CN117272394B

CN117272394B - Bond market data sharing method and device, storage medium and electronic equipment

Info

Publication number: CN117272394B
Application number: CN202311550878.8A
Authority: CN
Inventors: 王延昭; 唐华云; 贾晨; 丁杭超; 吕文哲; 高兰兰; 孙爽
Original assignee: China Bond Jinke Information Technology Co ltd
Current assignee: China Bond Jinke Information Technology Co ltd
Priority date: 2023-11-21
Filing date: 2023-11-21
Publication date: 2024-02-02
Anticipated expiration: 2043-11-21
Also published as: CN117272394A

Abstract

The invention provides a bond market data sharing method and device, a storage medium and electronic equipment, wherein the method comprises the following steps: when a data sharing request is received, determining a sharing task category and request information; determining at least one data provider according to the request information; determining a privacy calculation task and a target public key according to the privacy calculation strategy corresponding to the request information and the sharing task category; the target public key and the privacy calculation task are sent to a data provider, so that the data provider performs task processing based on bond data, and the task processing result is encrypted through the target public key to obtain a secret state task result; carrying out result fusion on the secret state task results of each data provider to obtain target task results; and sending the target task result to the data requiring party so as to obtain the data sharing result. By applying the method, the data demand can be realized by applying the multiparty bond data, which is beneficial to improving the data application effect and mining the data value of the bond market.

Description

Bond market data sharing method and device, storage medium and electronic equipment

Technical Field

The present invention relates to the field of financial science and technology, and in particular, to a method and apparatus for sharing market data of bonds, a storage medium, and an electronic device.

Background

The bond market is one of important components of the financial market, and along with the development of financial science and technology, bond data elements become important data resources gradually, and in various scenes such as financial model training, data analysis and the like, bond data participation is required.

Currently, bond data for each financial institution in the bond market is typically managed by each institution itself, and bond data for each financial institution is not communicated with each other. When the financial institution generates the data application demand, the data of the bonds mastered by the financial institution or the data of the bonds publically issued in the bond market can be acquired only through the information platform of the financial institution, and the data of the bonds mastered by the financial institution or the data publically issued are applied to the demand processing.

In the application scenario of bond data, the requirements of the financial institutions on the data application effect of the bond data are increasing. Based on the existing processing mode, facing the data application requirement, the processing process of the financial institutions can only adopt self-mastered bond data or publicly issued data, so that the requirements of all financial institutions are difficult to meet, and the data value of the bond market is not easy to mine.

Disclosure of Invention

In view of the above, the embodiment of the invention provides a method for sharing data of bond market, so as to solve the problem that a financial institution can only apply self-mastered bond data or public data and is difficult to meet data requirements.

The embodiment of the invention also provides a device for sharing the data of the bond market, which is used for ensuring the practical realization and application of the method.

In order to achieve the above object, the embodiment of the present invention provides the following technical solutions:

a bond market data sharing method, comprising:

when a data sharing request sent by a data requiring party is received, determining a sharing task category and request information corresponding to the data sharing request; the sharing task category characterizes a federation learning task or a data query task;

determining a data provider set according to the request information; the set of data providers includes at least one data provider;

determining a privacy calculation strategy corresponding to the sharing task category;

determining a privacy calculation task and a target public key corresponding to the data sharing request according to the privacy calculation strategy and the request information;

for each data provider, the target public key and the privacy calculation task are sent to the data provider, so that the data provider performs task processing on the privacy calculation task based on bond data owned by the data provider, the task processing result is encrypted through the target public key, and the encrypted result is used as a secret state task result corresponding to the data provider;

Receiving a secret state task result corresponding to each data provider, and carrying out result fusion processing on each secret state task result to obtain a target task result;

and sending the target task result to the data demand party, so that the data demand party determines a data sharing result corresponding to the data sharing request based on the target task result.

In the above method, optionally, if the sharing task class characterizes a federal learning task, determining, according to the privacy calculation policy and the request information, a privacy calculation task and a target public key corresponding to the data sharing request includes:

determining federal learning model information corresponding to the data sharing request based on the request information;

establishing a transverse federal learning model according to a preset federal learning algorithm and federal learning model information;

generating a federation learning model training task corresponding to the transverse federation learning model, and taking the federation learning model training task as the privacy calculation task;

acquiring a first public key carried in the request information, and taking the first public key as the target public key; the first public key is a public key in a first key pair, and the first key pair calls a key pair generated by a preset homomorphic encryption service for the data requiring party.

In the above method, optionally, the process that the data provider performs task processing on the privacy calculation task based on the bond data owned by the data provider, encrypts the task processing result by using the target public key, and uses the encrypted result as a secret state task result corresponding to the data provider includes:

model training is carried out on the transverse federal learning model based on the bond data, and model parameters of the trained transverse federal learning model are used as task processing results of the privacy calculation task;

and encrypting the model parameters through the target public key to obtain the secret state model parameters corresponding to the model parameters, and taking the secret state model parameters as secret state task results corresponding to the data provider.

The above method, optionally, performs a result fusion process on each of the close task results to obtain a target task result, including:

and carrying out parameter aggregation on the secret state model parameters contained in each secret state task result according to a preset federation aggregation strategy to obtain a parameter aggregation result, and taking the parameter aggregation result as the target task result so that the data demand party decrypts the target task result through the private key in the first key pair to obtain a model parameter plaintext corresponding to the parameter aggregation result.

In the above method, optionally, if the sharing task category characterizes the data query task, determining, according to the privacy calculation policy and the request information, the privacy calculation task and the target public key corresponding to the data sharing request includes:

determining a query attribute set and a query object set corresponding to the data sharing request based on the request information; the query attribute set comprises a plurality of query attributes, and the query object set comprises abstract values corresponding to a plurality of target query objects; the plurality of target query objects comprise query objects for the data demander to select a query and various query objects determined based on a preset careless transmission strategy;

generating a combined hidden query task according to the query attribute set and the query object set, and taking the combined hidden query task as the privacy calculation task;

and generating a second key pair according to a preset encryption algorithm, and taking a public key in the second key pair as the target public key.

