Abstract:

One of the most important tasks for secure information retrieval in the public cloud is semantic searching over encrypted data. In order to make queries and search results flexible, it intends to give retrieval service to any arbitrary terms. Verifiable searching is not supported in the current semantic search schemes because the queries are expanded on plaintext and the exact matching is carried out by the extended semantically words with predefined keywords, which reduces their accuracy. The forecasted results from predefined keywords are also required to verify the search results from the cloud. We suggest a secure, verifiable semantic searching technique in this study. In order to calculate the minimum word transportation cost (MWTC), which measures how similar queries and documents are, we formulate the word transportation (WT) problem. We then propose a secure transformation to convert WT problems into random linear programming (LP) problems in order to obtain the encrypted MWTC. In order to construct a verification mechanism employing the intermediate data generated during the matching process to test the accuracy of the search results, we investigate the duality theorem of LP. Security research shows that our plan can ensure secrecy and verifiability. Results from two datasets used in the experiments reveal that our system is more accurate than other schemes

Description:

public cloud, results verifiable searching, secure semantic searching, word transportation.