Contextual Oblivious Similarity Searching for Encrypted Data on Cloud Storage Services

Abstract

With the development of collaborative cloud storage services, files have been typically stored and secured through encryption making them hard to retrieve and search. Search over encrypted cloud approaches have consequently been utilizing cryptographic and indexing procedures. The vast majority use exact matching to fulfill their search criteria, which is then expanded by incorporating similarity ranking algorithms. However, this complex expansion does not always succeed due to its dependence on third parties to evaluate the search and the possible compromise on the privacy of the stored information. It also requires significant computational resources. This work demonstrates novel approach to similarity search, known as Contextual Oblivious Similarity based Search (COS2). In the proposed system, authorized users can categories searches resilient to typing errors. COS2 also introduces browsing caches to improve subscriber experience. Dual encryption mechanisms improve the relevance in searches without revealing confidential data on untrusted cloud service providers. Finally, this contextual search thrives to reduce the computational overhead of the overall search procedure, leading to a 86% improvement in terms of search efficiency.