In the above method, optionally, the task processing is performed on the privacy calculation task by the data provider based on the bond data owned by the data provider, the task processing result is encrypted by the target public key, and the encrypted result is used as a secret state task result corresponding to the data provider, including:

Determining a data object set corresponding to the bond data; the data object set comprises abstract values corresponding to a plurality of data objects;

performing privacy intersection processing on the data object set and the query object set based on a preset privacy intersection algorithm to obtain data objects corresponding to each target query object; the preset privacy intersection algorithm is a privacy intersection algorithm adopting a barrel-division optimization strategy based on editing distance;

for each target query object, determining attribute data corresponding to a data object corresponding to the target query object in the bond data according to the query attribute set, and taking the attribute data as the attribute data corresponding to the target query object;

taking the attribute data corresponding to each target query object as a task processing result of the privacy calculation task;

and encrypting the attribute data corresponding to each target query object through the target public key to obtain the secret attribute data corresponding to each target query object, and taking each secret attribute data as a secret task result corresponding to the data provider.

Decrypting each secret state task result through a private key in the second key pair to obtain a result plaintext corresponding to each secret state task result; the result plaintext corresponding to each secret state task result comprises an attribute data plaintext corresponding to secret state attribute data in the secret state task result;

performing data splicing processing based on the result plaintext to obtain query attribute data corresponding to each target query object;

and carrying out data processing on query attribute data corresponding to each target query object based on the careless transmission strategy to obtain careless transmission data corresponding to each target query object, and taking the careless transmission data corresponding to each target query object as the target task result so that the data demander can obtain the query attribute data corresponding to the query object selected by the data demander from the target task result based on the careless transmission strategy.

A bond market data sharing apparatus comprising:

the first determining unit is used for determining a sharing task category and request information corresponding to a data sharing request when the data sharing request sent by a data requiring party is received; the sharing task category characterizes a federation learning task or a data query task;

A second determining unit, configured to determine a data provider set according to the request information; the set of data providers includes at least one data provider;

the third determining unit is used for determining a privacy calculation strategy corresponding to the sharing task category;

a fourth determining unit, configured to determine, according to the privacy calculation policy and the request information, a privacy calculation task and a target public key corresponding to the data sharing request;

the first sending unit is used for sending the target public key and the privacy calculation task to each data provider, so that the data provider performs task processing on the privacy calculation task based on bond data owned by the data provider, encrypts a task processing result through the target public key, and takes the encrypted result as a secret state task result corresponding to the data provider;

the receiving unit is used for receiving the secret state task results corresponding to each data provider and carrying out result fusion processing on each secret state task result to obtain a target task result;

and the second sending unit is used for sending the target task result to the data demand party, so that the data demand party determines a data sharing result corresponding to the data sharing request based on the target task result.

A storage medium comprising stored instructions, wherein the instructions, when executed, control a device on which the storage medium resides to perform a bond market data sharing method as described above.

An electronic device comprising a memory, and one or more instructions, wherein the one or more instructions are stored in the memory and configured to be executed by one or more processors as the bond market data sharing method described above.

Based on the above-mentioned method for sharing market data of bonds provided by the embodiment of the present invention, the method includes: when a data sharing request sent by a data requiring party is received, determining a sharing task category and request information corresponding to the data sharing request; the shared task category characterizes a federal learning task or a data query task; determining a data provider set comprising at least one data provider according to the request information; determining a privacy calculation strategy corresponding to the sharing task category; determining a privacy calculation task and a target public key corresponding to the data sharing request according to the privacy calculation strategy and the request information; for each data provider, sending a target public key and a privacy calculation task to the data provider, enabling the data provider to perform task processing on the privacy calculation task based on bond data owned by the data provider, performing encryption processing on a task processing result through the target public key, and taking the encryption result as a secret state task result corresponding to the data provider; receiving a secret state task result corresponding to each data provider, and carrying out result fusion processing on each secret state task result to obtain a target task result; and sending the target task result to the data demand party, so that the data demand party determines a data sharing result corresponding to the data sharing request based on the target task result. By applying the method provided by the embodiment of the invention, a financial institution in the bond market can be used as a data demand party to initiate a data sharing request of a federal learning task or a data query task, and can be matched with a corresponding privacy calculation strategy in response to the data sharing request, so that the privacy calculation task is established, and the data provider can execute the privacy calculation task based on the bond data of the data provider so as to obtain a data sharing result. The data demand party can apply the bond data of other service participants to realize the data demand of the federal learning task or the data query task, thereby being beneficial to improving the processing effect of the data application task of the financial institution, meeting the data demand of the financial institution and mining the data value of the bond market. And secondly, the data sharing process is realized based on a privacy computing technology and an encryption technology, the bond data of the data provider cannot be locally output, the data demand party can only obtain the required data sharing result and can not obtain all the original data participating in task processing, and the data sharing can be realized under the condition of guaranteeing the security of the bond data.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present invention, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flow chart of a method for sharing data of bond market according to an embodiment of the present invention;

FIG. 2 is a flowchart of another method for sharing data of bond market according to an embodiment of the present invention;

FIG. 3 is a flowchart illustrating another method of the present invention for sharing bond market data;

fig. 4 is a schematic structural diagram of a bond market data sharing device according to an embodiment of the present invention;

fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

In this application, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

The embodiment of the invention provides a data sharing method for a bond market, which can be applied to a data sharing system facing the bond market, wherein an execution subject of the method can be a server of the system, and a flow chart of the method is shown in fig. 1, and comprises the following steps:

s101: when a data sharing request sent by a data requiring party is received, determining a sharing task category and request information corresponding to the data sharing request; the sharing task category characterizes a federation learning task or a data query task;

the method provided by the embodiment of the invention can be applied to a pre-deployed data sharing system, the data sharing system can be realized based on a blockchain network, and each financial institution can be used as a service participant to deploy corresponding nodes to participate in data sharing service. The service participant who can provide the bond data for data sharing can register the bond data owned by the service participant in the data sharing service on demand in advance, such as registering which types of data they own in a contracted form. The bond data mentioned in the embodiments of the present invention refers to information data associated with a bond market, including basic data information of bonds, such as bond issuers, bond issue dates, bond issue amounts, bond types, and the like, and also includes data information related to bonds, such as judicial, business, tax, public opinion information, and the like of bond issuers.

The method provided by the embodiment of the invention can realize the processing of two types of data sharing tasks, one type is a federal learning task and the other type is a data query task. Federal learning is a distributed learning technology combining traditional cryptography and machine learning, and aims to perform joint model training on the premise of meeting data privacy safety for data providers who are reluctant or unable to expose plaintext data of themselves. The data inquiry is to inquire the needed data content in the bond data owned by each service participant.

