Research On Verifiable Ciphertext Retrieval Technology Based On Semantic Extension

Massive data outsourcing can cause serious conflicts between data security and data utility.On the one hand,private data stored in plaintext on cloud servers is vulnerable to attacks and leaks,so it needs to be encrypted and stored in ciphertext form for protection.On the other hand,ciphertext generated by a cryptographic system,consisting of characters or bitsets,is difficult to comprehend,which may affect data circulation and use.Therefore,how to accurately obtain target information from the massive data outsourced in ciphertext form has become a current research hotspot and difficulty.Searchable encryption is an effective way to achieve the retrieval of target information through ciphertext search.However,most current searchable encryption schemes only consider word frequency factors and cannot fully reflect user search intent.Furthermore,most existing schemes lack comprehensive verification mechanisms in terms of result trustworthiness.Based on this,this paper investigates the research of verifiable ciphertext retrieval technology based on semantic extension,with the following main contributions and innovations:(1)Propose a semantic extension-based ciphertext retrieval method.Based on the Sentence-BERT model,this method extracts semantic features from the document set through pre-training models and generates feature vectors of documents to improve retrieval accuracy and reduce space consumption;fast clustering algorithms are used to effectively improve retrieval speed.This method also supports dynamic updates of ciphertext.(2)Propose a semantic extension-based verifiable ciphertext retrieval method with edge collaboration.Under the semi-honest-but-curious model,this method utilizes the Merkle-B+tree and hash algorithm to verify the correctness and freshness of the retrieval result,and completes integrity verification by combining Pailliar homomorphic encryption.(3)Build a prototype system for verifiable ciphertext retrieval based on semantic extension.The system realizes semantic extension-based ciphertext retrieval and realizes credible verification of results.Experiments show that under the WikiQA dataset environment:compared with traditional bag-of-words models,semantic accuracy is increased by an average of 82.08%,while space overhead is reduced by an average of 99.36%.The system proves the reliability and feasibility of the method.