Ranked Keyword Search over Encrypted Cloud Data Through Machine Learning Method

Yinbin Miao,Wei Zheng,Xiaohua Jia,Ximeng Liu,Kim-Kwang Raymond Choo,Robert H. Deng,Robert Deng
DOI: https://doi.org/10.1109/tsc.2021.3140098
IF: 11.019
2022-01-01
IEEE Transactions on Services Computing
Abstract:Ranked keyword search over encrypted data has been extensively studied in cloud computing as it enables data users to find the most relevant results quickly. However, existing ranked multi-keyword search solutions cannot achieve efficient ciphertext search and dynamic updates with forward security simultaneously. To solve the above problems, we first present a basic Machine Learning-based Ranked Keyword Search (ML-RKS) scheme in the static setting by using the k-means clustering algorithm and a balanced binary tree. ML-RKS reduces the search complexity without sacrificing the search accuracy, but is still vulnerable to forward security threats when applied in the dynamic setting. Then, we propose an Enhanced ML-RKS (called ML-RKS$^{+}$+) scheme by introducing a permutation matrix. ML-RKS$^{+}$+ prevents cloud servers from making search queries over newly added files via previous tokens, thereby achieving forward security. The security analysis proves that our schemes protect the privacy of indexes, query tokens and keywords. Empirical experiments using the real-world dataset demonstrate that our schemes are efficient and feasible in practical applications.
What problem does this paper attempt to address?