Business processing method and device based on credit processing system, electronic equipment, storage medium and program product

Abstract

The embodiment of the invention provides a business processing method and device based on a credit processing system, electronic equipment, a storage medium and a program product, and relates to the field of financial science and technology or other related fields. The method is applied to a central server of the credit processing system, the credit processing system comprises the central server and a plurality of sub-servers, a credit data processing request is received, a local database is inquired according to user information in the credit data processing request, and credit data of a target user uploaded by each sub-server is obtained. And based on the historical prediction accuracy and credit data corresponding to each sub-server, determining comprehensive credit data of the target user, and sending the comprehensive credit data of the target user to the target sub-server, so that the target sub-server determines a business handling strategy for the target user. According to the method, the correctness of the obtained credit data is improved, so that the accuracy of a business handling strategy is improved.

Description

Technical Field

[0001] The present application relates to the field of financial technology or other related fields, and in particular to a business processing method, device, electronic device, storage medium and program product based on a credit processing system. Background Art

[0002] Banks and other institutions can play an important role in identifying credit risks of users in many aspects, such as providing banks with decision-making basis and helping banks predict users' loan amounts, interest rates and guarantee requirements.

[0003] At present, banks generally identify credit risks of users by using their own accumulated user data to build a credit risk identification model. The credit risk identification model is used to give the user's credit rating in the bank, thereby determining business decisions for the user.

[0004] However, the user data accumulated by a single bank is insufficient to identify abnormal behavior of users in other financial institutions, resulting in serious distortion in credit risk identification and the inability to make appropriate business decisions. Summary of the Invention

[0005] The present application provides a business processing method, device, electronic device, storage medium and program product based on a credit processing system to solve the technical problem of distorted credit risk identification, which leads to the inability to make appropriate business decisions.

[0006] In a first aspect, the present application provides a business processing method based on a credit processing system, which is applied to a central server of the credit processing system, wherein the credit processing system includes a central server and multiple sub-servers, wherein the multiple sub-servers are respectively deployed in various banking institutions, including:

[0007] Receiving a credit data processing request sent by any sub-server, wherein the credit data processing request is generated by the sub-server receiving a business processing request sent by a user terminal, obtaining user information of a target user in the business processing request, and generating the request based on the user information;

[0008] Obtaining user information from the credit data processing request, and querying a local database based on the user information to obtain the credit data of the target user uploaded by each sub-server; wherein the credit data corresponding to each sub-server is determined by the sub-server inputting historical transaction data with the target user into a locally deployed credit risk model;

[0009] Determine the historical prediction accuracy corresponding to each sub-server;

[0010] Determine the comprehensive credit data corresponding to the target user based on the historical prediction accuracy and credit data corresponding to each sub-server;

[0011] The comprehensive credit data corresponding to the target user is sent to the target sub-server, so that the target sub-server determines the business handling strategy for the target user based on the comprehensive credit data, wherein the target sub-server is the sub-server that generates the credit data processing request.

[0012] In a second aspect, the present application provides a business processing device based on a credit processing system, which is applied to a central server of the credit processing system, wherein the credit processing system includes a central server and multiple sub-servers, wherein the multiple sub-servers are respectively deployed in various banking institutions, including:

[0013] a receiving module, configured to receive a credit data processing request sent by any sub-server, wherein the credit data processing request is a business processing request received by any sub-server from a user terminal, obtain user information of a target user in the business processing request, and generate the request based on the user information;

[0014] The processing module is further configured to obtain user information in the credit data processing request, and query a local database based on the user information to obtain the credit data of the target user uploaded by each sub-server; wherein the credit data corresponding to each sub-server is determined by the sub-server inputting historical transaction data with the target user into a locally deployed credit risk model;

[0015] A processing module, used to determine the historical prediction accuracy corresponding to each sub-server;

[0016] The processing module is further configured to determine the comprehensive credit data corresponding to the target user based on the historical prediction accuracy and credit data corresponding to each sub-server;

[0017] The sending module is used to send the comprehensive credit data corresponding to the target user to the target sub-server, so that the target sub-server determines the business processing strategy for the target user based on the comprehensive credit data, wherein the target sub-server is the sub-server that generates the credit data processing request.