When the financial institution has the data requirement of the federal learning task or the data query task, the financial institution can be used as a data requirement party, and corresponding request data is input through the corresponding node client side, and a corresponding data sharing request is sent to the server side.

When the server receives a data sharing request sent by a data requiring party, the data sharing request can be analyzed to obtain the sharing task category and the request information corresponding to the request. The shared task category, i.e. the task category requested to be executed by the data sharing request, characterizes the federation learning task or the data query task in the embodiment of the present invention. The request information is task demand information corresponding to the data sharing request, and specific information content corresponds to the sharing task category. For example, if the shared task class is a federal learning task, the request information may include data such as an attribute of training data, a model output attribute, and a training target, and if the shared task class is a data query task, the request information may include data such as a query object, a query attribute, and the like, for example, query judicial information and public opinion information of a bond issuer, the query object may be a bond identifier (such as a bond code) that characterizes the bond, and the query attribute may be a data attribute that characterizes judicial information of the issuer and public opinion information of the issuer.

S102: determining a data provider set according to the request information; the set of data providers includes at least one data provider;

in the method provided by the embodiment of the invention, the server side can determine the data requirement of the data sharing request, namely which types of data need to be applied according to the request information. Based on the data registration information of the service participants providing the bond data at the service end, among the service participants, the service participant whose bond data resource matches the data requirement of the current data sharing request can be determined, and the matched service participant is regarded as the data provider of the current data sharing request, so that the data provider set can be determined. For example, currently, the judicial information and public opinion information of a bond issuer need to be queried, the service participant a owns judicial information of the bond issuer, the service participant B owns tax information of the bond issuer, the service participant C owns business information of the bond issuer, the service participant D owns public opinion information of the bond issuer, and by matching the service participant data resources with the data demands of the data demander, the service participant a and the service participant D can be determined to be data providers respectively, and the service participant a and the service participant D form a data provider set.

S103: determining a privacy calculation strategy corresponding to the sharing task category;

in the method provided by the embodiment of the invention, privacy calculation strategies corresponding to various data sharing tasks are preset. The privacy calculation is a technology integrating cryptography, statistics, artificial intelligence, big data, a computer system and other core technologies, and can effectively mine the value of the data on the premise of not invading the safety and privacy of the data, support the trusted sharing and safe circulation of the data and realize the 'data availability is invisible'. The privacy calculation strategy adopted by the embodiment of the invention can be realized based on the existing privacy calculation technology, and different privacy calculation strategies can be adopted for different types of tasks. For example, for the federal learning task, the homomorphic encryption technology and the federal learning technology can be adopted to realize the privacy calculation, and for the data query task, the careless transmission technology and the privacy intersection technology can be adopted to realize the privacy calculation.

In the method provided by the embodiment of the invention, the privacy calculation strategy corresponding to the current sharing task category can be determined based on the pre-configured strategy information.

S104: determining a privacy calculation task and a target public key corresponding to the data sharing request according to the privacy calculation strategy and the request information;

In the method provided by the embodiment of the invention, the privacy calculation task corresponding to the current data sharing request can be generated based on the privacy calculation strategy and the request information corresponding to the current sharing task category. In the method provided by the embodiment of the invention, each data provider adopts encryption transmission when feeding back the task result to the server, the encryption strategy of the process corresponds to the task type, the encryption mode of the data feedback process can be defined in the privacy calculation strategy, and the server determines the public key, namely the target public key, of the data provider for encrypting the task processing result according to the encryption mode defined in the privacy calculation strategy. The target public key can be generated by a server side or a data demand party and is determined by actual task demands.

S105: for each data provider, the target public key and the privacy calculation task are sent to the data provider, so that the data provider performs task processing on the privacy calculation task based on bond data owned by the data provider, the task processing result is encrypted through the target public key, and the encrypted result is used as a secret state task result corresponding to the data provider;

in the method provided by the embodiment of the invention, the privacy calculation task and the target public key can be respectively sent to each data provider, and it can be understood that the privacy calculation task and the target public key are specifically sent to the node corresponding to each data provider, and the task processing process of the data provider is also the operation performed on the corresponding node. After the data provider receives the privacy calculation task and the target public key, task processing can be performed based on the corresponding bond data to obtain a task processing result, then the task processing result is encrypted by the target public key, and the encrypted task processing result is used as a secret state task result of the data provider and is sent to the server.

S106: receiving a secret state task result corresponding to each data provider, and carrying out result fusion processing on each secret state task result to obtain a target task result;

in the method provided by the embodiment of the invention, the result fusion mode corresponding to various data sharing tasks can be preset. After the server receives the secret state task results corresponding to all the data providers, the result fusion processing is carried out on the secret state task results corresponding to all the data providers, and the fusion processing result is the target task result. The result fusion processing mode corresponds to the sharing task category, and the result fusion processing modes corresponding to the federal learning task and the data query task can be different.

S107: and sending the target task result to the data demand party, so that the data demand party determines a data sharing result corresponding to the data sharing request based on the target task result.

In the method provided by the embodiment of the invention, the target task result is used as feedback data and is sent to the data demand side, and the data demand side can perform data processing on the target task result according to the data reading mode corresponding to the privacy calculation strategy corresponding to the sharing task category, so that the data sharing result corresponding to the data sharing request, namely the data required by the data demand side, is obtained.

Based on the method provided by the embodiment of the invention, when a data sharing request sent by a data demand party is received, the sharing task category and request information corresponding to the data sharing request are determined; the shared task category characterizes a federal learning task or a data query task; determining a data provider set comprising at least one data provider according to the request information; determining a privacy calculation strategy corresponding to the sharing task category; determining a privacy calculation task and a target public key according to the privacy calculation strategy and the request information; the target public key and the privacy calculation task are sent to the data provider, so that the data provider performs task processing on the privacy calculation task based on bond data owned by the data provider, the task processing result is encrypted through the target public key, and the encrypted result is used as a secret state task result corresponding to the data provider; receiving each secret state task result, and carrying out result fusion processing on each secret state task result to obtain a target task result; and sending the target task result to the data demand party, so that the data demand party determines a data sharing result based on the target task result. By applying the method provided by the embodiment of the invention, a financial institution in the bond market can be used as a data demand party to initiate a data sharing request of a federal learning task or a data query task, and can be matched with a corresponding privacy calculation strategy in response to the data sharing request, so that the privacy calculation task is established, and the data provider can execute the privacy calculation task based on the bond data of the data provider so as to obtain a data sharing result. The data demand party can apply the bond data of other service participants to realize the data demand of the federal learning task or the data query task, thereby being beneficial to improving the processing effect of the data application task of the financial institution, meeting the data demand of the financial institution and mining the data value of the bond market. And secondly, the data sharing process is realized based on a privacy computing technology and an encryption technology, the bond data of the data provider cannot be locally output, the data demand party can only obtain the required data sharing result and can not obtain all the original data participating in task processing, and the data sharing can be realized under the condition of guaranteeing the security of the bond data.

