A joint rule product design method and system based on privacy protection

Abstract

This invention discloses a privacy-preserving federated rule product design method and system. The privacy-preserving federated rule product design system includes: a data source management module for creating data sources, including local assets, external assets, and API data sources; a variable management module connected to the data source management module for converting the created data sources into variables adapted to federated rules; and a decision flow management module connected to the variable management module for creating decision flows and configuring corresponding federated rules based on variables within the created decision flows to obtain the configured product decision flow.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a joint rule product design method and system based on privacy protection. Background Art

[0002] The goal of federated learning is to achieve joint modeling and improve the effectiveness of AI models on the basis of ensuring data privacy, security, and legal compliance. Federated learning is essentially a distributed machine learning technology, or a machine learning framework. The rule engine is developed from the inference engine and is a component embedded in the application. It separates business decisions from application code and writes business decisions using predefined semantic modules.

[0003] As data increases and becomes increasingly connected, many data cannot be interoperable due to issues such as user privacy and laws and regulations, resulting in many "data islands". The concept of Federated Learning was proposed by Google in 2017 to solve the problem of joint modeling across devices.

[0004] At present, the rule engines on the market can only be configured based on the data of the party. For example, if a commercial bank's risk control engine needs to introduce the data of the Social Security Bureau to set risk control rules, the Social Security Bureau needs to provide a plain text API interface for calling, or copy the social security data to the commercial bank for use. At present, it is impossible to protect the data of the Social Security Bureau and the rules configured by the commercial bank.

[0005] In response to the above technical problems, the present invention proposes a joint rule product design method and system based on privacy protection. Summary of the invention

[0006] The purpose of the present invention is to provide a joint rule product design method and system based on privacy protection in view of the defects of the prior art.

[0007] In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0008] A privacy-preserving joint rule product design system, comprising:

[0009] The data source management module is used to create data sources, including local assets, external assets, and API data sources;

[0010] A variable management module, connected to the data source management module, is used to convert the created data source into a variable that is compatible with the joint rule;

[0011] The decision flow management module is connected to the variable management module and is used to create a decision flow and obtain a configured product decision flow based on the federal rules corresponding to the variable configuration in the created decision flow.

[0012] Furthermore, the creation of local assets and external assets in the data source management module also includes: setting query fields corresponding to the local assets and external assets.

[0013] Furthermore, creating an API data source in the data source management module specifically includes: configuring an API request method, an API request address, and an API request parameter.

[0014] Furthermore, the conversion method in the variable management module includes configuration of variable name, variable description, and variable data source.

[0015] Furthermore, the decision flow created in the decision flow management module includes start, end, federation rules, branch components, and assignment.

[0016] Accordingly, a joint rule product design method based on privacy protection is also provided, including:

[0017] S1. Create data sources, including local assets, external assets, and API data sources;

[0018] S2. Convert the created data source into a variable that is compatible with the joint rule;

[0019] S3. Create a decision flow, and configure the corresponding federal rules based on the variables in the created decision flow to obtain the configured product decision flow.

[0020] Furthermore, the creation of the local assets and external assets in step S1 also includes: setting query fields corresponding to the local assets and external assets.

[0021] Furthermore, the creation of the API data source in step S1 specifically includes: configuring the API request method, the API request address, and the API request parameters.

[0022] Furthermore, the conversion method in step S2 includes configuration of variable name, variable description, and variable data source.

[0023] Furthermore, the decision flow created in step S3 includes start, end, federation rules, branch components, and assignment.

[0024] Compared with the prior art, the present invention solves the problem of calculating security rules for multiple parties' assets without leaving the domain, and can quickly obtain the rule calculation results of multiple parties' data. It can protect the rules configured by the party and the data of external data sources. This product solution supports real-time configuration changes on the front end, and the calculation results can be obtained in real time after submitting the calculation. It supports single rule configuration, rule group configuration, and conditional branch node configuration. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a structural diagram of a joint rule product design system based on privacy protection provided in Example 1;

[0026] Figure 2 It is a decision flow product design structure diagram provided in Example 1;

[0027] Figure 3 This is a schematic diagram of rule configuration design provided in Example 1;

[0028] Figure 4 The API data source provided in the first embodiment requires pre-designing data in the background;

[0029] Figure 5 It is a schematic diagram of the configuration of the variables provided in the first embodiment. DETAILED DESCRIPTION

[0030] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0031] The purpose of the present invention is to provide a privacy protection-based joint rule product design method and system to address the defects of the prior art.

