Methods and apparatuses for jointly processing data by two parties for data privacy protection

1 . A computer-implemented method for data privacy protection, comprising:

receiving, by a first party from a second party, an encrypted integrated vector, wherein the first party owns an original matrix formed by privacy data and the second party owns n original vectors formed by privacy data, wherein the encrypted integrated vector is obtained by encrypting an integrated vector based on a public key of the second party and a homomorphic encryption algorithm, the integrated vector is a combination of mapping results of the n original vectors in n computing spaces obtained by mapping the n original vectors to the n computing spaces, and each of the n computing spaces are different and isolated from each other, wherein n is an integer greater than 1, wherein the first party owns the original matrix comprising model parameter data of a data processing model, and wherein the n original vectors comprise feature data of one or more service objects for joint service processing using the data processing model, and wherein the n computing spaces are modulo spaces corresponding to n modulus values that are coprime numbers and wherein the integrated vector is an integrated result obtained by performing an encoding operation CRTEncode according to a Chinese remainder theorem;

determining an encrypted result vector based on performing a row-column homomorphic addition operation on plaintext elements in the original matrix and ciphertext elements in the encrypted integrated vector;

determining a data processing result based on homomorphically superimposing an integrated random vector onto the encrypted result vector; and

sending the data processing result to the second party for the second party to obtain a multiplication calculation result of the original matrix and the n original vectors based on decomposing the data processing result into the n computing spaces to obtain the multiplication calculation result, wherein decomposing comprises decrypting the data processing result and performing a decoding operation CRTDecode according to the Chinese remainder theorem, and wherein the second party uses the multiplication calculation result to process the feature data using the data processing model.

2 . The method according to claim 1 , wherein space sizes of the n computing spaces are determined based on maximum effective bit widths of element products of the original matrix and the n original vectors.

3 . The method according to claim 1 , wherein the determining a data processing result comprises:

generating n random vectors;

using opposite vectors of the n random vectors as n first-party fragments of the multiplication calculation result;

combining mapping results of the n random vectors in the n computing spaces to obtain an integrated random vector; and

determining the data processing result based on the integrated random vector and the encrypted result vector for the second party to obtain n second-party fragments of the multiplication calculation result.

4 . The method according to claim 3 , wherein the first party owns one original matrix, and the n computing spaces are n spaces with different quantities of bits in a current numeral system, and wherein the combining mapping results of the n random vectors in the n computing spaces comprises:

extending elements in an ith random vector in the n random vectors by a quantity of bits corresponding to an ith space to obtain an ith random mapping result; and

determining the integrated random vector based on a sum of random mapping results.

5 . The method according to claim 4 , wherein a difference between quantities of bits of the ith space and an (i−1)th space is determined based on excess mask bit widths predetermined for the n random vectors and a maximum effective bit width of an element product of the original matrix and an (i−1)th original vector.

6 . The method according to claim 1 , wherein the first party owns n original matrices, and wherein the determining an encrypted result vector comprises:

determining an integrated matrix by using a Chinese remainder theorem, wherein the integrated matrix satisfies that a remainder result of performing a modulo operation on an ith modulus value corresponding to an ith space is an ith original matrix; and

obtaining the encrypted result vector based on a homomorphic operation on the integrated matrix and the encrypted integrated vector.

7 . The method according to claim 3 , wherein the first party owns n original matrices, wherein

the determining an encrypted result vector comprises:

determining an integrated matrix by using the Chinese remainder theorem, wherein the integrated matrix satisfies that a remainder result of performing a modulo operation on an ith modulus value corresponding to an ith space is an ith original matrix; and

obtaining the encrypted result vector based on a homomorphic operation on the integrated matrix and the encrypted integrated vector; and wherein

the obtaining an integrated random vector comprises:

determining the integrated random vector by using the Chinese remainder theorem, so that the integrated random vector satisfies that a remainder result of performing a modulo operation on the ith modulus value corresponding to the ith space is an ith random vector.

8 . The method according to claim 7 , wherein a bit width of the ith modulus value is determined based on excess mask bit widths predetermined for the n random vectors and a maximum effective bit width of an element product of the ith original matrix and an ith original vector.

9 . The method according to claim 8 , wherein the determining the data processing result based on the integrated random vector and the encrypted result vector comprises:

generating a target random vector, wherein bit widths of elements in the target random vector are determined based on maximum effective bit widths, excess mask bit widths, and a quantity of original matrices;

multiplying the target random vector by the n modulus values;

superimposing the multiplication calculation result onto the integrated random vector to obtain an extended random vector; and

homomorphically superimposing the extended random vector onto the encrypted result vector to obtain the data processing result.

10 . A non-transitory, computer-readable medium storing one or more instructions executable by at least one processor to perform operations comprising:

receiving, by a first party from a second party, an encrypted integrated vector, wherein the first party owns an original matrix formed by privacy data and the second party owns n original vectors formed by privacy data, wherein the encrypted integrated vector is obtained by encrypting an integrated vector based on a public key of the second party and a homomorphic encryption algorithm, the integrated vector is a combination of mapping results of the n original vectors in n computing spaces obtained by mapping the n original vectors to the n computing spaces, and each of the n computing spaces are different and isolated from each other, wherein n is an integer greater than 1, wherein the first party owns the original matrix comprising model parameter data of a data processing model, and wherein the n original vectors comprise feature data of one or more service objects for joint service processing using the data processing model, wherein the n computing spaces are modulo spaces corresponding to n modulus values that are coprime numbers and wherein the integrated vector is an integrated result obtained by performing an encoding operation CRTEncode according to a Chinese remainder theorem;

determining an encrypted result vector based on performing a row-column homomorphic addition operation on plaintext elements in the original matrix and ciphertext elements in the encrypted integrated vector;

determining a data processing result based on homomorphically superimposing an integrated random vector onto the encrypted result vector; and

sending the data processing result to the second party for the second party to obtain a multiplication calculation result of the original matrix and the n original vectors based on decomposing the data processing result into the n computing spaces to obtain the multiplication calculation result, wherein decomposing comprises decrypting the data processing result and performing a decoding operation CRTDecode according to the Chinese remainder theorem, and wherein the second party uses the multiplication calculation result to process the feature data using the data processing model.