On the basis of the method shown in fig. 1 and in combination with the flowchart shown in fig. 2, the embodiment of the present invention provides a further method for sharing data in bond market, where in the method provided by the embodiment of the present invention, the sharing task category characterizes a federal learning task, and the process mentioned in step S104 of determining, according to the privacy calculation policy and the request information, a privacy calculation task and a target public key corresponding to the data sharing request includes:

s201: determining federal learning model information corresponding to the data sharing request based on the request information;

in the method provided by the embodiment of the invention, the sharing task category corresponding to the data sharing request characterizes the federation learning task, namely, the data sharing task corresponding to the data sharing request is the federation learning task. The request information contains data related to a model which needs to be trained by the data demand party through federal learning, such as data of training data attributes, model output attributes, model training algorithm types and the like. In the request processing process, relevant data about model training can be obtained from the request information to obtain federal learning model information.

S202: establishing a transverse federal learning model according to a preset federal learning algorithm and federal learning model information;

In the method provided by the embodiment of the invention, the server side can be preset with a federal learning algorithm, such as a federal convolutional neural network algorithm, a federal deep neural network algorithm, a federal decision tree algorithm and the like. In the request processing process, model initialization can be carried out according to a preset federal learning algorithm and federal learning model information, and a transverse federal learning model is established.

S203: generating a federation learning model training task corresponding to the transverse federation learning model, and taking the federation learning model training task as the privacy calculation task;

in the method provided by the embodiment of the invention, a federal learning model training task for indicating to train the transverse federal learning model is generated, and the federal learning model training task is the current privacy calculation task.

S204: acquiring a first public key carried in the request information, and taking the first public key as the target public key; the first public key is a public key in a first key pair, and the first key pair calls a key pair generated by a preset homomorphic encryption service for the data requiring party.

In the method provided by the embodiment of the invention, the federal learning task performs privacy calculation based on homomorphic encryption technology. When a data demand side needs to initiate a data sharing request of a federal learning task, the data demand side needs to call a preset homomorphic encryption service to generate a group of key pairs, and a public key of the key pairs is sent to a server side through the data sharing request so as to facilitate the homomorphic encryption processing of task data. When the server side performs request processing, the public key generated by the data requiring party can be obtained from the request information, and the public key is used as a target public key. Homomorphic encryption is a cryptographic technique that performs operations on encrypted data so as to complete various computing tasks on the premise of ensuring user data security. The encrypted data ciphertext can be directly processed by using functions or ciphertext state mapping based on homomorphic encryption technology to obtain a processing result in a secret state, and the processing result in the secret state is the same as the result obtained by directly processing the original data plaintext after decryption, so that the user data privacy is ensured.

Based on the method provided by the embodiment of the invention, when the data demand side initiates the data sharing request of the federal learning task, a transverse federal learning model can be established based on a preset federal learning algorithm to generate a corresponding federal learning model training task, so that the data demand side can apply bond data of all data providers to carry out joint model training, and the model training requirement of the data demand side is met. The federal learning supports joint modeling of multiple data providers, model joint training is completed while data plaintext is not exposed, the data providers only provide model parameters obtained through training to a server, and a homomorphic encryption technology is used, and intermediate parameters are encrypted by using a public key of a data demander, so that parameters in a modeling process can be prevented from being leaked, and the safety of original data is guaranteed.

On the basis of the method provided by the above embodiment, in the method provided by the embodiment of the present invention, the process of performing task processing on the privacy calculation task by the data provider in step S105 based on the bond data owned by the data provider, performing encryption processing on the task processing result by the target public key, and taking the encryption result as the secret state task result corresponding to the data provider includes:

in the method provided by the embodiment of the invention, under the condition that the shared task category characterizes the federation learning task, the privacy calculation task received by the data provider is the federation learning model training task. When the data provider receives the federation learning model training task, based on task requirements, the local bond data of the data provider is used for carrying out model training on the transverse federation learning model, and a federation learning algorithm adopted by the data provider for carrying out model training is the same as a federation learning algorithm adopted by the server for establishing the transverse federation learning model. And after the data provider finishes the model training process, taking the model parameters of the trained transverse federal learning model as the task processing result of the current privacy calculation task.

In the method provided by the embodiment of the invention, the data provider takes the target public key as the encryption key, and encrypts the model parameters of the trained transverse federal learning model to obtain the secret model parameters, wherein the secret model parameters are secret task results corresponding to the current data provider, namely the task processing results which need to be fed back to the server by the current data provider.

Based on the method provided by the above embodiment, in the method provided by the embodiment of the present invention, the process of performing the result fusion processing on each of the close task results mentioned in step S106 to obtain the target task result includes:

In the method provided by the embodiment of the invention, the federation aggregation strategy corresponding to the federation learning algorithm is preset and is used for aggregating the federation learning processing results fed back by each data provider. Specifically, the federation policy may use existing federation algorithms such as a federation average algorithm or an optimization algorithm thereof. In the request processing process, after receiving the secret state task results corresponding to all the data providers, the server performs parameter aggregation on secret state model parameters in all the secret state task results to obtain parameter aggregation results, namely aggregated secret state model parameters, and sends the parameter aggregation results to the data demander as target task results.

When the data demand side receives a target task result, namely a parameter aggregation result, the data demand side calls a private key in a secret key pair generated by homomorphic encryption service in advance, and decrypts the parameter aggregation result to obtain a model parameter plaintext, wherein the model parameter plaintext is an aggregation result of model parameters trained by each data provider. The data demander can construct the service model needed by the data demander based on the model parameter plaintext.

Further, in the practical application process, the model training process based on federal learning may need to be trained for multiple times, when the data demand party obtains the current model parameter plaintext, if the overall model training process is not finished, the data demand party can trigger the next training based on the current model parameter plaintext, the data demand party can send the current model parameter plaintext to the server, so that the server can build a transverse federal learning model to be trained again according to the model parameter plaintext, and trigger each data provider to continue model training on the basis of the current transverse federal learning model, so as to obtain new model parameters until the data demand party obtains the model parameter plaintext meeting the training requirement.

