Searchable Public Key Encryption Supporting Semantic Multi-Keywords Search

Searchable public key encryption (SPE) that supports multi-keywords search, allows data users to retrieve encrypted files of interest efficiently, and thus it has been intensively studied during recent years. However, most existing SPE solutions focus on the exact keyword matching, which fails to capture the semantic information of documents. In this paper, we develop a novel SPE scheme supporting semantic multi-keywords search over the encrypted data. Our solution is mainly built on two techniques: one is a shallow neural network model called "word2vec" for capturing the semantic keywords from documents; the other is a keywords conversion method which can convert keywords into a set of vectors. We then utilize an efficient inner product encryption scheme to encrypt these converted vectors and develop the target SPE scheme, which is proven to be secure against chosen keywords attacks. Moreover, we also present both theoretical and experimental analysis to verify the efficiency and accuracy of this scheme. The experiments over a real-world dataset demonstrate that our scheme can obtain a practical performance in terms of time and space complexities. To the best of our knowledge, it is the first time to construct semantic keywords search scheme over encrypted data in the public key setting.