Research On Searchable Encryption Scheme Based On Untrusted Cloud Environment

With the popularization of Internet big data and cloud computing as a multi-technology integration innovation,enterprises and individuals are increasingly willing to apply various convenient and efficient cloud services for massive data storage and sharing.But this also brings about the problem of privacy leakage.It is the simplest and most effective way for users to encrypt local plaintext information before uploading data.Searchable encryption technology(SE)can solve the problem that plaintext retrieval technology cannot directly retrieve the ciphertext stored in the cloud.However,when the traditional searchable encryption sorting scheme uses keyword frequency weights for retrieval and sorting,it only returns all the retrieved ciphertext information in a singular manner without considering the user's retrieval habits and interest preferences,which not only increases the user's retrieval information but also the redundancy caused a waste of network and local resources.The main work of this paper is as follows:(1)Aiming at the waste of resources caused by the traditional encryption retrieval and sorting methods that all return ciphertext retrieval results,the SE retrieval and sorting scheme based on MapReduce is designed.This solution adopts the document set fragmentation mode,and combines the file management system HDFS and parallel programming MapReduce technology to perform parallel retrieval and sorting operations on Hadoop,which has advantages for processing large-scale document sets.(2)Aiming at the problem that traditional encryption retrieval and ranking methods cannot satisfy personal retrieval preferences,a SE retrieval ranking scheme based on a personalized interest model in a cloud environment is designed.The program establishes an interest model for each user,quantifies the generation of different priorities corresponding to "keyword-document",combines the processing value of the TF-IDF cleaning algorithm and the priority of the interest model keyword,and uses Secure KNN to calculate that meets the user's preference relevance score,realizes the personalized expression of the user's retrieval needs.Experimental simulation shows that the proposed scheme saves the resource cost of processing data,and improves the retrieval efficiency and accuracy of returning k document data with the highest relevance when retrieving ciphertext.