[0032] Embodiment 1

[0033] This embodiment provides a joint rule product design system based on privacy protection, such as Figure 1 As shown, including:

[0034] The data source management module 11 is used to create data sources, where the data sources include local assets, external assets, and API data sources;

[0035] The variable management module 12 is connected to the data source management module 11 and is used to convert the created data source into a variable that is compatible with the joint rule;

[0036] The decision flow management module 13 is connected to the variable management module 12, and is used to create a decision flow, and obtain a configured product decision flow based on the federal rules corresponding to the variable configuration in the created decision flow.

[0037] In the data source management module 11, a data source is created, where the data source includes local assets, external assets, and API data sources.

[0038] Local assets can upload calculation files locally, and the calculation nodes can automatically identify the feature columns and feature types of the files.

[0039] External assets are data that external partners authorize us to use, but the actual data still exists in external computing nodes.

[0040] In this embodiment, when creating local assets and external assets on the page, it is necessary to set a query field button on the page in advance, and query the local assets or external assets by clicking the query field button; the query field serves as an index to obtain data, and the local node or external node will obtain data based on the query field when performing data source calculations.

[0041] The data acquisition logic is as follows: if the data is acquired at the local node, the query request is passed to the local node, the local node loads the local file in DataFrame format, and completes the value acquisition from the DataFrame according to the value of the query field, and stores the value in the local node. If the data is acquired at an external node, the query request is sent to the external node, the external node loads the external file in DataFrame format, completes the value acquisition from the DataFrame according to the query field and query value, and stores the value in the external node.

[0042] The API data source needs to pre-design the request method, request address, query parameters, output parameter analysis and other parts in the background. In this embodiment, when creating an API data source, you need to add the API request method (GET / POST / PUT / FETCH), API request address, API request parameters, and design the interface for initiating the call. After the interface call is successful, select the output fields and save to complete the configuration of the API data source.

[0043] The API data source needs to pre-design the API request method, request address, query parameters, and parameter parsing method in the background. Figure 4 shown.

[0044] In the variable management module 12, the created data sources are converted into variables adapted to the joint rules.

[0045] Variable management is based on data sources. Successfully created data sources are packaged into variables that can be referenced by joint rules. The process of packaging variables is as follows. First, on the variable management page, you need to set a button to create variables in advance. When you click the button to create variables, a create variable edit box pops up. The edit box contains options for variable name, variable description, and variable data source configuration. Figure 5 shown.

[0046] The definition of the variable name must be globally unique and non-repetitive; the options for variable data source configuration include local assets, external assets, or API data sources. When selecting the configuration, you can choose any of the three. After selecting the required data source, you need to select the output column of the data source. The output column represents the feature name of the local asset or external asset, or the output parameter in the API data source. Then, when the final joint rule is calculated, the variable will give the output column as a calculation parameter to the data source node for calculation.

[0047] In the decision flow management module 13, a decision flow is created, and corresponding federal rules are configured based on variables in the created decision flow to obtain a configured product decision flow.

[0048] In terms of decision flow management, this embodiment is designed in a canvas mode (but not limited to this), such as Figure 2 As shown in the figure, the decision flow includes the configuration of basic information, decision flow design, running records, version information and operation records. On the decision flow canvas, there are five components: start, end, rule, branch and assignment, and the decision flow canvas supports drag-and-drop drag-and-drop and connection.

