Secure search method and secure search device
AssigneeHitachi, Ltd.
InventorsYasuhiro Fujii, Susumu Serita
Life of this patent
Estimating the term…
Abstract
In search that uses searchable code, the search query and the secure index is collated in round-robins fashion, registering the required secure index and a characteristic quantity of deposited data in a database server to realize the searchable code. The server uses the characteristic quantity to perform clustering on the secure index. For search, collation is first performed only for representative data of a cluster. For a hit, the collation priority is raised for all the data included in the cluster. When there is no hit, the priority is lowered. After calculating the priority, collation is performed sequentially for all the data on the basis of the priority.
Claims7
A secure search device that receives data from a registered client and receives information for a search from a search client, the secure search device comprising:
- a receiver that receives, from the registered client, a set of encrypted data acquired by encrypting the data, a secure index acquired by securing an index extracted from the data and a characteristic quantity for calculating similarity between data pieces;
- a similarity calculator that calculates the similarity of two data pieces based upon the characteristic quantity received from the registered client;
- a clustering unit that clusters the encrypted data received from the registered client based upon the similarity calculated by the similarity calculator;
- a priority calculator that receives a trapdoor acquired by securing a search keyword included in a search query for searching data registered in the secure search device from the search client and calculates the priority of collation of the clustered encrypted data and the trapdoor based upon a result of clustering generated by the clustering unit;
- a collator that collates the secure index received from the registered client and the trapdoor;
- and
- a search unit that collates the encrypted data and the trapdoor by the collator based upon the priority calculated by the priority calculator in the order of clusters having higher priority by a predetermined frequency when the trapdoor is received from the search client, and returns the encrypted data that hits the trapdoor to the search client.
Dependencies are inferred from the claim text.