TWI835300B - A data matching method, device, equipment and medium - Google Patents

Abstract

The invention discloses a data matching method, device, equipment and medium. In the method, the first data and the second data are respectively logged into the pre-trained vector transformation model to obtain the corresponding first vector and the second vector, and Obtain the distance between the first vector and the second vector encrypted by the first target public key, determine the target distance between the first vector and the second vector based on the encrypted distance and the first target private key, and determine based on the target distance Whether the two data match or not, when the two data are not exactly the same, fuzzy matching of the data can also be achieved, broadening the usage scenarios, and in the fuzzy matching process, the first target public and private key pairs are introduced for homomorphic encryption respectively. and decryption, realizing safe intersection and ensuring the security of the matching process. During the entire matching process, the data has not left the corresponding device in the form of original data, realizing fuzzy matching without leaving the original data. The security of the matching process is guaranteed.

Description

A data matching method, device, equipment and medium

The present invention belongs to the field of data processing technology, and in particular relates to a data matching method, device, equipment and medium.

Current privacy computing technology is mainly used in secure intersection and federated learning. The so-called secure friendship refers to identifying the intersection of data from both parties, for example, identifying the shared users of organization A and organization B. Secure friendship is also the first step in vertical federated learning. That is to say, first identify the mobile phone number and ID card. Key information such as ID number, business license number, etc. are securely exchanged before proceeding to the next step of joint modeling and other steps.

In related technologies, in order to identify the intersection of the two parties' data or achieve the matching of the two parties' data, common secure intersection algorithms include secure intersection algorithms based on the RSA encryption algorithm, and secure intersection algorithms based on the Oblivious Transfer (OT) protocol. Intersection algorithm and so on. However, the current safe intersection algorithm can only match successfully when the data of both parties are exactly the same, that is to say, when the data type of the two parties' data and the number of characters contained in the data are exactly the same. However, in actual business, there are often many usage scenarios that need to be matched when the data are not exactly the same. Therefore, the secure intersection algorithm in the existing technology greatly limits its usage scenarios and affects the business scope of matching.

The present invention provides a data matching method, device, equipment and medium to solve the problem that the secure intersection algorithm in the prior art can only perform secure intersection when the data of both parties are exactly the same, and the usage scenarios are limited, which affects the business scope of data matching. problem.

The present invention provides a data matching method, which is applied to the first device. The method includes: Log the first data to be matched into the pre-trained vector transformation model to obtain the first vector corresponding to the first data; Use the self-generated first target public key to perform homomorphic encryption on the first vector to generate a first encryption vector, and send the first target public key to the second device; Obtain the distance between the encrypted first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, wherein the second encrypted vector is obtained by synchronizing the second vector with the first target public key. obtained after state encryption; the second vector is obtained by logging the second data into the pre-trained vector transformation model in the second device; Based on the encrypted distance between the first vector and the second vector and the first target private key corresponding to the first target public key, determine the target distance between the first vector and the second vector, based on the target distance and a preset first distance threshold to determine whether the first data and the second data match.

The present invention provides a data matching method, which is applied to the second device. The method includes: Log the second data to be matched into the pre-trained vector transformation model to obtain the second vector corresponding to the second data; Receive the first target public key sent by the first device, and use the first target public key to perform homomorphic encryption on the second vector to generate a second encryption vector; Obtain the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, wherein the first encrypted vector is obtained by encrypting the first vector using the first target public key. , the first vector is obtained by logging the first data into the pre-trained vector transformation model in the first device; Based on the target distance and the preset first distance threshold, it is determined whether the first data and the second data match.

The invention also provides a data matching device, which includes: The first acquisition module is used to log the first data to be matched into the pre-trained vector transformation model and obtain the first vector corresponding to the first data; The first processing module is configured to perform homomorphic encryption on the first vector using the first target public key generated by itself to generate a first encryption vector, and send the first target public key to the second device; The first acquisition module is also used to acquire the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, wherein the second encrypted vector is obtained using the first target The second vector is obtained by homomorphically encrypting the second vector with the public key; the second vector is obtained by logging the second data into the pre-trained vector conversion model in the second device; A first determination module configured to determine the first vector and the second vector based on the encrypted distance between the first vector and the second vector and the first target private key corresponding to the first target public key. The target distance, based on the target distance and the preset first distance threshold, determine whether the first data and the second data match.

The invention also provides a data matching device, which includes: The second acquisition module is used to log the second data to be matched into the pre-trained vector transformation model and obtain the second vector corresponding to the second data; The second processing module is configured to receive the first target public key sent by the first device, and use the first target public key to perform homomorphic encryption on the second vector to generate a second encryption vector; The second acquisition module is also used to acquire the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, wherein the first encrypted vector is disclosed using the first target The first vector is obtained by encrypting the first vector with the key, and the first vector is obtained by logging the first data into the pre-trained vector conversion model in the first device; The second determination module is used to determine whether the first data and the second data match based on the target distance and the preset first distance threshold.

The present invention also provides an electronic device. The electronic device includes a processor. The processor is configured to implement the steps of any of the above data matching methods when executing a computer program stored in the memory.

The present invention also provides a computer-readable storage medium that stores a computer program that can be executed by a terminal. When the program is run on the terminal, it causes the terminal to execute any one of the above steps of the data matching method.

The present invention also provides an electronic device. The electronic device includes a processor. The processor is configured to implement the steps of any of the above data matching methods when executing a computer program stored in the memory.

The present invention also provides a computer-readable storage medium that stores a computer program that can be executed by a terminal. When the program is run on the terminal, it causes the terminal to execute any one of the above steps of the data matching method.

In the present invention, the first data to be matched is logged into the pre-trained vector conversion model, the first vector corresponding to the first data is obtained, and the first target public key generated by itself is used to homomorphically encrypt and generate the first vector. the first encrypted vector, and sends the first target public key to the second device to obtain the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, where the second encrypted vector It is obtained by homomorphically encrypting the second vector using the first target public key. The second vector is obtained by logging the second data into the pre-trained vector conversion model of the second device. Based on the encrypted The distance between the first vector and the second vector and the first target private key corresponding to the first target public key determine the target distance between the first vector and the second vector. Based on the target distance and the preset first distance threshold, determine Whether the first data and the second data match. In this embodiment of the present invention, the first data and the second data to be matched are respectively logged into the vector transformation model that has been trained in advance, and the first vector corresponding to the first data and the second vector corresponding to the second data are obtained. , and obtain the first encrypted vector encrypted based on the first vector, and the second encrypted vector encrypted based on the second vector, determine the distance between the encrypted first vector and the second vector, and obtain the encrypted first vector based on the encrypted second vector. The distance between the first vector and the second vector and the first target private key generated by itself determine the target distance between the first vector and the second vector, and determine the first data and the second data based on the target distance and the preset first distance threshold. Whether the data matches, that is, when the first data and the second data are not exactly the same, fuzzy matching between the first data and the second data can be achieved, broadening the usage scenarios, and introducing the first target disclosure during the fuzzy matching process The golden key and the first target private key are homomorphically encrypted and decrypted respectively, achieving safe intersection and ensuring the security of the matching process. During the entire matching process, neither the first data nor the second data are used as original data. The form has left the corresponding first device and second device, realizing fuzzy matching without leaving the original data out of the database, further ensuring the security of the matching process.

In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be described in further detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making any progressive efforts fall within the scope of protection of the present invention.

In order to ensure that matching can be performed even when the data of both parties are not exactly the same, and to broaden the business scope of data matching, embodiments of the present invention provide a data matching method, device, equipment and medium.

In the present invention, the first data to be matched is logged into the pre-trained vector conversion model, the first vector corresponding to the first data is obtained, and the first target public key generated by itself is used to homomorphically encrypt the first vector. Generate a first encrypted vector, send the first target public key to the second device, and obtain the distance between the encrypted first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, where the second encrypted The vector is obtained by homomorphically encrypting the second vector using the first target public key. The second vector is obtained by logging the second data into the pre-trained vector conversion model of the second device. Based on the encrypted The distance between the first vector and the second vector and the first target private key corresponding to the first target public key, determine the target distance between the first vector and the second vector, based on the target distance and the preset first distance threshold, Determine whether the first information and the second information match.

Example 1: Figure 1 is a schematic diagram of a data matching process provided by an embodiment of the present invention. The process includes the following steps: S101: Log the first data to be matched into the pre-trained vector transformation model, and obtain the first vector corresponding to the first data.

The data matching method provided by the embodiment of the present invention is applied to a first device, which may be a smart terminal, a PC, a server, or other devices.

In order to ensure that fuzzy matching can be achieved even when the data of both parties are not exactly the same, in the embodiment of the present invention, a pre-trained vector transformation model is deployed in the first device. The pre-trained vector transformation model is used to obtain the data to be matched. Corresponding vectors, and for different data, the dimensions of the vectors output by the pre-trained vector transformation model are the same.

In order to obtain the first vector corresponding to the first data to be matched, the first data is logged into the pre-trained vector transformation model, and the pre-trained vector transformation model outputs the first vector corresponding to the first data. Each component in the first vector is a number, that is, the first data is quantified through the pre-trained vector transformation model.

S102: Use the self-generated first target public key to perform homomorphic encryption on the first vector to generate a first encryption vector, and send the first target public key to the second device.

In this embodiment of the present invention, in order to improve security, the first device generates a first target public and private key pair, where the first target public and private key pair includes the first target public key and the first target private key. key, and encrypts the first vector according to the self-generated first target public key to generate a first encryption vector. The first target public and private key pair may be a symmetric public and private key pair or an asymmetric public and private key pair. Specifically, the target public and private key pair may be set according to requirements.

The process of generating the first target public-private key pair is an existing technology and will not be described in detail here.

Since the second data to be matched with the first data is obtained by the second device, in order to facilitate subsequent determination of the target distance between the first vector corresponding to the first data and the second vector corresponding to the second data, in this article In the embodiment of the invention, the first device also sends the first target public key to the second device, so that the second device can homomorphically encrypt and generate the second vector corresponding to the second material based on the first target public key. Second encryption vector. Specifically, in the embodiment of the present invention, when performing homomorphic encryption on the first vector and the second vector, each component in the first vector and each component in the second vector are encrypted based on the first target public key. Perform homomorphic encryption on each component of , and then obtain the first encryption vector and the second encryption vector.

S103: Obtain the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, where the second encrypted vector is the second vector using the first target public key. Obtained after homomorphic encryption; the second vector is obtained by logging the second data into the pre-trained vector transformation model in the second device.

In order to achieve fuzzy matching between the first data and the second data, a pre-trained vector conversion model is also deployed in the second device to obtain the second vector corresponding to the second data to be matched. That is to say, the second data is The second data is logged into the pre-trained vector conversion model. The pre-trained vector conversion model outputs the second vector corresponding to the second data. The second device is based on the received first target public key sent by the first device. Perform homomorphic encryption on the second vector to obtain a second encrypted vector.

Since the first vector and the second vector are respectively encrypted using the first target public key generated by the first device, in order to determine the target distance of the first vector and the second vector, in the embodiment of the present invention, A device may first receive the second encryption vector sent by the second device, and decrypt the second encryption vector based on the first target private key in the first target public-private key pair generated by the first device itself to obtain the second encryption vector. two vectors, and based on the first vector and the second vector, determine the target distance of the first vector and the second vector.

In order to improve security, in the embodiment of the present invention, in order to determine the target distance of the first vector and the second vector, the encrypted first vector and the second vector determined based on the first encrypted vector and the second encrypted vector may first be obtained. distance. Among them, the encrypted distance between the first vector and the second vector is not a certain numerical value, but a certain calculation formula that needs to be decrypted to obtain the target distance between the first vector and the second vector, wherein the encrypted distance The distance between the first vector and the second vector may be determined by the first device, or may be determined by the second device and then sent to the first device.

S104: Based on the encrypted distance between the first vector and the second vector and the first target private key corresponding to the first target public key, determine the target distance between the first vector and the second vector. Based on the The target distance and the preset first distance threshold determine whether the first data and the second data match.

In the embodiment of the present invention, in order to determine whether the first data and the second data match, after determining the distance between the encrypted first vector and the second vector, the distance between the encrypted first vector and the second vector is first determined. Decrypt and determine the target distance between the first vector and the second vector. Since both the first encrypted vector and the second encrypted vector are encrypted and generated by the first target public key generated by the first device, in order to determine the target distance between the first vector and the second vector, in the embodiment of the present invention, The distance between the encrypted first vector and the second vector can be decrypted according to the first target private key in the first target private key pair generated by the first device, and the distance between the first vector and the second vector can be determined. target distance.

In order to determine whether the first data and the second data match, in the embodiment of the present invention, the target distances of the first vector and the second vector are compared with a preset first distance threshold, and based on the comparison result, the target distance is determined. Whether the first information and the second information match. Among them, the smaller the target distance is, the better the match between the first data and the second data is.

In this embodiment of the present invention, the first data and the second data to be matched are respectively logged into the vector transformation model that has been trained in advance, and the first vector corresponding to the first data and the second vector corresponding to the second data are obtained. , and obtain the first encrypted vector encrypted based on the first vector, and the second encrypted vector encrypted based on the second vector, determine the distance between the encrypted first vector and the second vector, and obtain the encrypted first vector based on the encrypted second vector. The distance between the first vector and the second vector and the first target private key generated by itself determine the target distance between the first vector and the second vector, and determine the first data and the second data based on the target distance and the preset first distance threshold. Whether the data matches, that is, when the first data and the second data are not exactly the same, fuzzy matching between the first data and the second data can be achieved, broadening the usage scenarios, and introducing the first target disclosure during the fuzzy matching process The golden key and the first target private key are homomorphically encrypted and decrypted respectively, achieving safe intersection and ensuring the security of the matching process. During the entire matching process, neither the first data nor the second data are used as original data. The form has left the corresponding first device and second device, realizing fuzzy matching without leaving the original data out of the database, further ensuring the security of the matching process.

Example 2: In order to determine the first vector corresponding to the first data, based on the above embodiment, in the embodiment of the present invention, the first data to be matched is logged into the pre-trained vector transformation model to obtain the first data The corresponding first vector includes: Determine the first target data type corresponding to the first data to be matched; Determine the pre-completed first target vector transformation model corresponding to the first data according to the corresponding relationship between the first target data type and the pre-saved data type and the pre-trained vector transformation model; Log the first data into the pre-trained first target vector transformation model to obtain the first vector corresponding to the first data.

In this embodiment of the present invention, the first data to be matched may be text data, for example, the first data may be name, gender, address, etc., or it may be digital data, for example, the first data may be ID number, bank card number, admission ticket number, etc. Therefore, for different data types of first data, the pre-trained vector conversion models that obtain the corresponding first vectors are also different.

Specifically, the first device may store the corresponding relationship between the data type and the pre-trained vector conversion model, and use the corresponding pre-trained data type according to the first target data type corresponding to the obtained first data to be matched. The vector transformation model obtains the first vector corresponding to the first data; wherein the corresponding pre-trained vector transformation model is also the pre-trained first target vector transformation model.

In order to accurately determine the vector conversion model for converting the first data into the first vector, based on the above embodiments, in the embodiment of the present invention, if the first target data type is a text type, the corresponding pre-training is completed The first target vector conversion model is a word vector model or a sentence vector model; if the first target data type is a numeric type, the corresponding pre-trained first target vector conversion model is a One-Hot encoding model.

Specifically, if the first data is text data, that is, the first target data type of the first data is text type, then the first target is determined based on the correspondence between the pre-saved data type and the pre-trained vector transformation model. A pre-trained first target vector conversion model corresponding to the data type. The first target vector conversion model is a word vector model or a sentence vector model. Based on the pre-trained word vector model or sentence vector model, the first target vector conversion model is obtained. The first vector corresponding to the data; if the first data is digital data, that is, the first target data type of the first data is a digital type, then based on the correspondence between the pre-saved data type and the pre-trained vector conversion model, determine The pre-trained first target vector conversion model corresponding to the first target data type, the first target vector conversion model is a pre-trained One-Hot encoding model, and based on the pre-trained One-Hot encoding model, Obtain the first vector corresponding to the first data.

