A Privacy-Preserving Framework for Large-Scale Content- Based Information Retrieval

It is very necessary to protect personal confidential data that we share or search through web. Earlier there are number of privacy preserving mechanism has been developed. In this project, we develop a new privacy protection framework for huge- content-based information retrieval. We are contributing protection in two layers. Originally, robust hash values are taken as queries to avoid revealing of unique features or content. Then, the client has to select to skip some of the bits in a hash value for increasing the confusion for the server. Due to the suppressed information, it is computationally difficult for the server to know the client’s concern. The server has to return the hash values of all desirable candidates to the client. The client executes a search within the candidate list to find the best match. Subsequently only hash values are exchanged between the client and the server, the privacy of both parties is protected. We imported the concept of tunable privacy, where the privacy protection level can be adjusted according to a policy. It is concluded through hash-based piecewise inverted indexing. The concept is to divide a feature vector into pieces and index each piece with a sub hash value. Every sub hash value is associated with an inverted index list. The framework has been majorly tested using a large image database. We have calculated both retrieval performance and privacy-preserving performance for a particular content identification application. Both algorithms illustrate satisfactory performance in comparison with state-of-the-art retrieval schemes. The results show that the privacy enhancement somewhat improves the retrieval performance. We consider the majority voting attack for reckoning the query category and identification. Experiment results show that this attack is a threat when there are near-duplicates, but the success rate reduces with the number of omitted bits and the number of distinct items.