[0018] In a third aspect, the present application provides an electronic device, comprising: a processor, and a memory communicatively connected to the processor;

[0019] The memory stores computer-executable instructions;

[0020] The processor executes the computer-executable instructions stored in the memory to implement the above method.

[0021] In a fourth aspect, the present application provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to implement the method as described above when executed by a processor.

[0022] In a fifth aspect, the present application provides a computer program product, characterized in that it includes a computer program, which implements the method described above when executed by a processor.

[0023] The business processing method, device, electronic device, storage medium and program product based on the credit processing system provided by the present application obtain comprehensive credit data about the user by integrating the credit data of the user from the sub-servers of each bank, and sends the comprehensive credit data to the target sub-server so that the target sub-server can determine the business processing strategy for the user based on the comprehensive credit data of the user. This solves the disadvantage that a single bank cannot provide accurate credit data, improves the correctness of the obtained credit data, and thus improves the accuracy of determining the business processing strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0025] Figure 1 A schematic diagram of an application scenario of the prior art is provided as an example;

[0026] Figure 2 Schematic diagram of the business processing method based on the credit processing system provided for this application Figure 1 ;

[0027] Figure 3 A schematic diagram of an exemplary credit processing system;

[0028] Figure 4 Schematic diagram of the business processing method based on the credit processing system provided for the application Figure 2 ;

[0029] Figure 5 Schematic diagram of the business processing method based on the credit processing system provided for the application Figure 3 ;

[0030] Figure 6 Schematic diagram of the business processing method based on the credit processing system provided for the application Figure 4 ;

[0031] Figure 7 Schematic diagram of the structure of the multi-bank collaborative training learning model;

[0032] Figure 8The overall flow chart of the model training for the example;

[0033] Figure 9 Schematic diagram of the business processing method based on the credit processing system provided for this application Figure 5 ;

[0034] Figure 10 A schematic diagram of the structure of a business processing device based on a credit processing system provided in this application;

[0035] Figure 11 This is a schematic diagram of the structure of the electronic device provided in this application.

[0036] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0037] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0038] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with the relevant laws, regulations and standards of the relevant regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0039] In addition, this application involves conducting big data analysis of user information (including but not limited to personal biometrics, identity data, consumption data, asset data, electronic terminal operation data, etc.), and using artificial intelligence technology to make automated decisions, and making technical solutions that have a significant impact on personal rights and interests based on the results of automated decisions. The application provides users with corresponding operation entrances for them to choose to agree or reject the results of automated decisions; if the user chooses to reject, the expert decision-making process will be entered.

[0040] It should be noted that the business processing methods, devices, electronic devices, storage media and program products based on the credit processing system provided in this application can be used in the field of financial technology, and can also be used in any field other than the field of financial technology. The application fields of the business processing methods, devices, electronic devices, storage media and program products based on the credit processing system in this application are not limited.

[0041] Figure 1 Schematic diagram of application scenario of the prior art as an example, Figure 1 As shown, a user terminal sends a business request, such as a loan request, to a server at a bank or other institution. The bank's server analyzes the user's historical transaction data to identify the user's credit risk and determine their credit rating. This information is then used to determine the user's business processing strategy, such as the loan amount and interest rate. Currently, banks typically use their accumulated user data to build a credit risk identification model. This model then uses the user's credit rating and other credit data to determine business decisions for the user. However, individual banks often lack sufficient accumulated user data to identify anomalous user behavior at other financial institutions. This leads to significant distortion in credit risk identification and, consequently, the inability to make appropriate business decisions.

[0042] The business processing method, device, electronic device, storage medium and program product based on the credit processing system provided in this application are intended to solve the above technical problems in the prior art.

[0043] This application uses a central server to integrate the credit data of each sub-server on the user, determine the user's comprehensive credit data, and send the comprehensive credit data corresponding to the user to the bank's sub-server, so that the sub-server can determine the business processing strategy for the user based on the user's comprehensive credit data.