Taking the dimension of the vector output by the pre-trained vector conversion model as 5 and the pre-trained vector conversion model as a word vector model as an example, if the first data is text data, and the first data is "Shanghai Pudong New District Sunny Shop", then input "Shanghai Pudong New District Sunny Shop" into the pre-trained word vector model, and the output first vector corresponding to "Shanghai Pudong New District Sunny Shop" is (1.0, 2.0, 1.5, 2.0, 3.5).

If the pre-trained vector conversion model is a One-Hot encoding model, the corresponding one-hot encoding can be set in advance for each number. For example, if the number contains 0-9, then among the numbers 0-9, 0 corresponds to The one-hot encoding is 0000000001, the one-hot encoding corresponding to 1 is 0000000010, the one-hot encoding corresponding to 2 is 0000000100, the one-hot encoding corresponding to 3 is 0000001000, the one-hot encoding corresponding to 4 is 0000010000, and the one-hot encoding corresponding to 5 is 0000100000 , The one-hot encoding corresponding to 6 is 0001000000, the one-hot encoding corresponding to 7 is 0010000000, the one-hot encoding corresponding to 8 is 0100000000, and the one-hot encoding corresponding to 9 is 1000000000. Log the digital data into the One-Hot encoding model, and each first component in the first vector output by the One-Hot encoding model is the one-hot encoding of each corresponding number in the first data.

If the first data is digital data, and the digital data is "12345", then "12345" is input into the pre-trained One-Hot encoding model, and the first vector corresponding to the output "12345" is (0000000010, 0000000100, 0000001000, 0000010000, 0000100000).

Among them, when training the vector conversion model, each data and the corresponding label vector can be pre-labeled, each data and the corresponding label vector are input into the original vector conversion model, and the prediction vector output by the original vector conversion model is and the corresponding annotation vector, adjust the parameters of the original vector transformation model, and when the convergence conditions are met, it is determined that the training of the vector transformation model is completed.

In the embodiment of the present invention, regardless of whether the first data and the second data are digital data or text data, fuzzy matching can be achieved, further broadening the scenario application.

Example 3: In order to determine the distance between the encrypted first vector and the second vector, based on the above embodiments, in the embodiment of the present invention, the encrypted vector determined based on the first encrypted vector and the second encrypted vector is obtained. The distance between the first vector and the second vector includes: Receive the second encryption vector sent by the second device, wherein the second encryption vector is obtained by the second device after homomorphically encrypting the second vector based on the first target public key; Based on the first encrypted vector and the second encrypted vector, the distance between the encrypted first vector and the second vector is determined.

In the embodiment of the present invention, the distance between the encrypted first vector and the second vector obtained by the first device and determined based on the first encrypted vector and the second encrypted vector may be determined by the first device, or may also be determined by the second device. and sent to the first device.

If the distance between the encrypted first vector and the second vector is determined by the first device, in order to determine the distance between the encrypted first vector and the second vector, the first device needs to obtain the second encrypted vector sent by the second device. , specifically in order to obtain the second encryption vector, after the first device sends the first target public key to the second device, the second device receives the first target public key sent by the first device, based on the first target After encrypting the second vector with the public key, the second encrypted vector is obtained and sent to the first device. The second vector is obtained by logging the second data to be matched into the pre-trained vector transformation model in the second device. vector.

The first device receives the second encryption vector sent by the second device, and determines the encrypted first vector and the second vector locally on the first device based on the received second encryption vector and the first encryption vector determined by the first device. distance.

If each component in the first vector and each component in the second vector are homomorphically encrypted based on the first target public key to obtain the first encrypted vector and the second encrypted vector, then in order to determine the encryption The distance between the encrypted first vector and the second vector. In a possible implementation, the first device determines the encrypted first vector and the second encrypted vector based on the first encrypted vector and the second encrypted vector and the Euclidean distance formula. vector distance. specific, based on , determine the distance between the encrypted first vector and the second vector, where, the is the i-th component in the first encrypted vector, is the i-th component in the second encrypted vector, the is the distance between the encrypted first vector and the second vector, N is the number of components included in the first encrypted vector or the second encrypted vector, and the number of components included in the first encrypted vector is the same as the second encrypted component. The number of components contained in is the same, that is, the length of the first encryption vector is equal to the length of the second encryption vector.

In another possible implementation, the first device may also determine the distance between the encrypted first vector and the second vector based on the first encrypted vector and the second encrypted vector and the cosine distance formula or the Hamming distance formula.

It should be noted that since each component of the first encrypted vector and the second encrypted vector is homomorphically encrypted by the first target public key, the determined encrypted first vector and the second vector are Distance is not an actual numerical value, but a deterministic operation.

Since in the embodiment of the present invention, the second data has not left the second device in the form of original data, fuzzy matching can be achieved without the original data leaving the database, further ensuring the security of the matching process.

Example 4: In order to determine the distance between the encrypted first vector and the second vector, based on the above embodiments, in this embodiment of the present invention, sending the first target public key to the second device includes: Send the first encryption vector and the first target public key to the second device; The obtaining the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes: Receive the encrypted distance between the first vector and the second vector sent by the second device and determined based on the first encrypted vector and the second encrypted vector, where the second encrypted vector is determined by the second device based on the The first target public key is obtained by homomorphically encrypting the second vector.

In order to determine the distance between the encrypted first vector and the second vector, in the embodiment of the present invention, the distance between the encrypted first vector and the second vector obtained by the first device can also be determined for the second device and sent to The first device.

Specifically, in order to ensure that the second device can obtain the encrypted distance between the first vector and the second vector, in this embodiment of the present invention, when the first device sends the generated first target public key to the second device, The first encryption vector can also be sent to the second device together. After receiving the first target public key and the first encryption vector sent by the first device, the second device pairs the second vector with the first target public key based on the first target public key. Perform homomorphic encryption to generate a second encrypted vector, and the second device determines the distance between the encrypted first vector and the second vector based on the second encrypted vector and the received first encrypted vector sent by the first device, and adds the encrypted vector to the second encrypted vector. The encrypted distance between the first vector and the second vector is sent to the first device, and the first device obtains the encrypted distance between the first vector and the second vector sent by the second device.

If each component in the first vector and each component in the second vector are homomorphically encrypted based on the first target public key to obtain the first encrypted vector and the second encrypted vector, then in order to determine the encryption The distance between the encrypted first vector and the second vector. In a possible implementation, the second device determines the encrypted first vector and the second vector based on the first encrypted vector and the second encrypted vector and the Euclidean distance formula. distance. specific, based on , determine the distance between the encrypted first vector and the second vector, where, the is the i-th component in the first encrypted vector, is the i-th component in the second encrypted vector, the is the distance between the encrypted first vector and the second vector, N is the number of components included in the first encrypted vector or the second encrypted vector, and the number of components included in the first encrypted vector is the same as the second encrypted component. The number of components contained in is the same, that is, the length of the first encryption vector is equal to the length of the second encryption vector.

In another possible implementation, the second device may also determine the distance between the encrypted first vector and the second vector based on the first encrypted vector and the second encrypted vector and the cosine distance formula or the Hamming distance formula.

It should be noted that since each component of the first encrypted vector and the second encrypted vector is homomorphically encrypted by the first target public key, the determined encrypted first vector and the second vector are Distance is not an actual numerical value, but a deterministic operation.

Since in the embodiment of the present invention, none of the first data has left the first device in the form of original data, fuzzy matching can be achieved without the original data leaving the database, further ensuring the security of the matching process.

Example 5: In order to enable the second device to also determine whether the first data and the second data match, based on the above embodiments, the method further includes: Receive the third encryption vector sent by the second device and the second target public key generated by the second device; wherein the third encryption vector is the second device using the second target public key to encrypt the second vector. Obtained after homomorphic encryption; Perform homomorphic encryption on the first vector based on the second target public key to generate a fourth encrypted vector; Based on the third encrypted vector and the fourth encrypted vector, determine the distance between the encrypted second vector and the first vector, and send the encrypted distance between the second vector and the first vector to the third encrypted vector. Two devices, so that the second device calculates the encrypted second vector based on the distance between the encrypted second vector and the first vector and the second target private key corresponding to the second target public key. Decrypt the distance from the first vector to determine the target distance between the second vector and the first vector, and determine the third vector based on the target distance between the second vector and the first vector and the preset first distance threshold. Whether the first data and the second data match.

In order for both the first device and the second device to determine whether the first data and the second data match, in the embodiment of the present invention, the second device also needs to obtain the target distance between the first vector and the second vector, that is, The first device needs to synchronize data with the second device. The target distance between the first vector and the second vector may be determined by the second device itself, or may be sent to the second device after the first device determines the target distance.

If the target distance between the first vector and the second vector is determined by the second device itself, specifically, the first device can receive the second target public key sent by the second device and the third encryption vector sent by the second device. , wherein the second target public key is generated by the second device itself, and the third encryption vector is obtained by the second device using the second target public key to homomorphically encrypt the second vector. After receiving the second target public key sent by the second device, the first device performs homomorphic encryption on the first vector based on the second target public key to generate a fourth encryption vector. In order to determine the distance between the encrypted first vector and the second vector, in the embodiment of the present invention, the first device determines the encrypted vector based on the third encrypted vector sent by the second device and the fourth encrypted vector generated by itself. The distance between the first vector and the second vector is determined, and the determined distance between the encrypted first vector and the second vector is sent to the second device, so that the second device determines the distance between the encrypted first vector and the second vector based on the received encrypted first vector and the second vector. The distance between the two vectors and the second target private key corresponding to the second target public key generated by the second device itself, decrypt the distance between the encrypted first vector and the second vector, and determine the distance between the second vector and the second target public key. The target distance of the first vector, and based on the target distance of the second vector and the first vector and the preset first distance threshold, determine whether the first data and the second data match.

The following is explained with a specific example: The first device logs the first data to be matched into the pre-trained vector transformation model deployed in the first device, and obtains the first vector corresponding to the first data. The second device logs the second data to be matched. Go to the pre-trained vector conversion model deployed in the second device to obtain the second vector corresponding to the second data. If the first vector corresponding to the first data U1 is (x1, x2, x3..., xm), the second vector corresponding to the second data U2 is (y1, y2, y3..., ym).

The first device generates a first target public-private key pair A(pka1, ska1), where pka1 is the first target public key and ska1 is the first target private key. Based on the first target public key pair A vector is homomorphically encrypted to generate the first encrypted vector. The first encrypted vector corresponding to the first vector (x1, x2, x3..., xm) is ( , , ), and send the first encryption vector and the first target public key to the second device.

After receiving the first target public key and the first encryption vector sent by the first device, the second device performs homomorphic encryption on the second vector based on the first target public key to obtain the second encryption vector. Specifically, the second encrypted vector corresponding to the second vector (y1, y2, y3..., ym) is ( , , ), the second device determines the distance between the encrypted first vector and the second vector based on the second encrypted vector and the received first encrypted vector and sends it to the first device.

The second device generates a second target public-private key pair B(pka2, ska2), where pka2 is the second target public key and ska2 is the second target private key. Based on the second target public key pair Two vectors are homomorphically encrypted to generate a third encrypted vector. The third encrypted vector corresponding to the second vector (y1, y2, y3..., ym) is ( , , ), the second device sends the second target public key and the third encryption vector to the first device.

After receiving the second target public key and the third encryption vector sent by the second device, the first device performs homomorphic encryption on the first vector based on the second target public key to obtain a fourth encryption vector. Specifically, , the fourth encrypted vector corresponding to the first vector (x1, x2, x3..., xm) is ( , , ). The first device determines the encrypted distance between the first vector and the second vector based on the fourth encrypted vector and the third encrypted vector, and sends the encrypted distance between the first vector and the second vector to the second device.

After receiving the distance between the encrypted first vector and the second vector, the first device generates the encrypted first vector and the second vector according to the first target private key corresponding to the first target public key generated by itself. Decrypt the distance to determine the target distance between the first vector and the second vector. The first device determines whether the first data and the second data are based on the target distance between the second vector and the first vector and the preset first distance threshold. match.

After receiving the distance between the encrypted first vector and the second vector, the second device generates the encrypted first vector and the second vector according to the second target private key corresponding to the second target public key generated by itself. Decrypt the distance to determine the target distance between the first vector and the second vector, and the second device determines whether the first data and the second data are based on the target distance between the second vector and the first vector and the preset first distance threshold. match.

Example 6: In order to enable the second device to also determine whether the first data and the second data match, based on the above embodiments, in this embodiment of the present invention, after determining the target distance between the first vector and the second vector, The method also includes: Send the target distance of the first vector and the second vector to the second device, so that the second device determines whether the first data and the second data match based on the target distance and the preset first distance threshold. .

In order to enable the second device to also determine whether the first data and the second data match, in the embodiment of the present invention, the second device also needs to obtain the target distance between the first vector and the second vector. Specifically, the second device obtains The target distance between the first vector and the second vector may be obtained by the first device and sent to the second device. After receiving the target distance of the first vector and the second vector sent by the first device, the second device determines whether the first data and the second data match based on the target distance and the preset first distance threshold.

The following is a specific example to illustrate: The first device logs the first data to be matched into the pre-trained vector transformation model deployed in the first device, and obtains the first vector corresponding to the first data. The second device logs the second data to be matched. Go to the pre-trained vector conversion model deployed in the second device to obtain the second vector corresponding to the second data. If the first vector corresponding to the first data U1 is (x1, x2, x3..., xm), the second vector corresponding to the second data U2 is (y1, y2, y3..., ym).

The first device generates a first target public-private key pair A(pka1, ska1), where pka1 is the first target public key and ska1 is the first target private key. Based on the first target public key pair A vector is homomorphically encrypted to generate the first encrypted vector. The first encrypted vector corresponding to the first vector (x1, x2, x3..., xm) is ( , , ), and send the first encryption vector and the first target public key to the second device.

After receiving the first target public key and the first encryption vector sent by the first device, the second device performs homomorphic encryption on the second vector based on the first target public key to obtain the second encryption vector. Specifically, the second encrypted vector corresponding to the second vector (y1, y2, y3..., ym) is ( , , ), the second device determines the distance between the encrypted first vector and the second vector based on the second encrypted vector and the received first encrypted vector and sends it to the first device.

After receiving the distance between the encrypted first vector and the second vector, the first device generates the encrypted first vector and the second vector according to the first target private key corresponding to the first target public key generated by itself. Decrypt the distance to determine the target distance of the first vector and the second vector. The first device can determine whether the first data and the second data match based on the target distance and the preset first distance threshold, and the first device will The target distance between the first vector and the second vector is sent to the second device. The second device determines whether the first data and the second data match based on the target distance between the second vector and the first vector and the preset first distance threshold. .

Example 7: In order to determine the target distance between the first vector and the second vector, on the basis of the above embodiments, in the embodiment of the present invention, the distance based on the encrypted first vector and the second vector and the first target The first target private key corresponding to the public key, and determining the target distance between the first vector and the second vector includes: Use the first target private key corresponding to the first target public key generated by the first device itself to decrypt the distance between the encrypted first vector and the second vector to determine the first vector and the second vector. target distance.

In the embodiment of the present invention, since the distance between the encrypted first vector and the second vector obtained by the first device is determined based on the first encrypted vector and the second encrypted vector, and the first encrypted vector and the second encrypted vector The vector is determined based on the first target public key generated by the first device. Therefore, in order to determine the target distance between the first vector and the second vector, the first device obtains the encrypted distance between the first vector and the second vector. Then, use the first target private key corresponding to the first target public key generated by the first device to decrypt the encrypted distance between the first vector and the second vector, and determine the target distance between the first vector and the second vector. .

