A Verifiable and Efficient Multi-Keyword Fuzzy Rank Search Scheme Over Encrypted Data With Privacy-Preserving

Abstract

Searchable Encryption (SE) enables searching over encrypted data. Exact keyword search is supported in most SE schemes, which achieve higher search accuracy but suffer from lower completeness due to the inability to handle similar expressions. To realize fuzzy keyword search, some schemes employ Bloom Filters (BFs), but these may incur high false positive rates and risk exposing the Bloom Filter's internal values to cloud servers (CS). Besides, most existing schemes ignore the fact that CS may engage in malicious behaviors (e.g., undercounting parameters or forging results). To address these issues, we propose an efficient and verifiable ranked fuzzy multi-keyword search scheme based on BFs. We propose a Twin Bloom Filter (TBF) to conceal insertion positions and introduce random numbers to obfuscate uninserted bits. Search results are ranked using Term Frequency-Inverse Document Frequency (TF-IDF) scores to improve relevance. To ensure correctness and integrity, we employ Real Homomorphic Message Authentication Codes (RealHomMAC) and a random challenge technique, respectively. Security analysis proves that our scheme remains secure under both the known-ciphertext model and the known-background model. Theoretical and experimental performance analysis confirms that our scheme achieves efficient and accurate keyword search.