[0044] The following specific embodiments describe in detail the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0045] Figure 2 Schematic diagram of the business processing method based on the credit processing system provided for this application Figure 1 , a business processing method based on a credit processing system is applied to a central server of the credit processing system, wherein the credit processing system includes a central server and a plurality of sub-servers, wherein the plurality of sub-servers are respectively deployed in various banking institutions, such as Figure 2 Shown, including:

[0046] S201. Receive a credit data processing request sent by any sub-server, wherein the credit data processing request is a business processing request sent by a user terminal received by any sub-server, obtains user information of a target user in the business processing request, and generates the request based on the user information.

[0047] Combined with the scene example, Figure 3 A schematic diagram of a credit processing system is shown as an example. Figure 3 As shown, the credit processing system includes a central server and sub-servers corresponding to each sub-bank. The central server is the main execution entity of this application. A user sends a business processing request to a sub-bank through a user terminal. Based on the business processing request, the sub-server obtains the user information contained in the business processing request, which can include the user's name, mobile phone number, login account, etc. The sub-server generates a credit data processing request for the user based on the user information and sends it to the central server. The central server can accept credit data processing requests from any sub-server.

[0048] S202. Obtain user information in the credit data processing request, and query a local database based on the user information to obtain the credit data of the target user uploaded by each sub-server; wherein the credit data corresponding to each sub-server is determined after the sub-server inputs historical transaction data with the target user into a locally deployed credit risk model.

[0049] In this scenario example, since the credit data processing request is generated by a sub-server based on user information, the central server can parse the request to obtain the user information contained therein. The central server's local database contains user information and credit data corresponding to each user. Therefore, the central server can query the local database for the credit data corresponding to the target user associated with that user information. In the local database, each user's credit data includes credit data uploaded by each sub-server. Each sub-server can input locally stored historical transaction data of that user into a credit risk model, which then outputs credit data related to that user based on that transaction data. Therefore, each sub-server can input locally stored historical transaction data of a target user into the credit risk model, which then outputs credit data related to that target user. Credit scoring models are key tools for credit risk management. These models use machine learning algorithms to analyze multi-dimensional characteristics of users (including but not limited to historical repayment records, debt-to-equity ratios, transaction activity, and other transaction data) and output quantitative credit data or default probability estimates.

[0050] S203: Determine the historical prediction accuracy corresponding to each sub-server.

[0051] In combination with the scenario example, the central server can determine the historical prediction accuracy based on the total number of historical predictions and the number of correct predictions of each sub-server.

[0052] S204: Determine the comprehensive credit data corresponding to the target user based on the historical prediction accuracy and credit data corresponding to each sub-server.

[0053] Combined with the scenario example, the credit data corresponding to each sub-server can be weighted according to the historical prediction accuracy of each sub-server, so as to obtain the total comprehensive credit data that combines the credit data of the target user on each sub-server.

[0054] S205. Send the comprehensive credit data corresponding to the target user to a target sub-server, so that the target sub-server determines a business handling strategy for the target user based on the comprehensive credit data, wherein the target sub-server is the sub-server that generates the credit data processing request.

[0055] Combined with the scene example, combined with Figure 3 The central server can identify the sub-server that sent the credit data processing request as the target sub-server, and then return the obtained comprehensive credit data to the target sub-server. The target sub-server can then determine the service processing policy for the target user based on the target user's corresponding comprehensive credit data. For example, if the service processing request sent by the user terminal is for loan processing, the target sub-server can determine the service processing policy for the target user based on the target user's corresponding comprehensive credit data, such as the loan amount and repayment interest rate.

[0056] Based on the method provided in this example, the central server can obtain comprehensive credit data about the user, which solves the problem that a single bank cannot provide accurate credit data, improves the correctness of the obtained credit data, and thus improves the accuracy of determining business processing strategies.

[0057] Optional, Figure 4 Schematic diagram of the business processing method based on the credit processing system provided for this application Figure 2 ,like Figure 4 As shown, S203 includes:

[0058] S401. If there is no historical credit data processing request, the initial prediction accuracy corresponding to the credit risk model locally deployed on each sub-server is determined as the historical prediction accuracy corresponding to each sub-server, wherein the initial prediction accuracy corresponding to the credit risk model is determined by a preset test set.