In order to better illustrate the method provided by the embodiment of the invention, the data sharing scene of the bang learning task is further illustrated on the basis of the method provided by the embodiment. In the method provided by the embodiment of the invention, the realization process of data sharing of the federal learning task mainly comprises the following steps:

each data provider deploys distributed nodes for horizontal federation learning, takes data owned by the distributed nodes as an identification by using a data sample ID, registers the data sample ID in a data sharing service, calculates a data sample ID abstract by using an SM3 algorithm when registering the data sample ID, and then imports the abstract.

And the data demand side initiates a data sharing request of the federal learning task and sends a public key generated by calling the homomorphic encryption service to the server. The server determines each data provider participating in the current federal learning, collects and deduplicates the data sample ID abstracts imported by the data providers, and confirms the total amount of federal data samples used for the current model training.

The server initializes a transverse federal learning model, and specifically, federal learning algorithms such as federal convolutional neural networks, federal deep neural networks, federal decision trees and the like can be adopted.

And the server starts a training task and sends the public key of the data requiring party and the transverse federal learning model to each data provider.

The data provider performs model training based on local bond data, encrypts generated model parameters with a public key of the data demand party after the current training process is completed, and sends the encrypted model parameters to the server.

After receiving the encrypted model parameters of all data providers, the server uses a federal aggregation method to aggregate the parameters, the aggregation mode generally corresponds to the federal learning algorithm, for example, federal average aggregation can be used to add and average all the encrypted model parameters, and because the model parameters use homomorphic encryption, the method supports addition and multiplication under ciphertext, and can directly calculate the encrypted model parameters.

And the server side sends the model parameter ciphertext after federation aggregation to a data demand party.

The data demand side uses a private key generated by calling homomorphic encryption service in advance to decrypt the model parameter ciphertext, and at the moment, if training is not finished (federal learning usually requires multiple training rounds, depending on the data amount of the data provider and the data amount used in each training round), the model parameter plaintext is sent to the service side, the service side sends the model parameter plaintext to all the data providers again through tasks, the data provider executes the training tasks again on the basis of the model parameters, and then the data demand side can obtain the model parameters obtained by aggregation after a new training round. At this time, if training is finished, the data demander can obtain a required service model for own service.

The homomorphic encryption service in the method provided by the embodiment of the invention can select to use CKKS, DGK, paillier and other existing homomorphic encryption frameworks. The various frames have the advantages and can be set as required. For example, the CKS framework is based on a lattice password, so that the anti-quantum attack performance is achieved, and the encryption and decryption efficiency of the DGK framework is higher. Different frameworks have different supporting strengths for homomorphic addition and multiplication, so that the effect of homomorphic encryption service is also related to what parameter aggregation mode is used by the federal learning algorithm, and the homomorphic encryption framework can be selected in a self-adaptive mode according to the federal learning algorithm selected by the data demand side.

On the basis of the method shown in fig. 1 and in combination with the flowchart shown in fig. 3, the embodiment of the present invention provides a further method for sharing bond market data, where in the method provided by the embodiment of the present invention, the sharing task category characterizes a data query task, and the process mentioned in step S104 of determining, according to the privacy calculation policy and the request information, a privacy calculation task and a target public key corresponding to the data sharing request includes:

s301: determining a query attribute set and a query object set corresponding to the data sharing request based on the request information; the query attribute set comprises a plurality of query attributes, and the query object set comprises abstract values corresponding to a plurality of target query objects;

In the method provided by the embodiment of the invention, the sharing task category characterizes the data query task, namely the data sharing task corresponding to the data sharing request is the data query task. The processing of the data query task is realized based on an careless transmission technology, the careless transmission is a multiparty security calculation technology, the technology is based on a ciphertext data transmission means that different public keys are used for encrypting the data of a sender, and a receiver only has a private key of the data required by the receiver, so that the sender does not know the specific information wanted by the receiver, the receiver does not know the plaintext of other information of the sender, and the query intention of the sender and the data security of the receiver are simultaneously ensured in the process. That is, the data query provided by the embodiment of the invention is a joint hidden query, the server side and each data provider cannot know the object actually queried by the data demander, and the data demander can only acquire the related data of the actual query object.

When a data-sharing request of a data query task is required to be initiated by a data-requesting party, the data-requesting party needs to input an object actually queried (i.e. a query condition, such as a code of a certain bond), query attributes (i.e. query contents, i.e. query data of which aspect, such as bond basic data, company judicial, company tax, industry and commerce, public opinion, etc.), and a security coefficient of unintentional transmission, and call a preset unintentional transmission service, and among the objects of the same type, a plurality of query objects for unintentional transmission are determined, and an object with higher similarity to the actual query object is usually selected, for example, a plurality of bond codes are selected among all bond codes. Specifically, each query object for unintentional transmission may be determined based on an Edit Distance (Edit Distance) algorithm, such as measuring similarity between candidate objects and actual query objects using an Edit Distance algorithm based on a levenstein Distance (Levenshtein Distance).

The client may perform hash operation on the actual query object and the query object determined by invoking the careless transmission service through a preset hash algorithm (for example, an SM3 cryptographic algorithm or a hash function and other existing cryptographic algorithms) to obtain a digest value corresponding to each query object. And taking the actual query object and each query object determined based on the careless transmission service as target query objects, and generating a data sharing request based on the abstract values, the query attributes and other data corresponding to all the target query objects.

In this scenario, the request information corresponding to the data sharing request includes abstract values corresponding to the plurality of target query objects and each query attribute. Through information extraction, a query attribute set and a query object set corresponding to the data sharing request can be obtained. The target query objects comprise the query object of the data demander selection query and the query object determined based on a preset careless transmission strategy.

S302: generating a combined hidden query task according to the query attribute set and the query object set, and taking the combined hidden query task as the privacy calculation task;

in the method provided by the embodiment of the invention, the summary values corresponding to all the target query objects in the query object set are used as query conditions, and all the query attributes in the query attribute set are used as query contents, so that a joint hidden query task is generated, namely, attribute data related to each query attribute is indicated to query each target query object. The combined hidden inquiry task is the privacy calculation task of this time.

S303: and generating a second key pair according to a preset encryption algorithm, and taking a public key in the second key pair as the target public key.

