Research On Semantic Search Scheme Based On Conceptual Graph Over Encrypted Cloud Data

Nowadays, more and more data owners choose to outsource their data to cloud which can help them enjoy the on-demand high-quality applications and service and reduce data management cost and storage facility spending. How to store and use data securely puzzles data owners a lot. The solution to solve the security problem is data encryption. However, these complex techniques hold back the efficient utilization of data.Existing schemes almost choose keyword as document feature which can help construct indexes quickly when meets the requirement of data owners. Although keyword search can be adapted to encrypted forms, keywords carry quite a little semantic information which affect the precision of retrieval. In addition, almost few schemes can support completely semantic search in the encrypted form currently. To achieve the goal of realize a semantic scheme, in this paper, we study semantic search based on conceptual graphs over encrypted cloud data which contains two parts:First, enabling semantic search based on conceptual graphs over encrypted outsourced data was proposed. In this scheme, we extract conceptual graphs from documents. Because the problem of graph match is a NP hard problem, we choose to transfer conceptual graphs into vectors for realizing graph match in the encrypted form. Meanwhile, we ranked the final results for data user to provide better user experience. The security analysis of the scheme proves the scheme is secure and reliable. The study performs performance test on real data set which proves the scheme is effective.Second, enabling semantic search based on conceptual graphs over encrypted outsourced data efficiently was proposed. Considering the efficiency and security of the above scheme, compared with keyword search, it's still a little worse than keyword. So in this paper, we propose an improved scheme based on the above scheme. We still employ conceptual graphs as knowledge representation. The scheme constructs indexes in different ways according to different scenes to realize high precision. Enhanced schemes are proposed against different threat models. The security analysis proves our scheme is secure and reliable. Performance test on a real data and experiment results show the scheme has the same efficiency as keyword search but high precision.