Example 8: In order to determine the first encryption vector, based on the above embodiments, in the embodiment of the present invention, the first data to be matched is logged into the pre-trained vector transformation model, and the third data corresponding to the first data is obtained. A vector consists of: For each first sub-data in the first data, log the first sub-data into the pre-trained vector transformation model to obtain the first sub-vector corresponding to the first sub-data; wherein, each first sub-data is The length of the first sub-vector corresponding to the sub-data is the first preset length; The first sub-vectors corresponding to each first sub-data are spliced to obtain the first vector corresponding to the first data.

In the embodiment of the present invention, a first data may include one first sub-data, or may include multiple first sub-data. For example, the first data includes a first sub-data of "Shanghai Pudong New District Sunny Shop" Data, the first data can also include three first sub-data. For example, the three first sub-data are: "Shanghai Pudong New Area Qingtian Shop", "Shanghai Tiantian Restaurant", and "Gaoke Road" Yang Guofu Malatang".

In order to determine the first vector corresponding to the first data, for each first sub-data in the first data, the first sub-data can be logged into the pre-trained vector transformation model, and the corresponding first sub-data can be obtained. The first sub-vector, in which the length of the characters, numbers or characters contained in each first sub-data may be different, but the length of the first sub-vector corresponding to each first sub-data is the first preset length, The first preset length may be 3, 4, 6, etc. Specifically, the first preset length may be set according to requirements.

Taking the dimension of the vector output by the pre-trained vector conversion model as 5 and the pre-trained vector conversion model as a word vector model as an example to illustrate, if the first data contains three first sub-data, each first sub-data The data are all text data. The three first sub-data are "Shanghai Pudong New Area Qingtian Canteen", "Shanghai Tiantian Restaurant" and "Gaoke Road Yang Guofu Malatang", then "Shanghai Pudong New District Qingtian Canteen" Input into the pre-trained word vector model, and the first subvector corresponding to the output "Shanghai Pudong New District Sunny Shop" is (1.0, 2.0, 1.5, 2.0, 3.5), input "Shanghai Tiantian Restaurant" In the word vector model that has been trained in advance, the first subvector corresponding to the output "Shanghai Tiantian Restaurant" is (3.0, 4.0, 2.5, 2.5, 1.5), and "Gaoke Road Yang Guofu Malatang" is input into the pre-trained word vector model. In the word vector model that has been trained, the first subvector corresponding to "Gaoke Road Yang Guofu Malatang" output is (4.5, 5.5, 7.5, 1.5, 0.5).

If the first data contains three first sub-data, each first sub-data is digital data, and the three digital data are "12345", "11111", and "22233" respectively, then enter "12345" Into the pre-trained word vector model, the first subvector corresponding to the output "12345" is (0000000010, 0000000100, 0000001000, 0000010000, 0000100000). Input "11111" into the pre-trained word vector model. The first subvector corresponding to the output "11111" is (0000000010, 0000000010, 0000000010, 0000000010, 0000000010). Input "22233" into the pre-trained word vector model, and the first subvector corresponding to the output "22233" The subvectors are (0000000100, 0000000100, 0000000100, 0000001000, 0000001000).

In order to determine the first vector corresponding to the first data, in the embodiment of the present invention, after obtaining the first sub-vector corresponding to each first sub-data in the first data, the first vector corresponding to each first sub-data is obtained. The sub-vectors are spliced, and the splicing result is determined to be the first vector corresponding to the first data. Specifically, each first sub-data contained in the first data can be randomly sorted, and according to the first sub-data As a sorting result, the first sub-data corresponding to the first sub-vector is sorted and spliced to obtain the first vector.

For example, when the first data contains three first sub-data: "Shanghai Pudong New Area Qingtian Canteen", "Shanghai Tiantian Restaurant", and "Gaoke Road Yang Guofu Malatang", the corresponding "Shanghai Pudong New Area Qingtian Canteen" The first subvector is (1.0, 2.0, 1.5), the first subvector corresponding to "Shanghai Tiantian Restaurant" is (3.0, 4.0, 2.5), and the first subvector corresponding to "Gaoke Road Yang Guofu Malatang" is ( 4.5, 5.5, 7.5), then each first sub-data contained in the first data can be randomly sorted, and the sorting results obtained are "Shanghai Pudong New District Sunny Shop", "Gaoke Road Yang Guofu Malatang", "Shanghai Tiantian Restaurant", according to the sorting results, the first vector corresponding to the first data obtained after splicing the first sub-vectors corresponding to the three first sub-data is (1.0, 2.0, 1.5, 4.5, 5.5 ,7.5,3.0,4.0,2.5).

In order to achieve fuzzy matching between the first data and the second data, based on the above embodiments, in the embodiment of the present invention, the lengths of the first vector and the second vector are both second preset lengths.

In the embodiment of the present invention, in order to achieve fuzzy matching between the first data and the second data, the length of the first vector corresponding to the obtained first data and the length of the second vector corresponding to the second data must be the same, and both is a second preset length, wherein the second preset length is not less than the first preset length, the second preset length is an integer multiple of the first preset length, and if the first data only contains one A sub-data, then the first preset length is equal to the second preset length.

Since the length of the first vector corresponding to the first data and the length of the second vector corresponding to the second data are both the second preset length, even if the first data and the second data are different, fuzzy matching can be achieved. Broadened usage scenarios.

Example 9: In order to determine the first encryption vector, on the basis of the above embodiments, in the embodiment of the present invention, using the self-generated first target public key to homomorphically encrypt the first vector to generate the first encryption vector includes: For each first component in the first vector, determine the first square component corresponding to the first component; Insert the first square component corresponding to each first component into the first vector according to the preset insertion rules, and update the vector obtained after inserting the first square component to the first vector; Homomorphic encryption is performed on each first component and each first square component in the first vector based on the first target public key to generate the first encrypted vector.

In order to generate the first encrypted vector, in the embodiment of the present invention, the first vector can be directly homomorphically encrypted based on the first target public key to obtain the encrypted first encrypted vector. In order to ensure that the distance between the encrypted first vector and the second vector can be determined based on the first encrypted vector and the second encrypted vector without decrypting the first encrypted vector and the second encrypted vector, in the implementation of the present invention In this example, for each first component in the first vector, the first square component corresponding to the first component may also be determined first.

For example, if the first vector is (1, 2, 4, 5, 3), then the first square component corresponding to the first component of 1 in the first vector is 1, and the first square component of the first vector is 2. The first square component corresponding to one component is 4, the first square component corresponding to the first component 4 in the first vector is 16, and the first square component corresponding to the first component 5 in the first vector is 25 , the first square component corresponding to the first component of 3 in the first vector is 9.

In the embodiment of the present invention, for each first component in the first vector, after determining the first square component corresponding to the first component, the first square component corresponding to each first component can be calculated according to a preset The rule of is inserted into the first vector, and the vector obtained after inserting the first square component is updated as the first vector. Specifically, for the first square component corresponding to each first component, the first square component corresponding to the first component can be The first square component is inserted into any position in the first vector. For example, the first square component corresponding to the first component is inserted in front of the first component in the first vector, or the first square component corresponding to the first component is inserted into the first vector. The square component is inserted into the first vector after the first component, or the first square component is inserted into the back of the first component in sequence, as long as it is ensured that the first device and the second device can identify the first component in each vector. component and the first square component.

For example, if the first vector is (1, 2, 4, 5, 3), after determining the first square component of each first component in the first vector, insert the first square component into the first vector to obtain The updated first vector is (1, 9, 2, 16, 4, 4, 5, 25, 1, 3).

In order to facilitate the subsequent determination of the distance between the encrypted first vector and the second vector based on the first vector inserted into the first square component, in the embodiment of the present invention, for each first component in the first vector, after determining the After the first square component corresponding to the first component, the first square component corresponding to each first component can also be inserted into the first vector according to the preset insertion rule, and the vector obtained after inserting the first square component is updated. is the first vector. Specifically, the first square component corresponding to the first component can be inserted into the first vector at a position behind the first component and adjacent to the first component.

For example, if the first vector is (1, 2, 4, 5, 3), after determining the first square component of each first component in the first vector, insert the first square component into the first vector to obtain The updated first vector is (1, 1, 2, 4, 4, 16, 5, 25, 3, 9).

After determining the updated first vector, in order to determine the first encrypted vector, in the embodiment of the present invention, each first component and each first square in the first vector can be determined based on the first target public key. The components are homomorphically encrypted respectively to generate a first encryption vector.

For example, if the updated first vector is (2, 4, 3, 9), then the first encrypted vector is ( , , , ), among which, the Characterizes the result of homomorphic encryption of the first component of 2 in the first vector based on the target public key, which Characterizes the result of homomorphic encryption of the first component of 4 in the first vector based on the target public key, which Characterizes the result of homomorphic encryption of the first component of 3 in the first vector based on the target public key, which Characterizes the result of homomorphically encrypting the first component of 9 in the first vector based on the target public key.

In order to determine the distance between the encrypted first vector and the second vector, based on the above embodiments, in the embodiment of the present invention, the encrypted vector determined based on the first encrypted vector and the second encrypted vector is obtained. The distance between the first vector and the second vector includes: According to the preset insertion rule and each first component in the first encryption vector, obtain each set of first encryption components and first encryption square components; and according to the preset insertion rule and the second encryption vector For each second component of , obtain each set of second encrypted components and second encrypted square components; According to the preset insertion rule, determine each corresponding group of first encrypted component, first encrypted square component, second encrypted component and second encrypted square component; Determine each encrypted sub-distance according to each group of the first encrypted component, the first encrypted square component, the second encrypted component and the second encrypted square component; The distance between the encrypted first vector and the second vector is determined based on the sum of each sub-distance.

In the embodiment of the present invention, in order to determine the distance between the encrypted first vector and the second vector, the first device can obtain each group of the first component according to the preset insertion rule and the first encrypted vector. An encrypted component and a first encrypted square component, wherein a set of the first encrypted component and the first encrypted square component is obtained by performing homomorphic encryption on the same component itself and the square component of the component. That is, in each group of the first encrypted component and the first encrypted square component, the first encrypted square component is obtained by performing homomorphic encryption on the first square component corresponding to the first component of the first encrypted component before encryption.

Specifically, if the preset insertion rule is to insert the first square component corresponding to the first component into the first vector at a position behind the first component and adjacent to the first component, then after determining each group When the first encryption component and the first encryption component are used, you can directly start from the first component of the first vector, and combine the component of the group to be determined with the next component of the group to be determined in the first vector and the component of the group to be determined. The components of the group are determined as one group, and are divided sequentially until the first encrypted component and the first encrypted square component corresponding to all the groups are determined.

For example, the first encryption vector is ( , , , ), then two sets of first encrypted components and first encrypted square components can be obtained, where the first group is , the second group is . Among them, the first group is the first encrypted component in the first group, is the first encrypted square component in the first group, and the is the first encrypted component in the second group, is the first encrypted square component in this second group.

In the embodiment of the present invention, after obtaining the second encryption vector sent by the second device, the first device can also obtain each group of second components based on the preset insertion rules and each second component in the second encryption vector. Encrypted components and second encrypted square components, where the corresponding preset insertion rules when obtaining the first vector are the same as the preset insertion rules when obtaining the second vector, and each group of second encrypted components and second The process of encrypting square components is the same as the process of obtaining each group of first encrypted components and first encrypted square components, and will not be described again here.

In order to determine the distance between the encrypted first vector and the second vector, in the embodiment of the present invention, each group of the first encrypted component, the first encrypted square component, the second encrypted component and the second encrypted square component can be determined. , determine each sub-distance after encryption. Specifically, for each group, you can first determine the target sum value of the first encrypted square component and the second encrypted square component in the group, and then determine the first encrypted component and the second encrypted square component in the group. The target product of the encryption component and the preset value, where, in the embodiment of the present invention, the preset value is 2, finally determine the target difference between the target sum value and the target product, and determine the target difference as The encrypted subdistance of this group.

After each group of encrypted sub-distances is determined, the distance between the encrypted first vector and the second vector can be determined based on the sum of each sub-distance.

For example, if it is determined that the first vector corresponding to the first data is (1, 5) and the second vector corresponding to the second data is (2, 3), after determining the first square component of each first component in the first vector , insert each first square component into the first vector, the updated first vector obtained is (1, 1, 5, 25), and the first encrypted vector is determined to be ( , , , ), after determining the second square component of each second component in the second vector, insert each second square component into the second vector, and the updated second vector obtained is (2, 4, 3, 9) , determine the second encryption vector as ( , , , ), then determine the distance between the encrypted first vector and the second vector as [ ]+[ ].

Since the first data to be matched may include multiple first sub-data, and the second data to be matched may also include multiple second sub-data, therefore, for each first sub-data and each second sub-data, It can also determine the distance between the first sub-vector of the encrypted first sub-data and the second sub-vector of the second sub-data, and then determine whether each first sub-data and each second sub-data match, therefore, For convenience, the encrypted distance matrix can be determined based on the distance between each encrypted first sub-vector and the second sub-vector, where each element in the encrypted distance matrix is the corresponding encrypted first sub-vector. distance from the second subvector. After decrypting the encrypted distance matrix based on the target private key, the target sub-distance matrix is determined, where each element in the target sub-distance matrix is the target sub-distance of the corresponding first sub-vector and second sub-vector. .

After determining the target sub-distance of each first sub-vector and each second sub-vector, the target distance of the first vector and the second vector may also be determined based on the sum of each target sub-distance, and may also be determined Whether each first sub-data matches each second sub-data. Specifically, a sub-distance threshold can be set in advance. For each first sub-vector and each second sub-vector, according to the first sub-vector and the The target sub-distance of the second sub-vector determines whether the target sub-distance is less than the preset sub-distance threshold. If so, determines the first sub-data corresponding to the first sub-vector and the second sub-data corresponding to the second sub-vector. match.

Figure 2a is a schematic display diagram of a target sub-distance provided by some embodiments of the present invention. Figure 2b is a schematic display diagram of a target sub-distance matrix provided by some embodiments of the present invention. Figure 3a is another target provided by some embodiments of the present invention. A schematic diagram of the display of sub-distances. Figure 3b is a schematic diagram of the display of another target distance sub-matrix provided by some embodiments of the present invention. Figures 2a, 2b, 3a and 3b will now be described.

D(x,y) represents the target sub-distance, the first vector represented by x, and the second vector represented by y. If there are three first sub-data in the first data, there are also three second sub-data in the second data. sub-data, and the first sub-vectors corresponding to the three first sub-data are represented as A1, A2 and A3 respectively, and the second sub-vectors corresponding to the three second sub-data are represented as B1, B2 and B3 respectively, A1 and B1 The corresponding target sub-distance is 1, the target sub-distance corresponding to A1 and B2 is 3.64, the target sub-distance corresponding to A1 and B3 is 7.66, the target sub-distance corresponding to A2 and B1 is 3.9, and the target sub-distance corresponding to A2 and B2 is 0, the target sub-distance corresponding to A2 and B3 is 5.7, the target sub-distance corresponding to A3 and B1 is 8.35, the target sub-distance corresponding to A3 and B2 is 5.16, and the target sub-distance corresponding to A3 and B3 is 8.18, as shown in Figure 2a is shown, and the corresponding target sub-distance matrix is shown in Figure 2b.

If there are three first sub-data in the first data and three second sub-data in the second data, and the first sub-vectors corresponding to the three first sub-data are respectively represented as A1, A2 and A3, the three The second sub-vectors corresponding to the second sub-data are represented as B1, B2 and B3 respectively. The target sub-distance corresponding to A1 and B1 is 2.82, the target sub-distance corresponding to A1 and B2 is 1, and the target sub-distance corresponding to A1 and B3 is 3.16, the target sub-distance corresponding to A2 and B1 is 3, the target sub-distance corresponding to A2 and B2 is 1.73, the target sub-distance corresponding to A2 and B3 is 3, the target sub-distance corresponding to A3 and B1 is 0, and the target sub-distance corresponding to A3 and B2 The corresponding target sub-distance is 2.83, the corresponding target sub-distance between A3 and B3 is 3, as shown in Figure 3a, and the corresponding target sub-distance matrix is shown in Figure 3b.