In the method provided by the embodiment of the invention, in the processing process of the joint hiding query task, the server generates the encryption key so that each data provider encrypts the task result fed back by the data provider. The server generates a key pair based on a preset encryption algorithm, and takes a public key in the generated key pair as a target public key. The encryption algorithm adopted by the server side can be an existing encryption algorithm such as SM2 national encryption algorithm.

Based on the method provided by the embodiment of the invention, when the data demand side needs to initiate the data sharing request of the data query task, the joint hidden query task can be generated based on the careless transmission service, so that the data demand side can query the required data under the condition that each data provider can not learn the real query intention of the data demand side, and meanwhile, the data demand side can not obtain the data outside the actual query content of the data demand side.

in the method provided by the embodiment of the invention, when the data provider receives the joint hidden query task, the abstract value corresponding to each target query object of the current required query can be obtained according to the task information of the joint hidden query task. The data provider can determine the abstract value corresponding to each data object corresponding to the bond data owned by the data provider according to the bond data resource owned by the data provider, and a data object set is obtained. For example, the bond code is taken as an object, and the data object set corresponding to the bond data owned by the data provider is the abstract value corresponding to all bond codes owned by the data provider and relevant bond data.

in the method provided by the embodiment of the invention, the data provider is pre-deployed with a privacy intersection algorithm of a barrel-division optimization strategy based on the editing distance. In the processing process of the joint hidden query task, two-party privacy intersection processing is performed with the server based on a privacy intersection algorithm, namely, privacy intersection processing is performed on a data object set corresponding to the data provider and a query object set, so that which data objects need to be queried are determined in bond data owned by the data provider, and accordingly, the data object corresponding to each target query object can be obtained.

in the method provided by the embodiment of the invention, the data provider can determine the data attribute which needs to be queried currently based on the query attribute set according to the type of the resource owned by the data provider. The data provider can acquire attribute data corresponding to the data attribute of the current query of each target query object by taking the data object corresponding to the target query object as an association in the bond data, and take the acquired attribute data as the attribute data corresponding to the target query object.

Specifically, the data query mode of the privacy intersection algorithm based on the bucket optimization strategy of the edit distance is that the digest value corresponding to one target query object is selected from the query object set at will, the edit distance between the digest value and the digest value corresponding to each data object is calculated, the digest values corresponding to each data object are classified according to the order of the edit distance from small to large, and a plurality of buckets are obtained, wherein the size of each bucket is the same as the number of the digest values in the query object set. The data object with the edit distance of zero between the abstract value and the abstract value corresponding to the current target query object is the data object matched with the current target query object, and the corresponding data can be obtained from the bond data. And then, for each other target query object, respectively performing editing distance calculation and comparison from the first bucket to find a data object with the abstract value editing distance of zero with the target query object, and if the data object is not found in the current bucket, performing calculation and comparison to the next bucket until the data object corresponding to the target query object is found.

In the method provided by the embodiment of the invention, the task processing result is composed of the attribute data corresponding to each target query object. And taking the target public key as an encryption key, carrying out data encryption processing on the attribute data corresponding to each target query object, wherein the encryption result of the attribute data of each target query object is the secret state attribute data corresponding to the target query object. And forming a secret state task result corresponding to the current data provider by all secret state attribute data.

Based on the method provided by the embodiment of the invention, the data query is realized through the privacy intersection technology, and the data provider does not need to worry about extra leakage of local data, thereby being beneficial to guaranteeing the data security. Secondly, in the privacy intersection process, a barrel-division optimization strategy based on the editing distance is adopted, so that the calculation cost of the privacy intersection can be reduced, and the system performance is improved.

in the method provided by the embodiment of the invention, when the data sharing request of the data query task is processed, the result of the secret state task received by the server is the secret state attribute data of each target query object determined by the corresponding data provider. After receiving all the secret state task results, the server decrypts each secret state attribute data in each secret state task result by taking a private key in a pre-generated key pair as a decryption key, and obtains an attribute data plaintext corresponding to the secret state attribute data in each secret state task result, namely, attribute data corresponding to each target query object obtained by each data provider in the bond data.

in the method provided by the embodiment of the invention, the server obtains the plaintext of each result through decryption, namely the attribute data corresponding to each target query object queried by each data provider. And representing the target query object by using the abstract value corresponding to the target query object, splicing all attribute data according to the target query object, namely splicing and integrating the attribute data corresponding to the target query object in all result text for each target query object, wherein the splicing result is the query attribute data corresponding to the target query object. For example, each target query object is each bond, and the query attributes are bond basic data, judicial data of a bond issuer, tax data of the bond issuer, business data of the bond issuer, and public opinion data of the bond issuer. The data provider A provides the basic data of the bonds, the data provider B provides the business data and tax data of the issuing mechanism of each bond, the data provider C provides the judicial data of the issuing mechanism of each bond, and the data provider D provides the public opinion data of the issuing mechanism of each bond. And the server extracts the corresponding basic data of the bond, the business data of the issuing organization, the tax data, the judicial data and the public opinion data for each bond, and performs data splicing on the data to obtain the query attribute data corresponding to the bond.

In the method provided by the embodiment of the invention, the data demand side and the service side feed back query results based on the careless transmission strategy, so that after the service side obtains query attribute data corresponding to all target query objects, the query attribute data corresponding to all target query objects are processed according to the preset careless transmission strategy to obtain data results for careless transmission, and the data results are used as target task results to feed back the data demand side. After receiving the target task result, the data demand side extracts data according to a preset careless transmission strategy, and based on the principle of the careless transmission technology, the data demand side can only obtain query attribute data corresponding to the query object actually queried by the data demand side, and cannot obtain actual plaintext for query attribute data corresponding to other query objects.

In order to better illustrate the method provided by the embodiment of the invention, the data sharing scene of the data query task is further illustrated on the basis of the method provided by the embodiment. In the method provided by the embodiment of the invention, the realization process of data sharing of the data query task mainly comprises the following steps:

the service end places the query service based on the unintentional transport technology. The service participants who can provide bond data to participate in data sharing send own data content category plaintext to the service side in advance, so that the service side knows which service participant should be found as the data provider when inquiring certain data. Meanwhile, each service participant who can provide bond data uses SM3 cryptographic algorithm to generate a digest (other hash algorithms such as hash function can be selected) of all data objects (usually ID of some bond market related data) available for query.

