemes support keyword-based fuzzy search. These schemes commonly ignore users’ different search needs and thereby cannot meet the increasing user search experience. Therefore, starting with user needs, we design the following schemes with privacy-preserving under different search needs and security models. The main work is described as follows:(1) A verifiable keyword-based semantic search over encrypted cloud dataWe notice that cloud service providers may be selfish to save its computation or download bandwidth. That is, the cloud server might conduct only a fraction of search operation or return a part of result honestly. Therefore, we introduce and improve the verifiable symmetric searchable encryption, to support the verification of completeness and correctness of search result. Such technique can effectively curb cloud service providers’ selfish operation. In the meantime, by building a semantic tree in real time based on users’ query, this scheme can realize semantic query, which can address the limitation of keyword exact match and also improve the user search experience. Security analysis and experimental measurement shows that our proposed scheme not only supports keyword-based semantic search over encrypted data, but also provides verifiable searchability with data privacy preserving.(2) A personalized multi-keyword ranked search over encrypted cloud dataSince the retrieval model of "one size fits all" in existing searchable encryption schemes ignores the users’ personalized search needs, we propose a personalized multi-keyword ranked search scheme over encrypted data. By recording and analyzing the user’s search history, with the help of semantic ontology WordNet, we build the user interest model upon keyword-based semantic relationships according to keyword access frequency. Through user interest model, we reformulate the keyword priority to realize the personalized ranking over encrypted data. In the meantime, we introduce MDB tree to generate the index and improve the search efficiency significantly. In additional, rigorous security certification and analysis proves that our proposed scheme is secure. Testing and performance analysis of real data shows that our scheme is practical.