| Cloud storage is a widely used service for both a personal and enterprise demands. However, despite its advantages, many potential users with sensitive data refrain from fully using the service due to valid concerns about data privacy. An established solution to this problem is to perform encryption on the client's end. This approach, however, restricts data processing capabilities (e.g. searching over the data). In particular, searching semantically with real-time response is of interest to users with big data. To address this, this thesis introduces an architecture for semantically searching encrypted data using cloud services. It presents a method that accomplishes this by extracting and encrypting key phrases from uploaded documents and comparing them to queries that have been expanded with semantic information and then encrypted. It presents an additional method that builds off of this and uses topic-based clustering to prune the amount of searched data and improve performance times for big-data-scale. Results of experiments carried out on real datasets with fully implemented prototypes show that results are accurate and searching is efficient. |
