The College of Education for Pure Sciences, Department of Computers, discussed a master's thesis on analogue research on encrypted data
The message presented by the researcher (Hussein Mazen Muhammad) included searchable encryption as an information technology that enables customers to upload their sensitive data in encrypted form to semi-reliable servers to enjoy unlimited storage services and enormous computing capabilities. This method allows a search to be performed on the data while it is encrypted.
Most current searchable coding approaches support a single word search, making them impossible in real applications.
In this work, a secure approach was designed and applied to search for similar encrypted files based on the presentation of several search words. The proposed approach uses the neighbor-sensitive sculpting functions and the Bloom filter to describe each file as a single vector. The proposed approach is able to deal with typing errors and is equipped with an improved security definition that hides access and search patterns, but at an additional cost of delay. For the purpose of considering similar search, we employed an efficient method of word representation. Such a method relies on the bi-gram representation to increase the accuracy of the search. In order not to associate search queries with search results, the proposed approach uses two non-complicit servers. Practical experiments on real data showed the practical ability of the proposed approach. A rigorous security analysis is presented to demonstrate the security of the proposed approach.
