Efficient privacy preserving ranked search over encrypted data

Cloud computing and its ever so increasing prominence has rendered it as an unavoidable component for data storage and other data services. The security challenges of storing sensitive data on the cloud is reduced to an extent by the Encryption of data, though in the process of encrypted data search, efficiency is compromised. The encrypted data on the cloud can be retrieved using Searchable Symmetric Encryption (SSE). The current work uses multi-keyword searchable encryption scheme with top-k retrieval to avoid compromises on data privacy occurred by using Order Preserving Encryption schemes. The encryption scheme uses homomorphic encryption and vector space model. The vector space model provides the required search accuracy. The homomorphic encryption allows majority of the computation to be done at the server side while concealing the sensitive data. The user alone can identify the final result of the relevance calculation and request for the actual file. In this paper, phrase searching is included to improve the search results on the encrypted data. To accomplish this we maintain a list of the keyword locations in the encrypted file index. The cloud server, which we assume to be honest-but-curious, operates on these encrypted values and identifies if the words occur in close proximity without knowing the actual locations of these words and the words itself.