Example 10: In order to determine whether the first data and the second data match, based on the above embodiments, in the embodiment of the present invention, it is determined based on the target distance and the preset first distance threshold that the first data and the third data are matched. Whether the two data match includes: Determine whether the target distance is less than a preset first distance threshold; If so, it is determined that the first data matches the second data; Otherwise, it is determined that the first data does not match the second data.

In order to determine whether the first data and the second data match, in the embodiment of the present invention, the target distance is compared with a preset distance threshold. If the target distance is less than the preset first distance threshold, it is determined that the target distance is smaller than the preset first distance threshold. The first data and the second data match. If the target distance is not less than the preset first distance threshold, it is determined that the first data and the second data do not match. The preset first distance threshold may be 1, 1.5, etc. Specifically, the preset first distance threshold may be set according to requirements. Wherein, the smaller the target distance is, the closer the first vector and the second vector match.

In order to determine whether the second data completely matches the first data, based on the above embodiments, in the embodiment of the present invention, after determining that the first data matches the second data, the method further includes: It is determined whether the target distance is equal to a preset second distance threshold, and if so, it is determined that the first data and the second data are the same.

In the embodiment of the present invention, if the target distance is equal to the preset second distance threshold, it means that the first data and the second data are the same, that is to say, the first data and the second data completely match, where , the preset second distance threshold is less than the preset first distance threshold, and the preset second distance threshold is equal to 0.

Example 11: In order to ensure that matching can be performed even when the data of both parties are not exactly the same, and to broaden the business scope of data matching, embodiments of the present invention provide a data matching method, device, equipment and medium.

Figure 4 is a schematic process diagram of a data matching method provided by an embodiment of the present invention. The process includes the following steps: S401: Log the second data to be matched into the pre-trained vector transformation model to obtain the second vector corresponding to the second data.

The data matching method provided by the embodiment of the present invention is applied to a second device. The second device may be a smart terminal, a PC, a server, etc., and the second device is a different device from the first device in the present invention.

In the embodiment of the present invention, in order to ensure that fuzzy matching can be achieved even when the data of both parties are not exactly the same, in the embodiment of the present invention, a pre-trained vector transformation model is deployed in the second device to obtain the data to be matched. Corresponding vectors, and for different data, the dimensions of the vectors output by the pre-trained vector transformation model are the same.

In order to obtain the second vector corresponding to the second data to be matched, the second data is logged into the pre-trained vector transformation model, and the pre-trained vector transformation model outputs the second vector corresponding to the second data.

In the embodiment of the present invention, the first data and the second data are generally the same type of data, for example, both are text data or both are digital data.

S402: Receive the first target public key sent by the first device, and use the first target public key to perform homomorphic encryption on the second vector to generate a second encryption vector.

In this embodiment of the present invention, in order to determine the second encryption vector, after receiving the first target public key sent by the first device, the second device uses the first target public key to homomorphically encrypt the second vector to generate the second Encrypted vector.

S403: Obtain the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, where the first encrypted vector is encrypted using the first target public key. Obtained later, the first vector is obtained by logging the first data into a pre-trained vector transformation model in the first device.

In this embodiment of the present invention, in order to enable the second device to also determine whether the first data matches the second data, the second device also obtains the first vector and the second vector determined based on the first encryption vector and the second encryption vector. The target distance of the two vectors, where the target distance can be determined by the first device and then sent to the second device, or can also be determined by the second device.

Wherein, the first encrypted vector is obtained by encrypting the first vector using the first target public key generated by the first device, and the first vector is pre-trained by logging the first data into the first device. Obtained from vector transformation model.

S404: Based on the target distance and the preset first distance threshold, determine whether the first data and the second data match.

In order to determine whether the first data and the second data match, in the embodiment of the present invention, the target distance is compared with a preset first distance threshold, and based on the comparison result, the first data and the second data are determined. Whether it matches.

In this embodiment of the present invention, the first data and the second data to be matched are respectively logged into the vector transformation model that has been trained in advance, and the first vector corresponding to the first data and the second vector corresponding to the second data are obtained. , and obtain the first encrypted vector encrypted based on the first vector, and the second encrypted vector encrypted based on the second vector, determine the distance between the encrypted first vector and the second vector, and obtain the encrypted first vector based on the encrypted second vector. The distance between the first vector and the second vector and the first target private key generated by itself determine the target distance between the first vector and the second vector, and determine the first data and the second data based on the target distance and the preset first distance threshold. Whether the data matches, that is, when the first data and the second data are not exactly the same, fuzzy matching between the first data and the second data can be achieved, broadening the usage scenarios, and introducing the first target disclosure during the fuzzy matching process The golden key and the first target private key are homomorphically encrypted and decrypted respectively, achieving safe intersection and ensuring the security of the matching process. During the entire matching process, neither the first data nor the second data are used as original data. The form has left the corresponding first device and second device, realizing fuzzy matching without leaving the original data out of the database, further ensuring the security of the matching process.

Example 12: In order to determine the second vector corresponding to the second data, on the basis of the above embodiments, in the embodiment of the present invention, the second data to be matched is logged into the pre-trained vector transformation model to obtain the The second vector corresponding to the second data includes: Determine the second target data type corresponding to the second data to be matched; Determine the pre-completed second target vector transformation model corresponding to the second data according to the corresponding relationship between the second target data type and the pre-saved data type and the pre-trained vector transformation model; Log the second data into the pre-trained second target vector transformation model to obtain the second vector corresponding to the second data.

In this embodiment of the present invention, the second data to be matched can be text data. For example, the first data can be name, gender, address, etc., or it can also be digital data. For example, the second data can be ID number, bank card number, admission ticket number, etc. Therefore, for different data types of second data, the pre-trained vector conversion models to obtain the corresponding second vectors are also different.

Specifically, the corresponding relationship between the data type and the pre-completed vector conversion model can be saved in the second device, and the corresponding pre-trained vector can be used according to the second target data type corresponding to the obtained second data to be matched. Transform the model to obtain the second vector corresponding to the second data; wherein the corresponding pre-trained vector transformation model is also the pre-trained second target vector transformation model.

Figure 5a is a schematic diagram of a process for obtaining vectors corresponding to text type data provided by some embodiments of the present invention. Figure 5b is a schematic diagram of a process for obtaining vectors corresponding to numeric type data provided by some embodiments of the present invention. Now with reference to Figure 5a and Figure 5b for explanation.

If the data to be matched is text data, and the data to be matched can be the first data or the second data, then the previously trained vector conversion model is a word vector model, and the text data is logged into the pre-trained word In the vector model, the word vector model outputs the vector corresponding to the text data.

If the data to be matched is digital data, the data to be matched can be the first data or the second data, then the pre-trained vector conversion model is a One-Hot encoding model, and the digital data is logged into the pre-trained In the One-Hot encoding model, the One-Hot encoding model outputs a vector corresponding to the digital data.

In order to accurately determine the model for converting the second data into the second vector, based on the above embodiments, if the second target data type is a text type, the corresponding pre-trained second target vector is converted into a word model Vector model or sentence vector model; if the second target data type is a numeric type, the corresponding pre-trained second target vector conversion model is a one-hot encoding model.

In this embodiment of the present invention, if the second data is text data, in order to obtain the second vector corresponding to the text data, the pre-trained vector conversion model deployed in the second device can be a word vector model or a sentence vector model. , if the second data is digital data, in order to determine the second vector corresponding to the digital data, the pre-trained vector conversion model deployed in the second device can be a One-Hot encoding model.

Specifically, if the second data is text data, the pre-completed second target vector conversion model corresponding to the second data is determined to be a word vector according to the corresponding relationship between the pre-saved data type and the pre-trained vector conversion model. model, and based on the pre-trained word vector model, obtain the second vector corresponding to the second data; if the second data is digital data, convert the model's correspondence according to the pre-saved data type and the pre-trained vector relationship, determine that the pre-completed second target vector conversion model corresponding to the second data is a One-Hot encoding model, and based on the pre-trained One-Hot encoding model, obtain the second vector corresponding to the second data.

Among them, when training the vector conversion model, each data and the corresponding label vector can be pre-labeled, each data and the corresponding label vector can be input into the original vector conversion model, and the prediction vector output by the original vector conversion model can be and the corresponding annotation vector, adjust the parameters of the original vector transformation model, and when the convergence conditions are met, it is determined that the training of the vector transformation model is completed.

Example 13: In order to obtain the target distance of the first vector and the second vector determined based on the first encryption vector and the second encryption vector, based on the above embodiments, receiving the target public key sent by the first device includes: Receive the first target public key and the first encryption vector sent by the first device, wherein the first encryption vector is obtained by homomorphically encrypting the first vector using the first target public key; The obtaining the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes: Determine the distance between the encrypted first vector and the second vector according to the first encrypted vector and the second encrypted vector; Send the encrypted distance between the first vector and the second vector to the first device; Receive the target distance sent by the first device, where the target distance is the encrypted sum of the first vector using the first target private key corresponding to the first target public key generated by the first device itself. The distance of the second vector is obtained after decryption.

In this embodiment of the present invention, the target distance of the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector obtained by the second device may be determined by the first device and sent to the second device, or may be determined for the second device.

Since the first encrypted vector and the second encrypted vector are generated by homomorphically encrypting the first vector and the second vector based on the first target public key generated by the first device, therefore, after determining the encrypted first vector and the second vector, After the distance between the two vectors, the encrypted distance between the first vector and the second vector also needs to be decrypted based on the first target private key generated by the first device. If the target distance between the first vector and the second vector If determined by the second device, the second device needs to receive the first target private key sent by the first device. If there is a vulnerability or attack during the sending process, the security of the information will be affected and the security will not be high.

In order to improve security, in this embodiment of the present invention, the target distance can be determined by the first device and sent to the second device. Specifically, the second device receives the first target public key sent by the first device. At the same time, the first encryption vector determined by the first device may also be received, wherein the first encryption vector is obtained by the first device using the first target public key to homomorphically encrypt the first vector.

After receiving the first encryption vector and the first target public key sent by the first device, the second device may first perform homomorphic encryption on the second vector based on the first target public key, determine the second encryption vector, and then perform homomorphic encryption on the second vector based on the first target public key. The first encryption vector and the second encryption vector determine the distance between the encrypted first vector and the second vector, because the first encryption vector and the second encryption vector are both based on the first target public and private key generated by the first device. The first target public key in the pair is encrypted and generated. Therefore, in the embodiment of the present invention, in order to determine the target distance between the first vector and the second vector, the second device may combine the determined encrypted first vector and the second vector. The distance between the two vectors is sent to the first device, so that the first device decrypts the encrypted distance between the first vector and the second vector according to the first target private key in the generated first target public private key pair. Generate a target distance and send the target distance to the second device, and the second device receives the target distance sent by the first device.

Since in the embodiment of the present invention, none of the first data has left the first device in the form of original data, fuzzy matching can be achieved without the original data leaving the database, further ensuring the security of the matching process.

Example 14: In order to obtain the target distance of the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, based on the above embodiments, the target distance determined based on the first encrypted vector and the second encrypted vector is obtained. The target distance between a vector and the second vector includes: Send the second encryption vector to the first device, so that the first device homomorphically encrypts the first vector based on the second encryption vector and the first target public key generated by itself. An encryption vector, determining the distance between the encrypted first encryption vector and the second encryption vector; Receive the target distance of the first vector and the second vector sent by the first device, where the target distance is the encrypted target private key corresponding to the first target public key generated by the first device. The distance between the first vector and the second vector is decrypted.

In the embodiment of the present invention, in order to obtain the target distance of the first vector and the second vector determined based on the first encryption vector and the second encryption vector, the first device may send the first target public key and the first target distance to the second device. An encryption vector, so that the second device determines the distance between the encrypted first vector and the second vector based on the first encryption vector and the second encryption vector generated by encrypting the second vector based on the first target public key, and A device can also only send the first target public key to the second device.

After receiving the first target public key sent by the first device, the second device performs homomorphic encryption on the second vector based on the first target public key, generates the second encrypted vector, and sends the second encrypted vector. To the first device, so that the first device determines the distance between the encrypted first encryption vector and the second encryption vector based on the second encryption vector and the first encryption vector, and based on the first target public secret generated by itself The first target private key in the key pair decrypts the encrypted distance between the first vector and the second vector, determines the target distance between the first vector and the second vector, and sends it to the second device, and the second device receives it The first device sends a target distance between the first vector and the second vector, and can determine whether the first data and the second data match based on the target distance.

Figure 6 is a schematic diagram of the overall process of fuzzy matching of data from both parties provided by some embodiments of the present invention. Figure 6 will now be described.

Log the first data and second data to be matched into the pre-trained vector transformation model, output the first vector corresponding to the first data to be matched, and the second vector corresponding to the second data to be matched, based on the A target public key performs homomorphic encryption on the first vector and the second vector respectively to obtain the first encrypted vector and the second encrypted vector. Based on the first encrypted vector and the second encrypted vector, the encrypted first vector is determined. The distance between the vector and the second vector is decrypted based on the first target private key, and the target distance between the first vector and the second vector is determined. Compare with the set first distance threshold to determine the matching result.

Figure 7 is a schematic diagram of a specific process for fuzzy matching of data from both parties provided by some embodiments of the present invention. Figure 7 will now be described.

The first device logs the first data to be matched into the pre-trained vector transformation model deployed in the first device, and obtains the first vector corresponding to the first data. The second device logs the second data to be matched. Go to the pre-trained vector conversion model deployed in the second device to obtain the second vector corresponding to the second data. As shown in Figure 7, the first data includes four first sub-data, namely U1, U2, U3, and U4, and the first sub-vector corresponding to U1 is (x11, x12, x13..., x1m), and the corresponding first sub-vector of U2 is The first subvector of U3 is (x21, x22, x23..., x2m), the first subvector corresponding to U3 is (x31, x32, x33..., x3m), and the first subvector corresponding to U4 is (x41, x42 ,x43...,x4m). The first vector corresponding to the first data is (x11, x12, x13..., x1m, x21, x22, x23..., x2m, x31, x32, x33..., x3m, x41, x42, x43..., x4m) . The second data includes four second sub-data, namely U5, U6, U7, and U8. The second sub-vector corresponding to U5 is (y11, y12, y13..., y1m), and the second sub-vector corresponding to U6 is (y21, y22, y23..., y2m), the second subvector corresponding to U7 is (y31, y32, y33..., y3m), and the second subvector corresponding to U8 is (y41, y42, y43..., y4m ), the second vector corresponding to the second data is (y11, y12, y13..., y1m, y21, y22, y23..., y2m, y31, y32, y33..., y3m, y41, y42, y43..., y4m).

The first device generates a first target public-private key pair A(pka, ska), where pka is the first target public key, ska is the first target private key, and the first target public-private key pair is Homomorphic encryption target public and private key pairs. The first vector is homomorphically encrypted based on the first target public key to generate a first encrypted vector, (x11, x12, x13..., x1m, x21, x22, x23..., x2m, x31, x32, x33... ,x3m,x41,x42,x43...,x4m) The corresponding first encryption vector is ( , , , , , , , , , , ). The first encrypted vector includes m/2 sets of encrypted first encrypted components and first encrypted square components.

The first device sends the first target public key and the first encryption vector to the second device. After receiving the first target public key and the first encryption vector, the second device generates data based on the first target public key. Perform homomorphic encryption on the second vector to generate the second encrypted vector, (y11, y12, y13..., y1m, y21, y22, y23..., y2m, y31, y32, y33..., y3m, y41, y42, y43 ..., y4m) corresponding to the second encryption vector is ( , , , , , , , , , ). The second encrypted vector includes m/2 sets of encrypted second encrypted components and second encrypted square components.