[0059] In this scenario example, if there are no historical credit processing requests, this means the central server is currently receiving its first credit data processing request. In this case, the historical prediction accuracy corresponding to each sub-server is the initial prediction accuracy of the credit risk model. The initial prediction accuracy of the credit risk model can be determined using a test set. The test set includes multiple user data items, each of which includes user information and standard credit data. Each sub-server can sequentially input the user information in each user data item into a locally deployed credit risk model, causing the credit risk model to output predicted credit data corresponding to the user information in each user data item. The sub-server compares the predicted credit data with the standard credit data. If the difference between the predicted credit data and the standard credit data is excessive, the predicted credit data is determined to be inaccurate. For example, the credit data can be in numerical form. If the difference between the predicted credit data and the standard credit data exceeds a preset range, for example, if the difference between the predicted credit data and the standard credit data exceeds 10%, the predicted credit data is determined to be inaccurate. The sub-server compares the total amount of inaccurate predicted credit data with the standard credit data to determine the initial prediction accuracy of the locally deployed credit risk model. The central server interacts with each sub-server to obtain the initial prediction accuracy of the credit risk model deployed on each sub-server and determines it as the historical prediction accuracy corresponding to each sub-server.

[0060] S402. If there is a historical credit data processing request, based on obtaining the historical user information corresponding to the historical target user in the historical credit data processing request, query the local database according to the historical user information to obtain the historical credit data of the historical target user uploaded by each sub-server, and determine the historical comprehensive credit data corresponding to the historical target user. Through the historical comprehensive credit data corresponding to the historical target user and the historical credit data about the historical target user uploaded by each sub-server, determine the historical prediction accuracy corresponding to each sub-server.

[0061] Combined with the scenario example, the situation where there is a historical credit processing request means that the credit data processing request currently received by the central server is not the first credit data processing request. At this time, the historical user information of each historical target user obtained when processing each previous historical credit data processing request and the historical comprehensive credit data of each historical target user can be obtained. Through the historical user information of each historical target user, the local database is queried to determine the historical credit data of each sub-server for each historical target user. For example, if there are 100 historical credit processing requests, the corresponding 100 historical target users can be determined through the 100 historical credit processing requests, so as to obtain the historical comprehensive credit data corresponding to each historical target user. For each historical target user, the local database can be queried based on the historical user information corresponding to the historical target user to determine the historical credit data of each sub-server for the historical target user.

[0062] For each historical target user, each sub-server compares the historical credit data for that historical target user with the historical comprehensive credit data for that historical target user. If the difference between the historical credit data for that historical target user and the historical comprehensive credit data for that historical target user on a particular sub-server exceeds 10%, the historical credit data for that historical target user on that sub-server is deemed inaccurate. In this way, the number of accurate historical credit data for all historical target users on each sub-server is determined. This number is then compared to 100 to obtain the corresponding historical prediction accuracy for that sub-server. If a sub-server has 90 accurate historical credit data for all historical target users, then the corresponding historical prediction accuracy for that sub-server is 90%.

[0063] Based on the method provided in this example, the historical preset accuracy rate of each sub-server is obtained based on the actual historical credit data corresponding to each sub-server, which can ensure the accuracy of the historical preset accuracy rate obtained in each case.

[0064] Optional, Figure 5 Schematic diagram of the business processing method based on the credit processing system provided for this application Figure 3 ,like Figure 5 As shown, S204 includes:

[0065] S501: Determine the historical prediction accuracy corresponding to each sub-server as the first weight corresponding to each sub-server.

[0066] Combined with the scenario example, the historical prediction accuracy corresponding to each sub-server can be recorded as Pi.

[0067] S502: Based on the first weight corresponding to each sub-server, perform weighted summation on the credit data corresponding to each sub-server to obtain a summation result.

[0068] In combination with the scenario example, the credit data corresponding to each sub-server can be recorded as Si, and the sum result can be obtained by referring to the following formula:

[0069]

[0070] S503: Determine a target number of credit data that meets preset conditions in the credit data corresponding to each sub-server.

[0071] In this scenario example, the precondition is non-zero credit data. In reality, the target user will conduct transactions with one or more banks, so the target user's transaction data will be stored locally on one or more sub-servers. Therefore, the sub-servers with the target user's transaction data can obtain non-zero credit data using the locally deployed credit risk model. Since the sub-servers with the target user's transaction data will obtain zero credit data for the target user, the non-zero target quantity can be determined from all the credit data for the target user.

