Classificatory Group Index Method For Ranked Search Of Encrypted Data

Today,with the rapid development of technology,more and more people choose to store their private data in the cloud,so that they can make full use of some high-quality cloud storage services.However,on the one hand,storing data directly in the cloud often causes privacy information to leak.Therefore,in order to protect the security of private information,the data needs to be encrypted before storage.On the other hand,owing to the enormous growth in data storage capacity arising from individual users and business organizations,existing encrypted search methods not only have a high encryption time complexity,but also the cloud server performs search operations with slow speed and efficiency.To solve this problem,we first propose a classificatory group index method(CGIM)for multikeyword ranked search of encrypted cloud data.Different from existing search methods,CGIM first classifies the documents,and then extracts keywords from each document to construct category keyword sets and a keyword set.Finally,the group vector is created by calculating the score of each keyword of category keyword set in the document,which implements block encryption.When encrypted,each group vector in the index corresponds to a block key.This can transform original high-dimensional secret key into several low-dimensional keys to accelerate the process of encrypting indexes.The creation of group vectors can also improve the flexibility of updating documents.When updating,we only need to update these group vectors corresponding to the changed category keyword sets,thereby saving the update time.In the search process of CGIM,we introduce a "targeted search" method according to the category characteristic of each group vector in the index.With this method,instead of calculating the whole products,the cloud server only needs to calculate some inner products of group vectors corresponding to query keywords in the trapdoor and each index to improve the search speed and efficiency.Secondly,in order to improve the performance of CGIM,we further proposes a future matching ranked search for classificatory group index method(FM_CGIM).FM_CGIM uses feature score algorithm to create indexes,so that multi-keywords extracted from a document are as a feature to be mapped to one dimension of the index.Thus the storage cost of indexes can be reduced and the efficiency of encryption can be improved.Moreover,FM_CGIM uses a matching score algorithm in generating trapdoor process,With the help of feature score algorithm,the matching score algorithm can rank the search results according to the type of match and the number of matching keywords,and therefore it is able to return results with higher ranking accuracy.At the same time,this paper gives privacy analysis,complexity analysis and experimental verification for the classificatory group index method.Comprehensive privacy analysis shows that our method is secure.As can be seen from complexity analysis and experimental results,our method can accelerate the encryption and search speed.After further introducing feature matching search,the method not only saves storage costs,but also significantly improves ranking accuracy.