For each encrypted first sub-vector and each corresponding encrypted second sub-vector, the second device determines the distance between the encrypted first sub-vector and the second sub-vector, based on each encrypted The distance between the first subvector and the second subvector determines an encrypted distance matrix, and sends the encrypted distance matrix to the first device.

After receiving the encrypted distance matrix sent by the second device, the first device decrypts the encrypted distance matrix based on the first target private key generated by the first device, that is, for each encrypted third The distance between a sub-vector and the second sub-vector is decrypted to obtain a target sub-distance matrix. Based on the target sub-distance matrix, the first device determines whether each first sub-data matches each second sub-data, where, the Each element in the target sub-distance matrix is a corresponding target sub-distance. The target sub-distance matrix is sent to the second device. The second device also obtains the target sub-distance matrix. The second device determines each target sub-distance matrix based on the target sub-distance matrix. Whether the first sub-data matches each second sub-data.

Example 15: In order to obtain the target distance of the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, based on the above embodiments, the target distance determined based on the first encrypted vector and the second encrypted vector is obtained. The target distance between a vector and the second vector includes: Perform homomorphic encryption on the second vector using the self-generated second target public key to obtain a third encryption vector, and send the second target public key and the third encryption vector to the first device; Receive the information sent by the first device based on the third encrypted vector and the fourth encrypted vector, and determine the distance between the encrypted second vector and the first vector, wherein the fourth encrypted vector is obtained using the second target public gold. The key is generated by homomorphic encryption of the first vector; Based on the encrypted distance between the first vector and the second vector and the second target private key corresponding to the second target public key, determine the target distance between the first vector and the second vector, based on the target distance and a preset first distance threshold to determine whether the first data and the second data match.

In order to enable both the first device and the second device to determine whether the first data and the second data match, in the embodiment of the present invention, the second device can also obtain the target distance of the first vector and the second vector, where the third The target distance between a vector and a second vector may be obtained by the second device decrypting the encrypted distance between the first vector and the second vector sent by the first device.

Specifically, the second device can homomorphically encrypt the second vector using the second target public key generated by itself, obtain the third encryption vector, and send the third encryption vector and the second target public key to the third device. A device. After receiving the third encryption vector and the second target public key, the first device performs homomorphic encryption on the first vector based on the second target public key to obtain a fourth encryption vector. The first device may determine the encrypted distance between the first vector and the second vector based on the third encrypted vector and the fourth encrypted vector, and send the encrypted distance between the first vector and the second vector to the second device, After receiving the distance between the encrypted first vector and the second vector, the second device uses the second target private key corresponding to the second target public key generated by itself to encrypt the first vector and the second vector. Decrypt the distance to obtain the target distance between the first vector and the second vector, and determine whether the first data and the second data match based on the target distance between the first vector and the second vector and the preset first distance threshold.

Among them, the second device determines whether the first data and the second data match based on the target distance of the first vector and the second vector and the preset first distance threshold, which is the same as the first device based on the first vector and the second vector. The target distance and the preset first distance threshold are the same. The process of determining whether the first data and the second data match is the same and will not be described again here.

Since in the embodiment of the present invention, the second data has not left the second device in the form of original data, fuzzy matching can be achieved without the original data leaving the database, further ensuring the security of the matching process.

Example 16: In order to obtain the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, based on the above embodiments, the second data to be matched is logged into the pre-trained vector transformation In the model, obtaining the second vector corresponding to the second data includes: For each second sub-data in the second data, log the second sub-data into the pre-trained vector transformation model to obtain the second sub-vector corresponding to the second sub-data; wherein, each second sub-data is The lengths of the second sub-vectors corresponding to the sub-data are all the first preset length; The second sub-vectors corresponding to each second sub-data are spliced to obtain the second vector corresponding to the second data.

In this embodiment of the present invention, a second data may include one second sub-data, or may include multiple second sub-data. For example, the second data may include a second sub-data "Shanghai Pudong New District Sunny Shop" Data, the second data can also include three second sub-data. For example, the three second sub-data are: "Shanghai Pudong New District Qingtian Shop", "Shanghai Tiantian Restaurant", and "Gaoke Road" Yang Guofu Malatang".

In order to determine the second vector corresponding to the second data, for each second sub-data in the second data, the second sub-data can be logged into the pre-trained vector transformation model to obtain the second vector corresponding to the second sub-data. The second sub-vector, in which the length of the characters, numbers or characters contained in each second sub-data may be different, but the length of the second sub-vector corresponding to each second sub-data is the first preset length, The first preset length may be 3, 4, 6, etc. Specifically, the first preset length may be set according to requirements.

For example, when the second data contains three second sub-data: "Shanghai Pudong New Area Qingtian Canteen", "Shanghai Tiantian Restaurant", and "Gaoke Road Yang Guofu Malatang", enter "Shanghai Pudong New Area Qingtian Canteen" In the pre-trained word vector model, the second subvector corresponding to the output "Shanghai Pudong New District Qingtian Canteen" is (2.0, 3.0, 2.5, 1.0, 1.5), and "Shanghai Tiantian Restaurant" is input into In the pre-trained word vector model, the output second sub-vector corresponding to "Shanghai Tiantian Restaurant" is (2.3, 4.4, 3.5, 4.5, 2.5), and "Gaoke Road Yang Guofu Malatang" is input into the pre-training In the completed word vector model, the second subvector corresponding to the output "Gaoke Road Yang Guofu Malatang" is (2.5, 2.7, 8.3, 4.5, 1.5).

If the second data contains three second sub-data, each second sub-data is digital data, and the three digital data are "54321", "00001", and "33322" respectively, then enter "54321" Into the pre-trained word vector model, the second subvector corresponding to the output "54321" is (0000100000, 0000010000, 0000001000, 0000000100, 0000000010). Input "00001" into the pre-trained word vector model. The second subvector corresponding to the output "000011" is (0000000000, 0000000000, 0000000000, 0000000010, 0000000010). Input "33322" into the pre-trained word vector model, and the second subvector corresponding to the output "22233" The subvectors are (0000000010, 0000000010, 0000000010, 0000000100, 0000000100).

In order to determine the second vector corresponding to the second data, in the embodiment of the present invention, after obtaining the second sub-vector corresponding to each second sub-data in the second data, the second vector corresponding to each second sub-data is obtained. The sub-vectors are spliced, and the splicing result is determined as the second vector corresponding to the second data. Specifically, each second sub-data contained in the second data can be randomly sorted first, and according to the second sub-data As a sorting result, the second sub-data is sorted and spliced into the corresponding second sub-vectors to obtain the second vector.

For example, when the second data contains three second sub-data: "Shanghai Pudong New Area Qingtian Canteen", "Shanghai Tiantian Restaurant", and "Gaoke Road Yang Guofu Malatang", the corresponding "Shanghai Pudong New Area Qingtian Canteen" The second subvector is (1.0, 2.0, 1.5), the second subvector corresponding to "Shanghai Tiantian Restaurant" is (3.0, 4.0, 2.5), and the second subvector corresponding to "Gaoke Road Yang Guofu Malatang" is ( 4.5, 5.5, 7.5), then each second sub-data contained in the second data can be randomly sorted, and the sorting results obtained are "Shanghai Pudong New District Sunny Shop", "Gaoke Road Yang Guofu Malatang", "Shanghai Tiantian Restaurant", according to the sorting results, the second vector corresponding to the second data obtained after splicing the second sub-vectors corresponding to the three second sub-data is (1.0, 2.0, 1.5, 4.5, 5.5 ,7.5,3.0,4.0,2.5).

In order to achieve fuzzy matching between the first data and the second data, based on the above embodiments, the lengths of the first vector and the second vector are both second preset lengths.

In the embodiment of the present invention, in order to achieve fuzzy matching between the first data and the second data, the length of the first vector corresponding to the obtained first data and the length of the second vector corresponding to the second data must be the same, and both is a second preset length, wherein the second preset length is not less than the first preset length, the second preset length is an integer multiple of the first preset length, and if the second data contains only one For the second sub-data, the first preset length is equal to the second preset length.

Since the length of the first vector corresponding to the first data and the length of the second vector corresponding to the second data are both the second preset length, even if the first data and the second data are different, fuzzy matching can be achieved. Broadened usage scenarios.

Example 17: In order to determine the second encryption vector, based on the above embodiments, using the first target public key to homomorphically encrypt the second vector to generate the second encryption vector includes: For each second component in the second vector, determine the second square component corresponding to the second component; Insert the second square component corresponding to each second component into the second vector according to the preset insertion rules, and update the vector obtained after inserting the second square component to the second vector; Homomorphic encryption is performed on each second component and each second square component in the second vector based on the first target public key to generate the second encrypted vector.

In order to generate the second encrypted vector, in the embodiment of the present invention, the second vector can be directly homomorphically encrypted based on the first target public key to obtain the encrypted second encrypted vector. In order to ensure that the distance between the encrypted first vector and the second vector can be determined based on the first encrypted vector and the second encrypted vector without decrypting the second encrypted vector and the second encrypted vector, in the implementation of the present invention In this example, for each second component in the second vector, the second square component corresponding to the second component may also be determined first.

For example, if the second vector is (1, 2, 4, 5, 3), then the second square component corresponding to the second component of 1 in the second vector is 1, and the second square component of the second vector is 2. The second square component corresponding to the two components is 4, the second square component corresponding to the second component 4 in the second vector is 16, and the second square component corresponding to the second component 5 in the second vector is 25 , the second square component corresponding to the second component of 3 in the second vector is 9.

In the embodiment of the present invention, for each second component in the second vector, after determining the second square component corresponding to the second component, the second square component corresponding to each second component can be calculated according to a preset The rule of is inserted into the second vector, and the vector obtained after inserting the second square component is updated to the second vector. Specifically, for the second square component corresponding to each second component, the second square component corresponding to the second component can be The second square component is inserted into any position in the second vector. For example, the second square component corresponding to the second component is inserted in front of the second component in the second vector, or the second square component corresponding to the second component is inserted into the second vector. The square component is inserted into the second vector after the second component, or the second square component is inserted into the back of the second component in sequence, as long as it is ensured that the first device and the second device can identify the second component in each vector. component and the second square component.

For example, if the second vector is (1, 2, 4, 5, 3), after determining the second square component of each second component in the second vector, insert the second square component into the second vector to obtain The updated second vector is (1, 9, 2, 16, 4, 4, 5, 25, 1, 3).

In order to facilitate the subsequent determination of the distance between the encrypted first vector and the second vector based on the second vector inserted into the second square component, in the embodiment of the present invention, for each second component in the second vector, after determining the After the second square component corresponding to the second component, the second square component corresponding to each second component can also be inserted into the second vector according to the preset insertion rules, and the vector obtained after inserting the second square component is updated. is the second vector. Specifically, the second square component corresponding to the second component can be inserted into the second vector at a position behind the second component and adjacent to the second component.

For example, if the second vector is (1, 2, 4, 5, 3), after determining the second square component of each second component in the second vector, insert the second square component into the second vector to obtain The updated second vector is (1, 1, 2, 4, 4, 16, 5, 25, 3, 9).

After determining the updated second vector, in order to determine the second encrypted vector, in the embodiment of the present invention, each second component and each second square in the second vector can be matched based on the first target public key. The components are homomorphically encrypted respectively to generate a second encryption vector.

For example, if the updated second vector is (2, 4, 3, 9), then the second encrypted vector is ( , , , ), among which, the Characterizes the result of homomorphic encryption of the second component of 2 in the second vector based on the target public key, which Characterizes the result of homomorphic encryption of the second component of 4 in the second vector based on the target public key, which Characterizes the result of homomorphic encryption of the second component of 3 in the second vector based on the target public key, which Characterizes the result of homomorphically encrypting the second component of 9 in the second vector based on the target public key.

Example 18: In order to obtain the target distance of the first vector and the second vector determined based on the first encryption vector and the second encryption vector, based on the above embodiments, the encryption vector is determined based on the first encryption vector and the second encryption vector. The distance between the first vector and the second vector includes: According to the preset insertion rule and each second component in the second encryption vector, obtain each set of second encryption components and second encryption square components; and according to the preset insertion rule and the first encryption vector For each first component of , obtain each group of first encrypted component and first encrypted square component; According to the preset insertion rule, determine each corresponding group of first encrypted component, first encrypted square component, second encrypted component and second encrypted square component; Determine each encrypted sub-distance according to each group of the first encrypted component, the first encrypted square component, the second encrypted component and the second encrypted square component; The distance between the encrypted first vector and the second vector is determined based on the sum of each sub-distance.

In this embodiment of the present invention, in order to determine the distance between the encrypted first vector and the second vector, the second device can obtain each group of the second component for the preset insertion rule and the second encrypted vector. The second encrypted component and the second encrypted square component are obtained by homomorphically encrypting the same component itself and the square component of the component. That is, in each group of the second encrypted component and the second encrypted square component, the second encrypted square component is obtained by performing homomorphic encryption on the second square component corresponding to the second component of the second encrypted component before encryption.

Specifically, if the preset insertion rule is to insert the second square component corresponding to the second component into the second vector at a position behind the second component and adjacent to the second component. Then when determining each group of second encryption components and second encryption components, you can directly start from the first component of the second vector, and combine the component of the group to be determined with the lower component of the group to be determined in the second vector. One component of the group to be determined is determined as a group, and is divided sequentially until the second encrypted component and the second encrypted square component corresponding to all the groups are determined.

For example, the second encryption vector is ( , , , ), then two sets of second encrypted components and second encrypted square components can be obtained, where the second group is , the second group is . Among them, in the second group is the second encrypted component in the second group, is the second encrypted square component in the second group, and in the second group is the second encrypted component in the second group, is the second encrypted square component in the second group.

In the embodiment of the present invention, after obtaining the first encryption vector sent by the first device, the second device can also obtain each group of first components based on the preset insertion rules and each first component in the first encryption vector. Encrypted components and first encrypted square components, where the corresponding preset insertion rules when obtaining the second vector are the same as the preset insertion rules when obtaining the first vector, and each group of second encrypted components and second The process of encrypting square components is the same as the process of obtaining each group of first encrypted components and first encrypted square components, and will not be described again here.

In order to determine the distance between the encrypted first vector and the second vector, in the embodiment of the present invention, each group of the first encrypted component, the first encrypted square component, the second encrypted component and the second encrypted square component can be determined. , determine each encrypted sub-distance. Specifically, for each group, you can first determine the target sum of the first encrypted square vector and the second encrypted square component in the group, and then determine the first encrypted component and the second encrypted square component in the group. The target product of the two encryption components and a preset value, where, in the embodiment of the present invention, the preset value is 2, finally determine the target difference between the target sum value and the target product, and determine the target difference The encrypted subdistance determined for the group.

After the encrypted sub-distances of each group are determined, the distance between the encrypted first vector and the second vector can be determined based on the sum of each sub-distance.

Example 19: In order to determine whether the first data and the second data match, based on the above embodiments, determining whether the first data and the second data match based on the target distance and the preset first distance threshold includes: Determine whether the target distance is less than a preset first distance threshold; If so, it is determined that the first data matches the second data; Otherwise, it is determined that the first data does not match the second data.

In order to determine whether the first data and the second data match, in the embodiment of the present invention, the target distance is compared with a preset distance threshold. If the target distance is less than the preset first distance threshold, it is determined that the target distance is smaller than the preset first distance threshold. The first data and the second data match. If the target distance is not less than the preset first distance threshold, it is determined that the first data and the second data do not match. The preset first distance threshold may be 1, 1.5, etc. Specifically, the preset first distance threshold may be set according to requirements. Wherein, the smaller the target distance is, the closer the first vector and the second vector match.