The data demand side initiates a data sharing request of a data query task through a client, and specifically comprises the following steps:

the data demander inputs a query object, query content (namely data attribute to be queried) and a security coefficient N which is transmitted carelessly, wherein the query object is an ID (identity) of certain bond market related data, such as bond codes, and the ID is usually public information of the whole market;

The method comprises the steps of generating a digest (namely a digest value) of a query object of a data demand party in a client by adopting an SM3 cryptographic algorithm (other hash algorithms such as hash functions can be selected as well) which is the same as the algorithm adopted by a service participant to generate the digest of the data object, and calculating the Leventan distance between the digest and the digest (such as all bond codes) of all similar condition objects by using an edit distance algorithm, namely the minimum edit operation number required by converting one character string into the other character string. Selecting 2N abstract values with the closest editing distance, randomly taking N-1 abstract values, mixing the abstract values with the real query object abstract, finally transmitting the N abstract values to a server together, and transmitting the query content of the plaintext to the server.

After receiving the data sharing request, the server determines all data providers participating in the current sharing process, and performs joint hidden query with each data provider to obtain a query result. Meanwhile, after receiving the data sharing request, the server side generates N public and private key pairs by using an SM9 cryptographic algorithm, each public and private key pair corresponds to a query object abstract value, and the N public keys are sent to the data requiring party.

The data demand party generates a random number x, encrypts the random number by using a public key corresponding to the real query object abstract in the N public keys, and sends the ciphertext result of the random number to the server.

After receiving the ciphertext result, the server decrypts the ciphertext by using the N private keys respectively to obtain N decrypted results. When the server receives the data fed back by all the data providers, and after the data are processed to obtain the query results corresponding to each query object, exclusive-or processing is carried out on the decryption results and the corresponding query results, and the N query results after exclusive-or processing are sent to the data demander.

The data demander uses the random number x to carry out exclusive or operation with the received N inquiry results, wherein only one piece of data can be decrypted into real data, namely the data actually inquired by the data demander, and the rest decrypted data are still random numbers.

In the method provided by the embodiment of the invention, the joint hiding query process of each data provider mainly comprises the following steps:

after receiving query contents of N query object abstracts and plaintext, which are close in editing distance, sent by a data demand party, a server confirms which data providers need to participate in the query service according to the query contents, then generates a public-private key pair by using a national encryption algorithm SM2, sends the public key to all the data providers participating in the service, and then respectively performs two-party privacy exchange with the data providers, wherein the specific flow is as follows:

The server side sends N query object abstracts to each data provider;

each data provider selects one from N inquiry object abstracts, calculates the editing distance between the data provider and the abstracts of all data objects (for example, the data provider A has a plurality of pieces of data, and performs editing distance calculation with the abstracts of a certain inquiry object provided by a data demander respectively according to the SM3 abstracts of bond codes of the data);

each data provider sorts according to the editing distance from small to large, and takes N as the size of a barrel to divide the data object abstracts of all data of the data provider into barrels, namely, the first barrel stores N data object abstracts with the smallest editing distance, the second barrel stores N data object abstracts with the second smallest editing distance, and so on. And comparing the abstract of the query object with the abstract value of the data object in the barrel in sequence, wherein the abstract of the data object with the distance of 0 is the object to be queried, and the related data of the object is recorded. Comparing the edit distances of the rest N-1 query object abstracts in parallel from the first barrel, finding a value with the distance of 0, if no matched data exists in the first barrel, searching in the second barrel until the matched object is found, and recording the corresponding data;

After the data provider finds all the data, N pieces of data corresponding to the query object abstracts are shared, the data are encrypted by using the SM2 public key sent by the server, and the encrypted data are sent back to the server.

After receiving the results returned by all the data providers, the server decrypts the results by using the private key, splices all the results according to the query objects (for example, splices the basic data of the bonds, the judicial data, tax data, business data, public opinion data and the like of the issuing institution into the same line corresponding to each bond), obtains query results corresponding to all the query objects, and feeds back the results to the data demander through an careless transmission technology based on the query results.

Based on the method provided by each embodiment, the embodiment of the invention provides a data sharing method based on technologies of federal learning, homomorphic encryption, careless transmission, privacy asking for exchange and the like, which is used for supporting a data sharing process of a bond market data service system. In the practical application scenario of the method provided by the embodiment of the invention, a service provider deploys homomorphic encryption service, federal learning service, carelessly transmitted service and privacy exchange service, and provides deployment materials required by using the above services, such as installation packages, deployment documents and the like, for service participants. The data provider deploys distributed nodes of federal learning and privacy trading using the materials described above.

The method provided by the embodiment of the invention can realize two types of data sharing services: first, support the service participant to carry on the joint modeling based on federal learning and homomorphic encryption. For the scenes that the data are distributed on different data providers, model joint training can be carried out by using federal learning and homomorphic encryption technology under the condition that original data are not local, and the market data value of bonds can be exerted on the premise that relevant laws and regulations of the data are met and the privacy safety of user data is guaranteed; and secondly, supporting the service participants to perform joint hidden inquiry based on careless transmission and privacy interaction. For the data query scenario where data are distributed in different data providers (for example, a data demander wants to query self data of a bond and related information of an issuer thereof, and the basic data of the bond, judicial, tax, business, public opinion of a company and the like belong to different data providers respectively), in general, the data provider does not want to expose self data, and the data demander does not want to expose query intention, the method provided by the embodiment of the invention performs joint hidden query involving multiple participants based on careless transmission and privacy finding technology. The inquirer can obtain the inquiring result of the actual inquiring from the inquirer, but cannot obtain other data results, and the server cannot know which one of the whole inquiring results is actually inquired by the inquirer. The whole data sharing service flow can meet the requirements of the bond market on the combination of multiple mechanisms to develop business under the condition of meeting the requirements of relevant laws and regulations and data privacy protection regulations, is favorable for meeting the requirements of the bond market on data element sharing service, can support the value of multi-party institutions for mining multi-source data, and establishes a machine learning model in a combined mode or develops the combined hidden inquiry of multiple participants under the condition that the data of all parties are not revealed, so that the foundation is laid for novel data service of the bond market in terms of breakthrough of the technical aspect.