[0049] In the rule component and branch component, it includes the introduction of variables, the configuration of federated rules, and the configuration of decision results. This embodiment creates a federated rule set in the rule component and branch component based on the introduced variables. The configuration of federated rules supports the comparison of variables and variables, and the comparison of variables and fixed values. The configuration of rule sets supports AND or OR logical operations between rules, such as Figure 3 shown.

[0050] After the joint decision canvas is configured, you can click the Run button to test it. Before running, the system needs to identify the input parameters of the decision flow calculation variables. After the input parameter values ​​are filled in, click the Execute button to output the joint rule results.

[0051] This embodiment can quickly obtain the rule calculation results of multiple data, which can protect the rules configured by the party and the data of external data sources. This product solution supports real-time configuration changes on the front end, and the calculation results can be obtained in real time after the calculation is submitted. It supports single rule configuration, rule group configuration, and conditional branch node configuration.

[0052] Accordingly, a joint rule product design method based on privacy protection is also provided, including:

[0053] S1. Create data sources, including local assets, external assets, and API data sources;

[0054] S2. Convert the created data source into a variable that is compatible with the joint rule;

[0055] S3. Create a decision flow, and configure the corresponding federal rules based on the variables in the created decision flow to obtain the configured product decision flow.

[0056] Furthermore, the creation of the local assets and external assets in step S1 also includes: setting query fields corresponding to the local assets and external assets.

[0057] Furthermore, the creation of the API data source in step S1 specifically includes: configuring the API request method, the API request address, and the API request parameters.

[0058] Furthermore, the conversion method in step S2 includes configuration of variable name, variable description, and variable data source.

[0059] Furthermore, the decision flow created in step S3 includes start, end, federation rules, branch components, and assignment.

[0060] Embodiment 2

[0061] The privacy protection-based joint rule product design system provided in this embodiment is different from the first embodiment in that:

[0062] This embodiment is described by taking a commercial bank and a social security bureau as an example.

[0063] Commercial banks need to configure risk control rules based on the social security payment year data from the Social Security Bureau.

[0064] Set customers with more than 3 years of service as approved and customers with less than 3 years of service as rejected. When configuring risk control rules, you only need to introduce the data assets of the partner Social Security Bureau, set the query field of the Social Security Bureau's assets to the user's identity code, and create a variable for the social security payment year.

[0065] Configure in the joint rule, the left value is the payment year of the social security bureau customer, the right value is set to 3, the logic is set to greater than or equal to, and the decision action is set to pass.

[0066] After the configuration is completed, click the Run button. The system needs to identify the input parameters of the decision flow calculation variables. The logic of input parameter identification is as follows: the system will identify all variables configured in the decision flow, and obtain the data source of the variable according to each variable, and then obtain the query parameter value corresponding to the data source according to the data source, and then obtain the parameter list that needs to be filled in the decision flow. In this example, the input parameter finally obtained by the decision flow is the user's identity code. Suppose we enter the user's identity code s and click the Execute button. The system recognizes that the corresponding data source is an external data source, and will pass the parameter to the external node for calculation. After the external node calculates, it is assumed that the value obtained is x. The local node takes the right value 3. After the external node obtains the value x, the two nodes perform a secure comparison based on the secret sharing protocol to obtain the comparison result. Assuming that the x obtained by the external node is 4, the result of the joint rule is passed.