In order to obtain the target distance of the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, based on the above embodiments, after it is determined that the first data matches the second data, the method Also includes: It is determined whether the target distance is equal to a preset second distance threshold, and if so, it is determined that the first data and the second data are the same.

In the embodiment of the present invention, if the target distance is equal to the preset second distance threshold, it means that the first data and the second data are the same, that is to say, the first data and the second data completely match, where , the preset second distance threshold is smaller than the preset first distance threshold, and the preset second distance threshold is equal to 0.

Example 20: Figure 8 is a schematic structural diagram of a data matching device provided by some embodiments of the present invention. The device includes: The first acquisition module 801 is used to log the first data to be matched into the pre-trained vector transformation model and obtain the first vector corresponding to the first data; The first processing module 802 is configured to use the first target public key generated by itself to perform homomorphic encryption on the first vector to generate a first encryption vector, and send the first target public key to the second device; The first obtaining module 801 is also used to obtain the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, wherein the second encrypted vector is obtained by using the first encrypted vector. The second vector is obtained by homomorphically encrypting the second vector with the target public key; the second vector is obtained by logging the second data into the pre-trained vector conversion model in the second device; The first determination module 803 is configured to determine the first vector and the second vector based on the encrypted distance between the first vector and the second vector and the first target private key corresponding to the first target public key. The target distance of the vector determines whether the first data and the second data match based on the target distance and the preset first distance threshold.

In a possible implementation, the first acquisition module 801 is specifically used to determine the first target data type corresponding to the first data; according to the first target data type, the pre-saved data type and the pre-trained The corresponding relationship of the vector transformation model is determined to determine the pre-trained first target vector transformation model corresponding to the first data; log the first data into the pre-trained first target vector transformation model to obtain the first data corresponding to the first vector.

In a possible implementation, the first processing module 802 is specifically configured to send the first encryption vector and the first target public key to the second device; The first acquisition module 801 is specifically configured to receive the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector sent by the second device, wherein, the The second encryption vector is obtained by the second device homomorphically encrypting the second vector based on the first target public key.

In a possible implementation, the first processing module 802 is also configured to receive the third encryption vector sent by the second device and the second target public key generated by the second device; wherein, the third encryption vector The vector is obtained by the second device using the second target public key to homomorphically encrypt the second vector; based on the second target public key, the first vector is homomorphically encrypted to generate a fourth encrypted vector; Based on the third encrypted vector and the fourth encrypted vector, determine the distance between the encrypted second vector and the first vector, and send the encrypted distance between the second vector and the first vector to the third encrypted vector. Two devices, so that the second device calculates the encrypted second vector based on the distance between the encrypted second vector and the first vector and the second target private key corresponding to the second target public key. Decrypt the distance from the first vector to determine the target distance between the second vector and the first vector, and determine the third vector based on the target distance between the second vector and the first vector and the preset first distance threshold. Whether the first data and the second data match.

In a possible implementation, the first processing module 801 is also used to send the target distance of the first vector and the second vector to the second device, so that the second device can base the target distance and the predetermined distance on the second device. A first distance threshold is set to determine whether the first data and the second data match.

In a possible implementation, the first acquisition module 801 is specifically configured to use the first target private key corresponding to the first target public key generated by the first device itself to encrypt the first vector. Decrypt the distance between the first vector and the second vector to determine the target distance between the first vector and the second vector.

In a possible implementation, the first acquisition module 801 is specifically configured to log the first sub-data into the pre-trained vector transformation model for each first sub-data in the first data, Obtain the first sub-vector corresponding to the first sub-data; wherein, the length of the first sub-vector corresponding to each first sub-data is a first preset length; convert the first sub-vector corresponding to each first sub-data into The vectors are spliced together to obtain the first vector corresponding to the first data.

In a possible implementation, the first processing module 802 is specifically configured to determine, for each first component in the first vector, the first square component corresponding to the first component; The corresponding first square component is inserted into the first vector according to the preset insertion rules, and the vector obtained after inserting the first square component is updated to the first vector; the first vector is updated based on the first target public key. Each first component and each first square component in the vector are respectively homomorphically encrypted to generate the first encrypted vector.

In a possible implementation, the first acquisition module 801 is specifically configured to acquire each group of first encrypted components and first components according to the preset insertion rule and each first component in the first encrypted vector. Encrypt square components; and obtain each set of second encrypted components and second encrypted square components according to the preset insertion rule and each second component in the second encrypted vector; determine the corresponding According to each group of the first encrypted component, the first encrypted square component, the second encrypted square component and the second encrypted square component, Determine each encrypted sub-distance; determine the distance between the encrypted first vector and the second vector based on the sum of each sub-distance.

In a possible implementation, the first determination module 803 is specifically used to determine whether the target distance is less than a preset first distance threshold; if so, determine that the first data matches the second data; otherwise, It is determined that the first information does not match the second information.

In a possible implementation, the first determination module 803 is also used to determine whether the target distance is equal to a preset second distance threshold, and if so, determine that the first data and the second data are the same.

Example 21: Figure 9 is a schematic structural diagram of a data matching device provided by some embodiments of the present invention. The device includes: The second acquisition module 901 is used to log the second data to be matched into the pre-trained vector transformation model and obtain the second vector corresponding to the second data; The second processing module 902 is configured to receive the first target public key sent by the first device, and use the first target public key to perform homomorphic encryption on the second vector to generate a second encryption vector; The second acquisition module 901 is also used to acquire the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, wherein the first encrypted vector is calculated using the first target. The first vector is obtained by encrypting the first vector with the public key, and the first vector is obtained by logging the first data into the pre-trained vector conversion model in the first device; The second determination module 903 is used to determine whether the first data and the second data match based on the target distance and the preset first distance threshold.

In a possible implementation, the second acquisition module 901 is specifically used to determine the second target data type corresponding to the second data; according to the second target data type, the pre-saved data type and the pre-trained The corresponding relationship of the vector transformation model is to determine the pre-trained second target vector transformation model corresponding to the second data; log the second data into the pre-trained second target vector transformation model to obtain the second data The corresponding second vector.

In a possible implementation, the second processing module 902 is specifically configured to receive the first target public key and the first encryption vector sent by the first device, wherein the first encryption vector is generated using the The first target public key is obtained by homomorphically encrypting the first vector; The second acquisition module 901 is specifically configured to determine the distance between the encrypted first vector and the second vector according to the first encrypted vector and the second encrypted vector; and combine the encrypted first vector and the second encrypted vector. Send the distance between the two vectors to the first device; receive the target distance sent by the first device, where the target distance is the first target private key corresponding to the first target public key generated by the first device itself. The key is obtained by decrypting the distance between the encrypted first vector and the second vector.

In a possible implementation, the second acquisition module 901 is specifically configured to send the second encryption vector to the first device, so that the first device can, based on the second encryption vector and the first encryption vector, Determine the distance between the encrypted first encrypted vector and the second encrypted vector; receive the target distance of the first vector and the second vector sent by the first device, wherein the target distance is generated by the first device based on itself The first target private key corresponding to the first target public key is obtained by decrypting the distance between the encrypted first encryption vector and the second encryption vector.

In a possible implementation, the second acquisition module 901 is specifically configured to homomorphically encrypt the second vector using the second target public key generated by itself, obtain the third encrypted vector, and convert the second Send the target public key and the third encryption vector to the first device; receive the information sent by the first device based on the third encryption vector and the fourth encryption vector, and determine the encrypted second vector and the first vector. distance, wherein the fourth encrypted vector is generated by homomorphically encrypting the first vector using the second target public key; based on the distance between the encrypted first vector and the second vector and the second target The second target private key corresponding to the public key determines the target distance between the first vector and the second vector, and determines whether the first data and the second data are based on the target distance and the preset first distance threshold. match.

In a possible implementation, the second acquisition module 901 is specifically configured to log the second sub-data into the pre-trained vector transformation model for each second sub-data in the second data, Obtain the second sub-vector corresponding to the second sub-data; wherein, the length of the second sub-vector corresponding to each second sub-data is the first preset length; convert the second sub-vector corresponding to each second sub-data into The vectors are spliced together to obtain the second vector corresponding to the second data.

In a possible implementation, the second processing module 902 is specifically configured to determine, for each second component in the second vector, the second square component corresponding to the second component; The corresponding second square component is inserted into the second vector according to the preset insertion rules, and the vector obtained after inserting the second square component is updated to the second vector; the second vector is updated based on the first target public key. Each second component and each second square component in the vector are respectively homomorphically encrypted to generate the second encrypted vector.

In a possible implementation, the second acquisition module 901 is specifically configured to acquire each group of second encrypted components and second components according to the preset insertion rule and each second component in the second encrypted vector. Encrypt square components; and obtain each group of first encrypted components and first encrypted square components according to the preset insertion rule and each first component in the first encrypted vector; determine the corresponding According to each group of the first encrypted component, the first encrypted square component, the second encrypted square component and the second encrypted square component, Determine each encrypted sub-distance; determine the distance between the encrypted first vector and the second vector based on the sum of each sub-distance.

In a possible implementation, the second determination module 903 is specifically used to determine whether the target distance is less than a preset first distance threshold; if so, determine that the first data matches the second data; otherwise, It is determined that the first information does not match the second information.

In a possible implementation, the second determination module 903 is also used to determine whether the target distance is equal to a preset second distance threshold, and if so, determine that the first data and the second data are the same.

Example 22: Based on the above embodiments, some embodiments of the present invention also provide an electronic device, as shown in Figure 10, including: a processor 1001, a communication interface 1002, a memory 1003 and a communication bus 1004, wherein the processor 1001. The communication interface 1002 and the memory 1003 complete communication with each other through the communication bus 1004.

The memory 1003 stores a computer program. When the program is executed by the processor 1001, the processor 1001 performs the following steps: Log the first data to be matched into the pre-trained vector transformation model to obtain the first vector corresponding to the first data; Use the self-generated first target public key to perform homomorphic encryption on the first vector to generate a first encryption vector, and send the first target public key to the second device; Obtain the distance between the encrypted first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, wherein the second encrypted vector is obtained by synchronizing the second vector with the first target public key. obtained after state encryption; the second vector is obtained by logging the second data into the pre-trained vector transformation model in the second device; Based on the encrypted distance between the first vector and the second vector and the first target private key corresponding to the first target public key, determine the target distance between the first vector and the second vector, based on the target distance and a preset first distance threshold to determine whether the first data and the second data match.

Further, the processor 1001 is also used to determine the first target data type corresponding to the first data; according to the first target data type and the corresponding relationship between the pre-saved data type and the pre-trained vector transformation model, determine The first pre-trained first target vector transformation model corresponding to the first data; log the first data into the pre-trained first target vector transformation model to obtain the first vector corresponding to the first data.

Further, the processor 1001 is also configured to receive the second encryption vector sent by the second device, wherein the second encryption vector is performed by the second device based on the first target public key. Obtained after homomorphic encryption; based on the first encrypted vector and the second encrypted vector, determine the distance between the encrypted first vector and the second vector.

Further, the processor 1001 is also configured to send the first encryption vector and the first target public key to the second device; receive the data sent by the second device based on the first encryption vector and the second encryption key. The distance between the encrypted first vector and the second vector determined by the vector, wherein the second encrypted vector is obtained by the second device after homomorphically encrypting the second vector based on the first target public key. .

Further, the processor 1001 is also configured to receive a third encryption vector sent by the second device and a second target public key generated by the second device; wherein the third encryption vector is generated by the second device using the The second target public key is obtained by homomorphically encrypting the second vector; the first vector is homomorphically encrypted based on the second target public key to generate a fourth encrypted vector; based on the third encrypted vector and the The fourth encrypted vector determines the distance between the encrypted second vector and the first vector, and sends the encrypted distance between the second vector and the first vector to the second device, so that the second The device calculates the distance between the encrypted second vector and the first vector based on the encrypted distance between the second vector and the first vector and the second target private key corresponding to the second target public key. Decrypt, determine the target distance between the second vector and the first vector, and determine whether the first data and the second data are based on the target distance between the second vector and the first vector and the preset first distance threshold. match.

Further, the processor 1001 is also configured to send the target distance of the first vector and the second vector to the second device, so that the second device determines based on the target distance and the preset first distance threshold. Whether the first data and the second data match.

Further, the processor 1001 is also configured to use the first target private key corresponding to the first target public key generated by the first device itself to decrypt the encrypted distance between the first vector and the second vector. , determine the target distance between the first vector and the second vector.

Further, the processor 1001 is also configured to log the first sub-data into the pre-trained vector transformation model for each first sub-data in the first data, and obtain the corresponding data of the first sub-data. The first sub-vector; wherein the length of the first sub-vector corresponding to each first sub-data is the first preset length; the first sub-vectors corresponding to each first sub-data are spliced to obtain the first The first vector corresponding to the data.

Further, the processor 1001 is further configured to determine, for each first component in the first vector, a first square component corresponding to the first component; and convert the first square component corresponding to each first component according to a predetermined Assume that the insertion rule is inserted into the first vector, and the vector obtained after inserting the first square component is updated to the first vector; based on the first target public key, each first component in the first vector is And each first square component is homomorphically encrypted separately to generate the first encrypted vector.

Further, the processor 1001 is also configured to obtain each group of first encrypted components and first encrypted square components according to the preset insertion rule and each first component in the first encrypted vector; and according to the preset Assuming the insertion rules and each second component in the second encryption vector, obtain each group of second encryption components and second encryption square components; according to the preset insertion rules, determine the corresponding group of first encryption components, The first encrypted square component, the second encrypted square component and the second encrypted square component; determine each encrypted sub-distance based on each group of the first encrypted component, the first encrypted square component, the second encrypted component and the second encrypted square component; The distance between the encrypted first vector and the second vector is determined based on the sum of each sub-distance.

Further, the processor 1001 is also used to determine whether the target distance is less than a preset first distance threshold; if so, determine that the first data matches the second data; otherwise, determine that the first data matches the third data. The two data do not match.

Further, the processor 1001 is also used to determine whether the target distance is equal to a preset second distance threshold, and if so, determine that the first data and the second data are the same.

The communication bus mentioned in the above-mentioned server can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into address bus, data bus, control bus, etc. For ease of presentation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

The communication interface 1002 is used for communication between the above-mentioned electronic device and other devices.

The memory may include random access memory (Random Access Memory, RAM) or non-volatile memory (Non-Volatile Memory, NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located far away from the aforementioned processor.

The above-mentioned processor can be a general-purpose processor, including a central processing unit, a network processor (Network Processor, NP), etc.; it can also be a digital signal processor (Digital Signal Processing, DSP), a special integrated circuit, or field programmable Gate array or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

Example 23: Based on the above embodiments, some embodiments of the present invention also provide an electronic device, as shown in Figure 11, including: a processor 1101, a communication interface 1102, a memory 1103 and a communication bus 1104, where the processor 1101. Communication interface 1102 and memory 1103 complete communication with each other through communication bus 1104.

The memory 1103 stores a computer program. When the program is executed by the processor 1101, the processor 1101 performs the following steps: Log the second data to be matched into the pre-trained vector transformation model to obtain the second vector corresponding to the second data; Receive the first target public key sent by the first device, and use the first target public key to perform homomorphic encryption on the second vector to generate a second encryption vector; Obtain the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, wherein the first encrypted vector is obtained by encrypting the first vector using the first target public key. , the first vector is obtained by logging the first data into the pre-trained vector transformation model in the first device; Based on the target distance and the preset first distance threshold, it is determined whether the first data and the second data match.

Further, the processor 1101 is also used to determine the second target data type corresponding to the second data; determine the second target data type according to the corresponding relationship between the second target data type and the pre-saved data type and the pre-trained vector transformation model. The pre-trained second target vector transformation model corresponding to the second data; log the second data into the pre-trained second target vector transformation model to obtain the second vector corresponding to the second data.

