Encrypted search

1. A method, comprising:

tokenizing a document;

generating an index key;

sharding the index key;

encrypting the index key shards with public keys of a plurality of devices;

distributing the encrypted shards;

constructing a secure index using the tokenized document and the key, the constructing comprising generating a trapdoor by hashing a token into an elliptic curve group and performing a point multiplication of the group by an integer representation of the index key; and

storing the secure index in a memory device.

2. The method of claim 1 , wherein the constructing further comprises inserting the trapdoor into a bloom filter.

3. The method of claim 2 , further comprising finalizing the bloom filter with a cryptographic blinding stage.

4. The method of claim 2 , wherein the bloom filter comprises hash functions that are pseudorandom functions with keys derived from index key.

5. The method of claim 1 , wherein the sharding includes secret sharing.

6. A method, comprising:

tokenizing, at a first device, a search query;

creating search requests and send to delegate devices, each search request including a public key encrypted message containing the tokenized search query and index identifiers of indices to be searched;

computing search responses to the search requests, each search response comprising a partial trapdoor computed per token per identifier;

transmitting the search responses to the first device;

recombining, at the first device, the search responses per identifier per token;

performing a ranked set of queries against the indices; and

returning the search results in order of relevancy.

7. The method of claim 6 , wherein the partial trapdoor is computed by hashing each token into the elliptic curve group and performing a point multiplication by the data of a share of an index key.

8. The method of claim 6 , wherein the indices are created by:

tokenizing a document;

generating the index key;

sharding the index key;

encrypting the index key shards with public keys of the delegate devices;

distributing the encrypted shards;

constructing an index using the tokenized document and the key, the constructing comprising generating a trapdoor by hashing a token into an elliptic curve group and performing a point multiplication of the group by an integer representation of the index key; and

storing the index in a memory device.

9. The method of claim 8 , wherein the constructing further comprises inserting the trapdoor into a bloom filter.

10. The method of claim 9 , further comprising finalizing the bloom filter with a cryptographic blinding stage.

11. The method of claim 9 , wherein the bloom filter comprises hash functions that are pseudorandom functions with keys derived from index key.

12. The method of claim 8 , wherein the sharding includes secret sharing.

13. The method of claim 6 , further comprising receiving a user selection of a document from the returned search results and decrypting the user-selected document.

14. A computing device comprising:

a processor; and

one or more memories that include instructions that, when executed by the processor, cause the computing device to:

tokenize, with the processor, a search query;

create search requests and send to delegate devices, each search request including a public key encrypted message containing the tokenized search query and index identifiers of indices to be searched;

cause the delegate devices to compute search responses to the search requests, each search response comprising a partial trapdoor computed per token per identifier;

cause the delegate devices to transmit the search responses to the processor;

recombine, with the processor, the search responses per identifier per token;

performing, with the processor, a ranked set of queries against the indices; and

returning, with the processor, the search results in order of relevancy.

15. The computing device of claim 14 , wherein the partial trapdoor is computed by hashing each token into the elliptic curve group and performing a point multiplication by the data of a share of an index key.

16. The computing device of claim 14 , wherein the indices are created by:

tokenizing a document;

generating the index key;

sharding the index key;

encrypting the index key shards with public keys of the delegate devices;

distributing the encrypted shards;

constructing an index using the tokenized document and the key, the constructing comprising generating a trapdoor by hashing a token into an elliptic curve group and performing a point multiplication of the group by an integer representation of the index key; and

storing the index in a memory device.

17. The computing device of claim 16 , wherein the constructing further comprises inserting the trapdoor into a bloom filter.

18. The computing device of claim 17 , wherein the instructions further cause a computer to finalize the bloom filter with a cryptographic blinding stage.

19. The computing device of claim 17 , wherein the bloom filter comprises hash functions that are pseudorandom functions with keys derived from the index key.

20. The computing device of claim 16 , wherein the sharding includes secret sharing.

21. The computing device of claim 14 , further comprising receiving a user selection of a document from the returned search results and decrypting the user-selected document.