[0072] S504: Determine the ratio between the sum result and the target number as the comprehensive credit data corresponding to the target user.

[0073] Combined with the scenario example, the above summation result is compared with the target number, and the obtained ratio is determined as the comprehensive credit data corresponding to the target user.

[0074] Based on the method provided in this example, taking into account the situation where the credit data is zero, the accuracy of the final comprehensive credit data can be improved by determining the ratio between the summation result and the target number as the comprehensive credit data.

[0075] Optionally, before querying the local database according to the user information, the method further includes:

[0076] Construct a local database, wherein the local database includes user information corresponding to each user and credit data of each user uploaded by each sub-server; wherein the credit data of each user corresponding to each sub-server is determined after the sub-server inputs historical transaction data with each user into a locally deployed credit risk model.

[0077] In this scenario example, each sub-server can input locally stored historical transaction data corresponding to each user into a credit risk model. The credit risk model can then output credit data about each user based on that historical transaction data. The sub-server then uploads the obtained credit data and user information about each user to the central server. The central server obtains the credit data and user information about each user uploaded by each sub-server and creates a corresponding sub-database for each sub-server. Each sub-database corresponding to each sub-server stores the credit data and user information about each user uploaded by that sub-server. Based on the sub-database corresponding to each sub-server, the central server constructs a master local database.

[0078] Based on the method provided in this example, the local database constructed by the central server can accommodate the credit data of each user of each sub-server of each sub-bank, thereby improving the convenience of obtaining the comprehensive credit data of the target user.

[0079] Optionally, after building the local database, the method further includes:

[0080] According to the preset processing cycle, the new credit data of each user uploaded by each sub-server in the current processing cycle is updated to the local database.

[0081] In conjunction with the scenario example, the processing cycle can be determined based on actual conditions. If the processing cycle is determined to be one month, each sub-server will locally generate new transaction data about each user within the past month. Therefore, each sub-server will input the new transaction data generated within the past month into the locally deployed credit risk model to obtain new credit data about each user. Each sub-server will upload the new credit data of each user obtained based on the new transaction data of each user this month to the central server at a preset time each month, such as on the last day of each month. The central server will update the local database with the new credit data of each user uploaded by each sub-server each month. When updating the new credit data to the local database, the corresponding update date can be recorded together, and the credit data corresponding to the old date can be determined as historical credit data.

[0082] Based on the method provided in this example, the local database can be updated periodically to ensure that the local database includes the latest credit data uploaded by each sub-server, which can improve the accuracy of the subsequent comprehensive credit data.

[0083] Optional, Figure 6 Schematic diagram of the business processing method based on the credit processing system provided for this application Figure 4 ,like Figure 6As shown, before receiving the credit data processing request sent by any sub-server, the method further includes:

[0084] S601. Build a learning model.

[0085] Combined with the scene example, combined with Figure 3 The credit processing system provided in this application is a "center-edge" architecture. The central server can be a federated learning server, so the central server creates an initial learning model and completes the training of the learning model through multi-bank collaboration to obtain a credit scoring model. Federated learning is a distributed machine learning paradigm. Multiple data holders (such as banks) can achieve the purpose of collaborative model training by encrypting and transmitting model parameters or gradient update information without directly sharing the original data. Multi-bank collaboration refers to the establishment of a secure computing alliance by multiple banking institutions, and the use of cryptographic technology to achieve cross-institutional joint modeling while ensuring that the user's transaction data is always retained locally.

[0086] S602: Send the learning model to each sub-server so that each sub-server deploys the learning model, and input historical transaction data with the user into the learning model so that the learning model outputs credit data corresponding to the user.

[0087] Combined with the scene example, Figure 7 The structural diagram of the multi-bank collaborative training learning model is shown as follows: Figure 7 As shown, the central server's functional modules include a model coordination module, a security aggregation module, and a credit data generation module. The model coordination module is responsible for model initialization and distribution; the security aggregation module is responsible for implementing gradient aggregation based on homomorphic encryption; and the credit data generation module is responsible for synthesizing the credit data of each sub-server to output the final comprehensive credit data. Each sub-server includes a data preprocessing module, a local training module, and a privacy protection module. The data preprocessing module preprocesses local historical transaction data; the local training module trains the learning model; and the privacy protection module implements gradient obfuscation and encrypted transmission.