Further, the processor 1101 is also configured to receive the first target public key and the first encryption vector sent by the first device, wherein the first encryption vector is the first target public key using the first target public key. The first vector is obtained by performing homomorphic encryption; according to the first encrypted vector and the second encrypted vector, determine the distance between the encrypted first vector and the second vector; and combine the encrypted first vector and the second encrypted vector. Send the distance between the two vectors to the first device; receive the target distance sent by the first device, where the target distance is the first target private key corresponding to the first target public key generated by the first device itself. The key is obtained by decrypting the distance between the encrypted first vector and the second vector.

Further, the processor 1101 is also configured to send the second encryption vector to the first device, so that the first device determines the encrypted first encryption vector based on the second encryption vector and the first encryption vector. The distance between the vector and the second encrypted vector; receiving the target distance between the first vector and the second vector sent by the first device, where the target distance is the first target public key generated by the first device based on itself The corresponding first target private key is obtained by decrypting the distance between the encrypted first encryption vector and the second encryption vector.

Further, the processor 1101 is also configured to perform homomorphic encryption on the second vector using the second target public key generated by itself, obtain a third encryption vector, and combine the second target public key and the third Send the encryption vector to the first device; receive the information sent by the first device based on the third encryption vector and the fourth encryption vector, and determine the distance between the encrypted second vector and the first vector, wherein the fourth encryption The vector is generated by homomorphically encrypting the first vector using the second target public key; based on the distance between the encrypted first vector and the second vector and the second target corresponding to the second target public key The private key determines the target distance between the first vector and the second vector, and determines whether the first data and the second data match based on the target distance and the preset first distance threshold.

Further, the processor 1101 is also configured to log the second sub-data into the pre-trained vector transformation model for each second sub-data in the second data, and obtain the second sub-data corresponding to the second sub-data. The second sub-vector; wherein, the length of the second sub-vector corresponding to each second sub-data is the first preset length; the second sub-vector corresponding to each second sub-data is spliced to obtain the second sub-vector. The second vector corresponding to the data.

Further, the processor 1101 is also configured to determine, for each second component in the second vector, a second square component corresponding to the second component; and convert the second square component corresponding to each second component according to a predetermined Assume that the insertion rule is inserted into the second vector, and the vector obtained after inserting the second square component is updated to the second vector; based on the first target public key, each second component in the second vector is And each second square component is homomorphically encrypted separately to generate the second encrypted vector.

Further, the processor 1101 is also configured to obtain each group of second encrypted components and second encrypted square components according to the preset insertion rule and each second component in the second encrypted vector; and according to the preset Assuming the insertion rules and each first component in the first encryption vector, obtain each group of first encryption components and first encryption square components; according to the preset insertion rules, determine the corresponding group of first encryption components, The first encrypted square component, the second encrypted square component and the second encrypted square component; determine each encrypted sub-distance based on each group of the first encrypted component, the first encrypted square component, the second encrypted component and the second encrypted square component; The distance between the encrypted first vector and the second vector is determined based on the sum of each sub-distance.

Further, the processor 1101 is also used to determine whether the target distance is less than a preset first distance threshold; if so, determine that the first data matches the second data; otherwise, determine that the first data matches the third data. The two data do not match.

Further, the processor 1101 is also used to determine whether the target distance is equal to a preset second distance threshold, and if so, determine that the first data and the second data are the same.

The communication bus mentioned in the above-mentioned server can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into address bus, data bus, control bus, etc. For ease of presentation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

The communication interface 1102 is used for communication between the above-mentioned electronic device and other devices.

The memory may include random access memory (Random Access Memory, RAM) or non-volatile memory (Non-Volatile Memory, NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located far away from the aforementioned processor.

The above-mentioned processor can be a general-purpose processor, including a central processing unit, a network processor (Network Processor, NP), etc.; it can also be a digital signal processor (Digital Signal Processing, DSP), a special integrated circuit, or field programmable Gate array or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

Example 24: Based on the above embodiments, embodiments of the present invention also provide a computer-readable storage medium. The computer-readable storage medium stores a computer program that can be executed by an electronic device. When the program is run on the electronic device, , so that the electronic device can implement the following steps when executing: Log the first data to be matched into the pre-trained vector transformation model to obtain the first vector corresponding to the first data; Use the self-generated first target public key to perform homomorphic encryption on the first vector to generate a first encryption vector, and send the first target public key to the second device; Obtain the distance between the encrypted first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, wherein the second encrypted vector is obtained by synchronizing the second vector with the first target public key. obtained after state encryption; the second vector is obtained by logging the second data into the pre-trained vector transformation model in the second device; Based on the encrypted distance between the first vector and the second vector and the first target private key corresponding to the first target public key, determine the target distance between the first vector and the second vector, based on the target distance and a preset first distance threshold to determine whether the first data and the second data match.

Further, logging the first data to be matched into the pre-trained vector transformation model and obtaining the first vector corresponding to the first data includes: Determine the first target data type corresponding to the first data; Determine the pre-trained first target vector transformation model corresponding to the first data according to the corresponding relationship between the first target data type and the pre-saved data type and the pre-trained vector transformation model; Log the first data into the pre-trained first target vector transformation model to obtain the first vector corresponding to the first data.

Further, the first target data type is a text type or a numeric type.

Further, if the first target data type is a text type, the corresponding pre-trained first target vector conversion model is a word vector model or a sentence vector model; if the first target data type is a numeric type, the corresponding pre-trained The completed first target vector conversion model is a one-hot encoding model.

Further, obtaining the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes: Receive the second encryption vector sent by the second device, wherein the second encryption vector is obtained by the second device after homomorphically encrypting the second vector based on the first target public key; Based on the first encrypted vector and the second encrypted vector, the distance between the encrypted first vector and the second vector is determined.

Further, sending the first target public key to the second device includes: Send the first encryption vector and the first target public key to the second device; The obtaining the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes: Receive the encrypted distance between the first vector and the second vector sent by the second device and determined based on the first encrypted vector and the second encrypted vector, where the second encrypted vector is determined by the second device based on the The first target public key is obtained by homomorphically encrypting the second vector.

Further, the method also includes: Receive the third encryption vector sent by the second device and the second target public key generated by the second device; wherein the third encryption vector is the second device using the second target public key to encrypt the second vector. Obtained after homomorphic encryption; Perform homomorphic encryption on the first vector based on the second target public key to generate a fourth encrypted vector; Based on the third encrypted vector and the fourth encrypted vector, determine the distance between the encrypted second vector and the first vector, and send the encrypted distance between the second vector and the first vector to the third encrypted vector. Two devices, so that the second device calculates the encrypted second vector based on the distance between the encrypted second vector and the first vector and the second target private key corresponding to the second target public key. Decrypt the distance from the first vector to determine the target distance between the second vector and the first vector, and determine the third vector based on the target distance between the second vector and the first vector and the preset first distance threshold. Whether the first data and the second data match.

Further, after determining the target distance between the first vector and the second vector, the method further includes: The target distance of the first vector and the second vector is sent to the second device, so that the second device determines whether the first data and the second data match based on the target distance and the preset first distance threshold. .

Further, determining the target distance between the first vector and the second vector based on the encrypted distance between the first vector and the second vector and the first target private key corresponding to the first target public key includes: Use the first target private key corresponding to the first target public key generated by the first device itself to decrypt the distance between the encrypted first vector and the second vector to determine the first vector and the second vector. target distance.

Further, logging the first data to be matched into the pre-trained vector transformation model and obtaining the first vector corresponding to the first data includes: For each first sub-data in the first data, log the first sub-data into the pre-trained vector transformation model to obtain the first sub-vector corresponding to the first sub-data; wherein, each first sub-data is The length of the first sub-vector corresponding to the sub-data is the first preset length; The first sub-vectors corresponding to each first sub-data are spliced to obtain the first vector corresponding to the first data.

Further, the lengths of the first vector and the second vector are both second preset lengths.

Further, using the self-generated first target public key to homomorphically encrypt the first vector to generate the first encrypted vector includes: For each first component in the first vector, determine the first square component corresponding to the first component; Insert the first square component corresponding to each first component into the first vector according to the preset insertion rules, and update the vector obtained after inserting the first square component to the first vector; Homomorphic encryption is performed on each first component and each first square component in the first vector based on the first target public key to generate the first encrypted vector.

Further, obtaining the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes: According to the preset insertion rule and each first component in the first encryption vector, obtain each set of first encryption components and first encryption square components; and according to the preset insertion rule and the second encryption vector For each second component of , obtain each set of second encrypted components and second encrypted square components; According to the preset insertion rule, determine each corresponding group of first encrypted component, first encrypted square component, second encrypted component and second encrypted square component; Determine each encrypted sub-distance according to each group of the first encrypted component, the first encrypted square component, the second encrypted component and the second encrypted square component; The distance between the encrypted first vector and the second vector is determined based on the sum of each sub-distance.

Further, determining whether the first data and the second data match based on the target distance and a preset first distance threshold includes: Determine whether the target distance is less than a preset first distance threshold; If so, it is determined that the first data matches the second data; Otherwise, it is determined that the first data does not match the second data.

Further, after determining that the first data matches the second data, the method further includes: It is determined whether the target distance is equal to a preset second distance threshold, and if so, it is determined that the first data and the second data are the same.

Example 25: Based on the above embodiments, embodiments of the present invention also provide a computer-readable storage medium. The computer-readable storage medium stores a computer program that can be executed by an electronic device. When the program is run on the electronic device, , so that the electronic device can implement the following steps when executing: Log the second data to be matched into the pre-trained vector transformation model to obtain the second vector corresponding to the second data; Receive the first target public key sent by the first device, and use the first target public key to perform homomorphic encryption on the second vector to generate a second encryption vector; Obtain the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, wherein the first encrypted vector is obtained by encrypting the first vector using the first target public key. , the first vector is obtained by logging the first data into the pre-trained vector transformation model in the first device; Based on the target distance and the preset first distance threshold, it is determined whether the first data and the second data match.

Further, logging the second data to be matched into the pre-trained vector transformation model, and obtaining the second vector corresponding to the second data includes: Determine the second target data type corresponding to the second data; According to the second target data type and the corresponding relationship between the pre-saved data type and the pre-trained vector transformation model, determine the pre-trained second target vector transformation model corresponding to the second data; Log the second data into the pre-trained second target vector transformation model to obtain the second vector corresponding to the second data.

Further, the second target data type is a text type or a numeric type.

Further, if the second target data type is a text type, the corresponding pre-trained second target vector conversion model is a word vector model or a sentence vector model; if the second target data type is a numeric type, the corresponding pre-trained The completed second target vector transformation model is a one-hot encoding model.

Further, receiving the first target public key sent by the first device includes: Receive the first target public key and the first encryption vector sent by the first device, wherein the first encryption vector is obtained by homomorphically encrypting the first vector using the first target public key; The obtaining the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes: Determine the distance between the encrypted first vector and the second vector according to the first encrypted vector and the second encrypted vector; Send the encrypted distance between the first vector and the second vector to the first device; Receive the target distance sent by the first device, where the target distance is the encrypted sum of the first vector using the first target private key corresponding to the first target public key generated by the first device itself. The distance of the second vector is obtained after decryption.

Further, obtaining the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes: Send the second encryption vector to the first device, so that the first device determines the distance between the encrypted first encryption vector and the second encryption vector based on the second encryption vector and the first encryption vector; Receive the target distance of the first vector and the second vector sent by the first device, wherein the target distance is the first target private key corresponding to the first target public key generated by the first device based on itself. The distance between the encrypted first encrypted vector and the second encrypted vector is obtained by decrypting.

Further, obtaining the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes: Perform homomorphic encryption on the second vector using the self-generated second target public key to obtain a third encryption vector, and send the second target public key and the third encryption vector to the first device; Receive the information sent by the first device based on the third encrypted vector and the fourth encrypted vector, and determine the distance between the encrypted second vector and the first vector, wherein the fourth encrypted vector is obtained using the second target public gold. The key is generated by homomorphic encryption of the first vector; Based on the encrypted distance between the first vector and the second vector and the second target private key corresponding to the second target public key, determine the target distance between the first vector and the second vector, based on the target distance and a preset first distance threshold to determine whether the first data and the second data match.

Further, logging the second data to be matched into the pre-trained vector transformation model, and obtaining the second vector corresponding to the second data includes: For each second sub-data in the second data, log the second sub-data into the pre-trained vector transformation model to obtain the second sub-vector corresponding to the second sub-data; wherein, each second sub-data is The lengths of the second sub-vectors corresponding to the sub-data are all the first preset length; The second sub-vectors corresponding to each second sub-data are spliced to obtain the second vector corresponding to the second data.

Further, the lengths of the first vector and the second vector are both second preset lengths.

Further, using the first target public key to perform homomorphic encryption on the second vector to generate a second encrypted vector includes: For each second component in the second vector, determine the second square component corresponding to the second component; Insert the second square component corresponding to each second component into the second vector according to the preset insertion rules, and update the vector obtained after inserting the second square component to the second vector; Homomorphic encryption is performed on each second component and each second square component in the second vector based on the first target public key to generate the second encrypted vector.

Further, determining the distance between the encrypted first vector and the second vector according to the first encrypted vector and the second encrypted vector includes:

According to the preset insertion rule and each second component in the second encryption vector, obtain each set of second encryption components and second encryption square components; and according to the preset insertion rule and the first encryption vector For each first component of , obtain each group of first encrypted component and first encrypted square component; According to the preset insertion rule, determine each corresponding group of first encrypted component, first encrypted square component, second encrypted component and second encrypted square component; Determine each encrypted sub-distance according to each group of the first encrypted component, the first encrypted square component, the second encrypted component and the second encrypted square component; The distance between the encrypted first vector and the second vector is determined based on the sum of each sub-distance.

Further, determining whether the first data and the second data match based on the target distance and a preset first distance threshold includes: Determine whether the target distance is less than a preset first distance threshold; If so, it is determined that the first data matches the second data; Otherwise, it is determined that the first data does not match the second data.

Further, after determining that the first data matches the second data, the method further includes: It is determined whether the target distance is equal to a preset second distance threshold, and if so, it is determined that the first data and the second data are the same.

In this embodiment of the present invention, the first data and the second data to be matched are respectively logged into the vector transformation model that has been trained in advance, and the first vector corresponding to the first data and the second vector corresponding to the second data are obtained. , and obtain the first encrypted vector after encrypting the first vector and the second encrypted vector after encrypting the second vector based on the first target public key, and determine the distance between the encrypted first vector and the second vector, And based on the encrypted distance between the first vector and the second vector and the first target private key generated by itself, the target distance between the first vector and the second vector is determined based on the target distance and the preset first distance threshold. Whether the first data and the second data match, that is, when the first data and the second data are not exactly the same, fuzzy matching between the first data and the second data can also be achieved, broadening the usage scenarios, and during the fuzzy matching process The first target public key and the first target private key are introduced to perform homomorphic encryption and decryption respectively, achieving safe intersection and ensuring the security of the matching process. During the entire matching process, the first data and the third Neither of the two data has left the corresponding first device and the second device in the form of original data, which enables fuzzy matching to be achieved without the original data leaving the database, further ensuring the security of the matching process.

Those skilled in the art will understand that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the invention may take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk memory, CD-ROM, optical memory, etc.) embodying computer-usable program code therein. .

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to the invention. It will be understood that each process and/or block in the flowchart illustrations and/or block diagrams, and combinations of processes and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing device to produce a machine that causes instructions to be executed by the processor of the computer or other programmable data processing device. Means are generated for implementing the functions specified in the process or processes of the flowchart diagram and/or the block or blocks of the block diagram.

These computer program instructions may also be stored in a computer-readable memory that causes a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including the instruction device. , the instruction device implements the functions specified in one process or multiple processes in the flow chart and/or one block or multiple blocks in the block diagram.

