Proficient Dual Secure Multi Keyword Search by Top-K Ranking Based on Synonym Index and DNN in Untrusted Cloud

Abstract

Recent developments in cloud services have increased number of data owners to store their encrypted data in the cloud whereas equal or more data users participate to retrieve data. Secure retrieval of relevant data has become a challenging issue. In this paper, a secure ranking based multi-keyword search using semantic index is being developed. Initially, owner builds an index file by semantic representation of keyword. Security key is provided by Trusted Authority (TA) for decrypting the obtained results at the user side. TA manages dual security processes such as managing secret keys and issuing security devices to the data users. User query reaches proxy server, and it checks whether any frequent keyword matches with given query using Boolean Search. If it does not match, query enters into the main server which stores all document and index files to obtain relevant result using Deep Learning Neural Network. In deep learning neural network, the query is processed with vector space model to retrieve the relevant documents. Finally, user decrypts the relevant results obtained from deep neural network. The experimental result shows that our proposed model provides better performance in terms of recall, ranking privacy, precision, and searching time.