[0067] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A privacy-preserving joint rule product design system, characterized in that: include: The data source management module is used to create data sources, including local assets, external assets, and API data sources; A variable management module, connected to the data source management module, is used to convert the created data source into a variable that is compatible with the joint rule; A decision flow management module, connected to the variable management module, is used to create a decision flow, and configure the corresponding federation rules based on the variables in the created decision flow to obtain a configured product decision flow; The creation of local assets and external assets in the data source management module further includes: setting query fields corresponding to the local assets and external assets; When the local asset or external asset performs data source calculation, data is retrieved according to the query field. The data retrieval logic is as follows: if data is retrieved from the local asset, the query request is passed to the local asset, the local asset loads the local file, and completes the value from the local file according to the value of the query field, and stores the value in the local asset; if data is retrieved from the external asset, the query request is sent to the external asset, the external asset loads the external file, completes the value from the external file according to the query field and the query value, and stores the value in the external asset; a secure comparison is performed between the local asset and the external asset based on the secret sharing protocol to obtain the comparison result; Creating an API data source in the data source management module specifically includes: configuring the API request method, API request address, and API request parameters; To convert variables, you need to set a button for creating variables in the variable management page in advance. When you click the button for creating variables, a variable creation edit box pops up, which contains options for variable name, variable description, and variable data source configuration. The definition of variable names must be globally unique and non-repetitive. The options for variable data source configuration include local assets, external assets, or API data sources. When selecting the configuration, you can choose any of the three. After selecting the required data source, you need to select the output column of the data source. The output column represents the feature name of the local asset or external asset, or the output parameter in the API data source. When the final joint rule is calculated, the variable will use the output column as a calculation parameter to the data source node for calculation; The decision flow created in the decision flow management module includes start, end, federation rules, branch components, and assignment; In the rule component and branch component, it includes the introduction of variables, the configuration of federated rules, and the configuration of decision results; based on the introduced variables, a federated rule set is created in the rule component and branch component; the configuration of federated rules supports the comparison between variables and between variables and between variables and fixed values, and the configuration of rule sets supports AND or OR logical operations between rules; The federal rules support real-time configuration changes on the front end, and obtain calculation results in real time after submitting the calculation. They support single rule configuration, rule group configuration, and conditional branch node configuration.

2. According to claim 1, a privacy protection-based joint rule product design system is characterized in that: The conversion method in the variable management module includes the configuration of variable name, variable description, and variable data source.

3. A method for designing a joint rule product based on privacy protection, characterized in that: include: S1. Create data sources, including local assets, external assets, and API data sources; S2. Convert the created data source into a variable that is compatible with the joint rule; S3. Create a decision flow, and configure the corresponding federation rules based on the variables in the created decision flow to obtain the configured product decision flow; The step S1 of creating the local assets and external assets also includes: setting query fields corresponding to the local assets and external assets; When the local asset or external asset performs data source calculation, the data is retrieved according to the query field, and the data retrieval logic is as follows: if the data is retrieved in the local asset, the query request is passed to the local asset, the local asset loads the local file, and completes the value retrieval from the local file according to the value of the query field, and stores the value in the local asset; if the data is retrieved in the external asset, the query request is sent to the external asset, the external asset loads the external file, completes the value retrieval from the external file according to the query field and the query value, and stores the value in the external asset; a security comparison is performed between the local asset and the external asset based on the secret sharing protocol to obtain the comparison result; the creation of the API data source in step S1 is specifically: configuring the API request method, API request address, and API request parameters; To convert variables, you need to set a button for creating variables in the variable management page in advance. When you click the button for creating variables, a variable creation edit box pops up, which contains options for variable name, variable description, and variable data source configuration. The definition of variable names must be globally unique and non-repetitive. The options for variable data source configuration include local assets, external assets, or API data sources. When selecting the configuration, you can choose any of the three. After selecting the required data source, you need to select the output column of the data source. The output column represents the feature name of the local asset or external asset, or the output parameter in the API data source. When the final joint rule is calculated, the variable will use the output column as a calculation parameter to the data source node for calculation; The decision flow created in the decision flow management module includes start, end, federation rules, branch components, and assignment; In the rule component and branch component, it includes the introduction of variables, the configuration of federated rules, and the configuration of decision results; based on the introduced variables, a federated rule set is created in the rule component and branch component; the configuration of federated rules supports the comparison between variables and between variables and between variables and fixed values, and the configuration of rule sets supports AND or OR logical operations between rules; The federal rules support real-time configuration changes on the front end, and obtain calculation results in real time after submitting the calculation. They support single rule configuration, rule group configuration, and conditional branch node configuration.

4. The method for designing a joint rule product based on privacy protection according to claim 3, characterized in that: The conversion method in step S2 includes configuration of variable name, variable description, and variable data source.

Images