These computer program instructions may also be loaded onto a computer or other programmable data processing device, causing a series of operating steps to be performed on the computer or other programmable device to produce computer-implemented processing, thereby causing the computer or other programmable device to perform a computer-implemented process. The instructions executed on provide steps for implementing the functions specified in a process or processes of the flow diagrams and/or a block or blocks of the block diagrams.

Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the invention. In this way, if these modifications and variations of the present invention fall within the patent scope of the present invention and the scope of equivalent technologies, the present invention is also intended to include these modifications and variations.

S101~S104, S401~S404: steps 801:First acquisition module 802: First processing module 803: First confirmed module 901: Second acquisition module 902: Second processing module 903: Second determined module 1001: Processor 1002: Communication interface 1003:Memory 1004: Communication bus 1101: Processor 1102: Communication interface 1103:Memory 1104: Communication bus

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings needed to describe the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the field, other drawings can also be obtained based on these drawings without incurring any progressive labor: Figure 1 is a schematic diagram of a data matching process provided by an embodiment of the present invention; Figure 2a is a schematic diagram showing a target sub-distance display provided by some embodiments of the present invention; Figure 2b is a schematic display diagram of a target sub-distance matrix provided by some embodiments of the present invention; Figure 3a is a schematic diagram showing another target sub-distance display provided by some embodiments of the present invention; Figure 3b is a schematic display diagram of another target sub-distance matrix provided by some embodiments of the present invention; Figure 4 is a schematic process diagram of a data matching method provided by an embodiment of the present invention; Figure 5a is a schematic diagram of a process for obtaining vectors corresponding to text type data provided by some embodiments of the present invention; Figure 5b is a schematic diagram of a process for obtaining vectors corresponding to digital data provided by some embodiments of the present invention; Figure 6 is a schematic diagram of the overall process of fuzzy matching of data from both parties provided by some embodiments of the present invention; Figure 7 is a schematic diagram of a specific process for fuzzy matching of data from both parties provided by some embodiments of the present invention; Figure 8 is a schematic structural diagram of a data matching device provided by some embodiments of the present invention; Figure 9 is a schematic structural diagram of a data matching device provided by some embodiments of the present invention; Figure 10 is a schematic structural diagram of an electronic device provided by some embodiments of the present invention; Figure 11 is a schematic structural diagram of an electronic device provided by some embodiments of the present invention.

S101~S104: Steps

Claims (30)

A data matching method, characterized in that, applied to a first device, the method includes: logging the first data to be matched into a vector transformation model that has been trained in advance, and obtaining a first vector corresponding to the first data; using its own The generated first target public key performs homomorphic encryption on the first vector to generate a first encryption vector, and sends the first target public key to the second device; obtaining the first encryption vector based on the first encryption vector and the second encryption vector. The determined distance between the encrypted first vector and the second vector, where the second encrypted vector is obtained by homomorphically encrypting the second vector using the first target public key; the second vector is the The second data is obtained by logging into the pre-trained vector conversion model in the second device; based on the encrypted distance between the first vector and the second vector and the first target corresponding to the first target public key The private key determines the target distance between the first vector and the second vector, and determines whether the first data and the second data match based on the target distance and the preset first distance threshold; the third data generated by itself is used. Homomorphically encrypting the first vector with a target public key to generate the first encrypted vector includes: for each first component in the first vector, determining the first square component corresponding to the first component; The first square component corresponding to a component is inserted into the first vector according to the preset insertion rules, and the vector obtained after inserting the first square component is updated to the first vector; based on the first target public key, the Each first component and each first square component in the first vector are respectively homomorphically encrypted to generate the first encrypted vector. The data matching method as described in claim 1, wherein logging the first data to be matched into the pre-trained vector transformation model and obtaining the first vector corresponding to the first data includes: determining the corresponding The first target data type; according to the first target data type and the corresponding relationship between the pre-saved data type and the pre-trained vector transformation model, determine the pre-trained first target vector transformation model corresponding to the first data; Log the first data into the pre-trained first target vector transformation model to obtain the first vector corresponding to the first data. The data matching method as described in claim 2, wherein the first target data type is a text type or a numeric type. The data matching method as described in claim 3, wherein if the first target data type is a text type, the corresponding pre-trained first target vector conversion model is a word vector model or a sentence vector model; if the first target The data type is a numeric type, and the corresponding pre-trained first target vector conversion model is a one-hot encoding model. The data matching method as described in claim 1, wherein the obtaining the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes: receiving the encrypted distance sent by the second device. The second encryption vector, wherein the second encryption vector is obtained by the second device homomorphically encrypting the second vector based on the first target public key; based on the first encryption vector and the second encryption vector, determine the distance between the encrypted first vector and the second vector. The data matching method as described in claim 1, wherein sending the first target public key to the second device includes: The first encrypted vector and the first target public key are sent to the second device; the obtaining the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes : Receive the distance between the encrypted first vector and the second vector determined based on the first encrypted vector and the second encrypted vector sent by the second device, where the second encrypted vector is based on The first target public key is obtained by homomorphically encrypting the second vector. The data matching method as described in request item 6, wherein the method further includes: receiving a third encryption vector sent by the second device and a second target public key generated by the second device; wherein, the third encryption vector It is obtained by homomorphically encrypting the second vector using the second target public key for the second device; homomorphically encrypting the first vector based on the second target public key to generate a fourth encrypted vector; based on The third encrypted vector and the fourth encrypted vector determine the distance between the encrypted second vector and the first vector, and send the encrypted distance between the second vector and the first vector to the second device, so that the second device calculates the sum of the encrypted second vector and the first vector according to the distance between the encrypted second vector and the first vector and the second target private key corresponding to the second target public key. Decrypt the distance of the first vector to determine the target distance between the second vector and the first vector, and determine the first distance based on the target distance between the second vector and the first vector and the preset first distance threshold. information and the second information match. The data matching method as described in claim 1, 5 or 6, wherein after determining the target distance between the first vector and the second vector, the method further includes: The target distance of the first vector and the second vector is sent to the second device, so that the second device determines whether the first data and the second data match based on the target distance and the preset first distance threshold. . The data matching method as described in claim 1, 5 or 6, wherein the first target private key corresponding to the first target public key is based on the encrypted distance between the first vector and the second vector, Determining the target distance between the first vector and the second vector includes: using the first target private key corresponding to the first target public key generated by the first device itself to calculate the encrypted first vector and the second vector. The distance is decrypted to determine the target distance between the first vector and the second vector. The data matching method as described in claim 1, wherein logging the first data to be matched into a pre-trained vector transformation model and obtaining the first vector corresponding to the first data includes: targeting the first data For each first sub-data, log the first sub-data into the pre-trained vector transformation model to obtain the first sub-vector corresponding to the first sub-data; where, the first sub-data corresponding to each first The lengths of the sub-vectors are all the first preset length; the first sub-vectors corresponding to each first sub-data are spliced to obtain the first vector corresponding to the first data. The data matching method as described in claim 1 or 10, wherein the lengths of the first vector and the second vector are both second preset lengths. The data matching method as described in claim 1, wherein the obtaining the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes: According to the preset insertion rule and each first component in the first encryption vector, obtain each set of first encryption components and first encryption square components; and according to the preset insertion rule and the second encryption vector For each second component of The second encrypted square component; according to each group of the first encrypted component, the first encrypted square component, the second encrypted component and the second encrypted square component, determine each encrypted sub-distance; according to the sum of each sub-distance, determine the encrypted The distance between the first vector and the second vector. The data matching method of claim 1, wherein determining whether the first data and the second data match based on the target distance and a preset first distance threshold includes: determining whether the target distance is less than a preset distance A first distance threshold; if yes, it is determined that the first data and the second data match; otherwise, it is determined that the first data and the second data do not match. The data matching method as described in claim 13, wherein after determining that the first data matches the second data, the method further includes: determining whether the target distance is equal to a preset second distance threshold, and if so, determining The first data is the same as the second data. A data matching method, characterized in that, applied to a second device, the method includes: logging the second data to be matched into a pre-trained vector transformation model, and obtaining a second vector corresponding to the second data; Receive the first target public key sent by the first device, use the first target public key to perform homomorphic encryption on the second vector to generate a second encryption vector; obtain the second encryption vector determined based on the first encryption vector and the second encryption vector. The target distance between the first vector and the second vector, wherein the first encrypted vector is obtained by homomorphically encrypting the first vector using the first target public key, and the first vector is the first data Obtained by logging into the pre-trained vector transformation model in the first device; based on the target distance and the preset first distance threshold, determine whether the first data and the second data match; use the first Homomorphically encrypting the second vector with the target public key to generate the second encrypted vector includes: for each second component in the second vector, determining the second square component corresponding to the second component; The second square component corresponding to the component is inserted into the second vector according to the preset insertion rules, and the vector obtained after inserting the second square component is updated to the second vector; based on the first target public key, the second square component is inserted into the second vector based on the first target public key. Each second component and each second square component in the two vectors are homomorphically encrypted respectively to generate the second encrypted vector. The data matching method as described in claim 15, wherein logging the second data to be matched into the pre-trained vector transformation model and obtaining the second vector corresponding to the second data includes: determining the corresponding The second target data type; according to the second target data type and the corresponding relationship between the pre-saved data type and the pre-trained vector transformation model, determine the pre-trained second target vector transformation model corresponding to the second data; Log the second data into the pre-trained second target vector transformation model to obtain the second vector corresponding to the second data. The data matching method as described in claim 16, wherein the second target data type is a text type or a numeric type. The data matching method as described in claim 17, wherein if the second target data type is a text type, the corresponding pre-trained second target vector conversion model is a word vector model or a sentence vector model; if the second target data type is The data type is a numeric type, and the corresponding pre-trained second target vector conversion model is a one-hot encoding model. The data matching method as described in claim 15, wherein receiving the first target public key sent by the first device includes: receiving the first target public key and the first encryption vector sent by the first device, wherein , the first encryption vector is obtained by homomorphically encrypting the first vector using the first target public key; the acquisition is based on the first vector and the second encryption vector determined based on the first encryption vector and the second encryption vector. The target distance of the vector includes: determining the encrypted distance between the first vector and the second vector based on the first encrypted vector and the second encrypted vector; and sending the encrypted distance between the first vector and the second vector. To the first device; receive the target distance sent by the first device, where the target distance is the encrypted value of the first target private key corresponding to the first target public key generated by the first device itself. It is obtained by decrypting the distance between the first vector and the second vector. The data matching method as described in claim 15, wherein obtaining the target distance of the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes: sending the second vector to the first device. Encrypt the vector so that the first device determines the distance between the encrypted first encryption vector and the second encryption vector based on the second encryption vector and the first encryption vector; receive the first encryption vector sent by the first device The target distance between the vector and the second vector, where the target distance is the encrypted first encryption vector and the first target private key corresponding to the first target public key generated by the first device. The second encrypted vector is obtained by distance decryption. The data matching method as described in claim 15, wherein obtaining the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector includes: using a self-generated second target public fund. Perform homomorphic encryption on the second vector with the key to obtain a third encryption vector, and send the second target public key and the third encryption vector to the first device; receive the information sent by the first device based on the third encryption vector. The encrypted vector and the fourth encrypted vector determine the distance between the encrypted second vector and the first vector, wherein the fourth encrypted vector is generated by homomorphic encryption of the first vector using the second target public key. based on the encrypted distance between the first vector and the second vector and the second target private key corresponding to the second target public key, determine the target distance between the first vector and the second vector, based on the The target distance and the preset first distance threshold determine whether the first data and the second data match. The data matching method as described in claim 15, wherein logging the second data to be matched into a pre-trained vector transformation model and obtaining the second vector corresponding to the second data includes: For each second sub-data in the second data, log the second sub-data into the pre-trained vector transformation model to obtain the second sub-vector corresponding to the second sub-data; wherein, each second sub-data is The lengths of the second sub-vectors corresponding to the sub-data are all the first preset length; the second sub-vectors corresponding to each of the second sub-data are spliced to obtain the second vector corresponding to the second data. The data matching method as described in claim 15 or 22, wherein the lengths of the first vector and the second vector are both second preset lengths. The data matching method of claim 19, wherein determining the distance between the encrypted first vector and the second vector based on the first encrypted vector and the second encrypted vector includes: based on the preset insertion rule and each second component in the second encryption vector, obtain each set of second encryption components and second encryption square components; and according to the preset insertion rule and each first component in the first encryption vector, Obtain each group of the first encrypted component and the first encrypted square component; according to the preset insertion rule, determine the corresponding group of the first encrypted component, the first encrypted square component, the second encrypted component and the second encrypted square component; according to For each group of the first encrypted component, the first encrypted square component, the second encrypted component and the second encrypted square component, each encrypted sub-distance is determined; based on the sum of each sub-distance, the encrypted first vector and the encrypted square component are determined. The distance of the second vector. The data matching method of claim 15, wherein determining whether the first data and the second data match based on the target distance and a preset first distance threshold includes: determining whether the target distance is less than a preset distance A first distance threshold; if yes, it is determined that the first data and the second data match; otherwise, it is determined that the first data and the second data do not match. The data matching method as described in claim 25, wherein after determining that the first data matches the second data, the method further includes: determining whether the target distance is equal to a preset second distance threshold, and if so, determining The first data is the same as the second data. A data matching device, characterized in that it is applied to a first device. The device includes: a first acquisition module for logging the first data to be matched into a pre-trained vector transformation model to obtain the first data. The corresponding first vector; the first processing module is used to perform homomorphic encryption on the first vector using the first target public key generated by itself to generate the first encryption vector, and send the first target public key to The second device; the first acquisition module is also used to acquire the encrypted distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, wherein the second encrypted vector is The first target public key is obtained by homomorphically encrypting the second vector; the second vector is obtained by logging the second data into the pre-trained vector transformation model in the second device; the first Determining module, configured to determine the target of the first vector and the second vector based on the encrypted distance between the first vector and the second vector and the first target private key corresponding to the first target public key. distance, based on the target distance and the preset first distance threshold, determine whether the first data and the second data match; the first processing module is specifically used to target each first component in the first vector , determine the first square component corresponding to the first component; insert the first square component corresponding to each first component into the first vector according to the preset insertion rules, and insert the first square component into the vector obtained updated to this A first vector; performing homomorphic encryption on each first component and each first square component in the first vector based on the first target public key to generate the first encrypted vector. A data matching device, characterized in that it is applied to a second device, and the device includes: a second acquisition module for logging the second data to be matched into a pre-trained vector transformation model to obtain the second data The corresponding second vector; a second processing module configured to receive the first target public key sent by the first device, and use the first target public key to perform homomorphic encryption on the second vector to generate a second encryption vector; The second acquisition module is also used to acquire the target distance between the first vector and the second vector determined based on the first encrypted vector and the second encrypted vector, wherein the first encrypted vector is disclosed using the first target The first vector is obtained by encrypting the first vector with the key, and the first vector is obtained by logging the first data into the pre-trained vector transformation model in the first device; the second determination module is used to base on the first vector The target distance and the preset first distance threshold are used to determine whether the first data and the second data match; the first processing module is specifically used to determine the third component for each second component in the second vector. The second square component corresponding to the two components; insert the second square component corresponding to each second component into the second vector according to the preset insertion rules, and update the vector obtained after inserting the second square component to the second square component. Two vectors; performing homomorphic encryption on each second component and each second square component in the second vector based on the first target public key to generate the second encrypted vector. An electronic device, characterized in that the electronic device includes a processor and a memory, the memory is used to store program instructions, and the processor is used to execute the computer program stored in the memory When implementing the steps of the data matching method as described in any one of claims 1 to 14 or the steps of the data matching method as described in any one of claims 15 to 26. A computer-readable storage medium, characterized in that it stores a computer program that, when executed by a processor, implements the steps of the data matching method as described in any one of claims 1 to 14 or as claimed in claims 15 to 14. The steps of the data matching method described in any one of 26.

Images