Corresponding to the method for sharing the bond market data shown in fig. 1, the embodiment of the invention also provides a device for sharing the bond market data, which is used for realizing the method shown in fig. 1, and the structure schematic diagram is shown in fig. 4, and comprises the following steps:

a first determining unit 401, configured to determine, when a data sharing request sent by a data demander is received, a sharing task class and request information corresponding to the data sharing request; the sharing task category characterizes a federation learning task or a data query task;

a second determining unit 402, configured to determine a set of data providers according to the request information; the set of data providers includes at least one data provider;

a third determining unit 403, configured to determine a privacy calculation policy corresponding to the sharing task category;

a fourth determining unit 404, configured to determine, according to the privacy calculation policy and the request information, a privacy calculation task and a target public key corresponding to the data sharing request;

a first sending unit 405, configured to send, for each data provider, the target public key and the privacy calculation task to the data provider, so that the data provider performs task processing on the privacy calculation task based on bond data owned by the data provider, and performs encryption processing on a task processing result through the target public key, and uses the encryption result as a secret task result corresponding to the data provider;

The receiving unit 406 is configured to receive a secret task result corresponding to each data provider, and perform a result fusion process on each secret task result to obtain a target task result;

the second sending unit 407 is configured to send the target task result to the data demander, so that the data demander determines a data sharing result corresponding to the data sharing request based on the target task result.

By applying the device provided by the embodiment of the invention, a financial institution in the bond market can be used as a data demand party to initiate a data sharing request of a federal learning task or a data query task, and can be matched with a corresponding privacy calculation strategy in response to the data sharing request, so that the privacy calculation task is established, and the data provider can execute the privacy calculation task based on the bond data of the data provider so as to obtain a data sharing result. The data demand party can apply the bond data of other service participants to realize the data demand of the federal learning task or the data query task, thereby being beneficial to improving the processing effect of the data application task of the financial institution, meeting the data demand of the financial institution and mining the data value of the bond market. And secondly, the data sharing process is realized based on a privacy computing technology and an encryption technology, the bond data of the data provider cannot be locally output, the data demand party can only obtain the required data sharing result and can not obtain all the original data participating in task processing, and the data sharing can be realized under the condition of guaranteeing the security of the bond data.

The device provided by the embodiment of the present invention may further extend a plurality of units on the basis of the device shown in fig. 4, and the functions of each unit may be referred to in the foregoing description of each embodiment provided by the bond market data sharing method, which is not further illustrated herein.

The embodiment of the invention also provides a storage medium, which comprises stored instructions, wherein the equipment where the storage medium is located is controlled to execute the bond market data sharing method when the instructions run.

The embodiment of the present invention further provides an electronic device, whose structural schematic diagram is shown in fig. 5, specifically including a memory 501, and one or more instructions 502, where the one or more instructions 502 are stored in the memory 501, and configured to be executed by the one or more processors 503, where the one or more instructions 502 perform the following operations:

In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for a system or system embodiment, since it is substantially similar to a method embodiment, the description is relatively simple, with reference to the description of the method embodiment being made in part. The systems and system embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.

Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative elements and steps are described above generally in terms of functionality in order to clearly illustrate the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A bond market data sharing method, comprising:

The target task result is sent to the data demand party, so that the data demand party determines a data sharing result corresponding to the data sharing request based on the target task result;

if the sharing task category characterizes the data query task, determining, according to the privacy calculation policy and the request information, the privacy calculation task and the target public key corresponding to the data sharing request includes: determining a query attribute set and a query object set corresponding to the data sharing request based on the request information; the query attribute set comprises a plurality of query attributes, and the query object set comprises abstract values corresponding to a plurality of target query objects; the plurality of target query objects comprise query objects for the data demander to select a query and various query objects determined based on a preset careless transmission strategy; generating a combined hidden query task according to the query attribute set and the query object set, and taking the combined hidden query task as the privacy calculation task; generating a second key pair according to a preset encryption algorithm, and taking a public key in the second key pair as the target public key;

The data provider performs task processing on the privacy calculation task based on the bond data owned by the data provider, encrypts the task processing result through the target public key, and takes the encrypted result as a secret state task result corresponding to the data provider, and the method comprises the following steps: determining a data object set corresponding to the bond data; the data object set comprises abstract values corresponding to a plurality of data objects; performing privacy intersection processing on the data object set and the query object set based on a preset privacy intersection algorithm to obtain data objects corresponding to each target query object; the preset privacy intersection algorithm is a privacy intersection algorithm adopting a barrel-division optimization strategy based on editing distance; for each target query object, determining attribute data corresponding to a data object corresponding to the target query object in the bond data according to the query attribute set, and taking the attribute data as the attribute data corresponding to the target query object; taking the attribute data corresponding to each target query object as a task processing result of the privacy calculation task; encrypting the attribute data corresponding to each target query object through the target public key to obtain the secret attribute data corresponding to each target query object, and taking each secret attribute data as a secret task result corresponding to the data provider;

And carrying out result fusion processing on each close-state task result to obtain a target task result, wherein the method comprises the following steps of: decrypting each secret state task result through a private key in the second key pair to obtain a result plaintext corresponding to each secret state task result; the result plaintext corresponding to each secret state task result comprises an attribute data plaintext corresponding to secret state attribute data in the secret state task result; performing data splicing processing based on the result plaintext to obtain query attribute data corresponding to each target query object; and carrying out data processing on query attribute data corresponding to each target query object based on the careless transmission strategy to obtain careless transmission data corresponding to each target query object, and taking the careless transmission data corresponding to each target query object as the target task result so that the data demander can obtain the query attribute data corresponding to the query object selected by the data demander from the target task result based on the careless transmission strategy.

2. The method according to claim 1, wherein if the shared task class characterizes a federal learning task, the determining, according to the privacy calculation policy and the request information, a privacy calculation task and a target public key corresponding to the data sharing request includes:

3. The method according to claim 2, wherein the process of the data provider performing task processing on the privacy calculation task based on the bond data owned by the data provider and encrypting the task processing result by the target public key, and using the encrypted result as the secret task result corresponding to the data provider includes:

4. A method according to claim 3, wherein the performing the result fusion processing on each of the close task results to obtain a target task result includes:

5. A bond market data sharing apparatus, comprising:

the second sending unit is used for sending the target task result to the data demand party, so that the data demand party determines a data sharing result corresponding to the data sharing request based on the target task result;

6. A storage medium comprising stored instructions, wherein the instructions, when executed, control a device in which the storage medium is located to perform the bond market data sharing method according to any one of claims 1 to 4.

7. An electronic device comprising a memory and one or more instructions, wherein the one or more instructions are stored in the memory and configured to be executed by the one or more processors to perform the bond market data sharing method of any one of claims 1-4.