FBPB-SE: A Forward and Backward Private Blockchain-based Searchable Encryption with Fuzzy Multi-keyword

Searching directly on encrypted data is a fundamental problem with many real-world applications, including cloud storage retrieval and image recognition. Most traditional searchable encryption schemes assume that cloud servers are honest and curious. However, in real applications, cloud servers can also be malicious. We establish a trust system through blockchain technology to enable fair transactions between several untrusted entities. In this paper, we design a reliable smart contract to perform ciphertext retrieval instead of a central cloud server to ensure the correctness of retrieval results and fairness of transactions. We further utilize Word2Vec to train the semantic model to transfer each keyword to the corresponding vector and map synonyms to the approximate vector to support synonym search. We propose a blockchain-based forward and backward private fuzzy multi-keyword searchable encryption (FBPB-SE). In terms of security, we show that FBBP-SE can provide both forward and backward security proofs. We experimentally show the feasibility and efficiency of the scheme.