11 . The non-transitory, computer-readable medium according to claim 10 , wherein space sizes of the n computing spaces are determined based on maximum effective bit widths of element products of the original matrix and the n original vectors.

12 . The non-transitory, computer-readable medium according to claim 10 , wherein the determining a data processing result comprises:

generating n random vectors;

using opposite vectors of the n random vectors as n first-party fragments of the multiplication calculation result;

combining mapping results of the n random vectors in the n computing spaces to obtain an integrated random vector; and

determining the data processing result based on the integrated random vector and the encrypted result vector for the second party to obtain n second-party fragments of the multiplication calculation result.

13 . The non-transitory, computer-readable medium according to claim 12 , wherein the first party owns one original matrix, and the n computing spaces are n spaces with different quantities of bits in a current numeral system, and wherein the combining mapping results of the n random vectors in the n computing spaces comprises:

extending elements in an ith random vector in the n random vectors by a quantity of bits corresponding to an ith space to obtain an ith random mapping result; and

determining the integrated random vector based on a sum of random mapping results.

14 . The non-transitory, computer-readable medium according to claim 13 , wherein a difference between quantities of bits of the ith space and an (i−1)th space is determined based on excess mask bit widths predetermined for the n random vectors and a maximum effective bit width of an element product of the original matrix and an (i−1)th original vector.

15 . The non-transitory, computer-readable medium according to claim 10 , wherein the first party owns n original matrices, wherein the determining an encrypted result vector comprises:

determining an integrated matrix by using a Chinese remainder theorem, wherein the integrated matrix satisfies that a remainder result of performing a modulo operation on an ith modulus value corresponding to an ith space is an ith original matrix; and

obtaining the encrypted result vector based on a homomorphic operation on the integrated matrix and the encrypted integrated vector.

16 . The non-transitory, computer-readable medium according to claim 12 , wherein the first party owns n original matrices,

the determining an encrypted result vector comprises:

determining an integrated matrix by using the Chinese remainder theorem, wherein the integrated matrix satisfies that a remainder result of performing a modulo operation on an ith modulus value corresponding to an ith space is an ith original matrix; and

obtaining the encrypted result vector based on a homomorphic operation on the integrated matrix and the encrypted integrated vector; and wherein

the obtaining an integrated random vector comprises:

determining the integrated random vector by using the Chinese remainder theorem, so that the integrated random vector satisfies that a remainder result of performing a modulo operation on the ith modulus value corresponding to the ith space is an ith random vector.

17 . The non-transitory, computer-readable medium according to claim 16 , wherein a bit width of the ith modulus value is determined based on excess mask bit widths predetermined for the n random vectors and a maximum effective bit width of an element product of the ith original matrix and an ith original vector.

18 . The non-transitory, computer-readable medium according to claim 17 , wherein the determining the data processing result based on the integrated random vector and the encrypted result vector comprises:

generating a target random vector, wherein bit widths of elements in the target random vector are determined based on maximum effective bit widths, excess mask bit widths, and a quantity of original matrices;

multiplying the target random vector by the n modulus values;

superimposing the multiplication calculation result onto the integrated random vector to obtain an extended random vector; and

homomorphically superimposing the extended random vector onto the encrypted result vector to obtain the data processing result.

19 . A computing device comprising:

at least one processor; and

one or more memories storing programming instructions for execution by the at least one processor to perform one or more operations comprising:

receiving, by a first party from a second party, an encrypted integrated vector, wherein the first party owns an original matrix formed by privacy data and the second party owns n original vectors formed by privacy data, wherein the encrypted integrated vector is obtained by encrypting an integrated vector based on a public key of the second party and a homomorphic encryption algorithm, the integrated vector is a combination of mapping results of the n original vectors in n computing spaces obtained by mapping the n original vectors to the n computing spaces, and each of the n computing spaces are different and isolated from each other, wherein n is an integer greater than 1, wherein the first party owns the original matrix comprising model parameter data of a data processing model, and wherein the n original vectors comprise feature data of one or more service objects for joint service processing using the data processing model, and wherein the n computing spaces are modulo spaces corresponding to n modulus values that are coprime numbers and wherein the integrated vector is an integrated result obtained by performing an encoding operation CRTEncode according to a Chinese remainder theorem;

determining an encrypted result vector based on performing a row-column homomorphic addition operation on plaintext elements in the original matrix and ciphertext elements in the encrypted integrated vector;

determining a data processing result based on homomorphically superimposing an integrated random vector onto the encrypted result vector; and

sending the data processing result to the second party for the second party to obtain a multiplication calculation result of the original matrix and the n original vectors based on decomposing the data processing result into the n computing spaces to obtain the multiplication calculation result, wherein decomposing comprises decrypting the data processing result and performing a decoding operation CRTDecode according to the Chinese remainder theorem, and wherein the second party uses the multiplication calculation result to process the feature data using the data processing model.

20 . The computing device according to claim 19 , wherein space sizes of the n computing spaces are determined based on maximum effective bit widths of element products of the original matrix and the n original vectors.