[0088] Figure 8 The overall flow chart of the model training for the example is as follows: Figure 8As shown, the central server distributes the initial learning model created through the model coordination module to each sub-server. Each sub-server deploys the received learning model locally, and each sub-server trains its local learning model in parallel. Each sub-server uses the data preprocessing module to cleanse and normalize the locally stored historical transaction data of each user. The local training module then inputs this preprocessed historical transaction data into the learning model, which outputs credit data for each user, which can be a quantified value.

[0089] S603: Obtain the gradient encrypted data sent by each sub-server, wherein the gradient encrypted data is obtained by each sub-server adding noise data to the gradient data of the model parameters in the learning model and performing encryption calculation using a preset encryption algorithm.

[0090] Combined with the scene example, combined with Figure 8 , each sub-server can calculate the gradient data of the model parameters of the learning model through back propagation according to the privacy protection module, and the gradient data can be recorded as The noise data can be Gaussian noise, and then the sub-server uses the gradient obfuscation technology to add Gaussian noise to the gradient data through the privacy protection module. The Gaussian noise can be recorded as , the gradient data after adding Gaussian noise can be recorded as ,so Gradient obfuscation is a technical method that interferes with the model gradient information to prevent attackers from inferring the original data. Then the sub-server uses the Paillier public key encryption algorithm to encrypt the data through the privacy protection module. Perform encryption calculation to obtain gradient encrypted data, which can be recorded as Finally, the sub-server sends the obtained gradient encrypted data to the central server, and the central server can obtain the gradient encrypted data sent by each sub-server.

[0091] S604: Based on the gradient encrypted data sent by each sub-server, the learning model is updated to obtain a new learning model.

[0092] Combined with the scene example, combined with Figure 8, the central server serially and securely aggregates the gradient encrypted data of each sub-server through the security aggregation module, and performs homomorphic encryption and decryption on the aggregated data to update the model parameters in the learning model, thereby obtaining a new learning model. Homomorphic encryption and decryption refers to a groundbreaking cryptographic scheme. Its uniqueness lies in allowing specific algebraic operations (such as addition or multiplication) to be performed directly on ciphertext, and the decryption result is consistent with the result of the same operation on the plaintext. Gradient aggregation is the core mechanism of iterative optimization of federated learning, which includes three key links: first, the central server receives the model gradient data of each sub-server, then adopts a weighted average strategy for aggregation, and finally distributes the aggregated results to all sub-servers for model update.

[0093] S605. Repeat the steps of sending the learning model to each sub-server and updating the learning model based on the gradient encrypted data sent by each sub-server until the loss values ​​corresponding to the credit data output by the learning models deployed on each sub-server converge.

[0094] In this scenario example, the central server uses the model coordination module to distribute the new learning model to each sub-server to complete a new round of updates. In each update round, each sub-server obtains its current and previous credit data and calculates the difference between the two to determine the credit loss. Training the learning model ends when the credit loss values ​​for each sub-server converge.

[0095] S606: Determine the most recently updated learning model as the credit scoring model, and send the credit scoring model to each sub-server so that each sub-server deploys the credit scoring model locally.

[0096] In combination with the scenario example, after the training of the learning model is completed, the most recently updated learning model is determined as the credit scoring model, and the credit scoring model is sent to each sub-server.

[0097] Based on the method provided in this example, the central server can combine the training results of the learning model of multiple sub-servers to update the learning model, thereby ensuring that the obtained credit scoring model is applicable to each sub-server.

[0098] Optional, Figure 9 Schematic diagram of the business processing method based on the credit processing system provided for this application Figure 5 ,like Figure 9 As shown, S604 includes:

[0099] S901: Determine a second weight corresponding to each sub-server.

[0100] In this scenario example, the second weight corresponding to each sub-server can be determined based on the amount of historical transaction data of users stored locally on each sub-server. For example, the total historical transaction data stored locally on all sub-servers can be obtained, and the amount of historical transaction data of users stored locally on each sub-server can be compared with the total historical transaction data. The resulting ratio can be used to determine the second weight.

[0101] S902: Based on the second weight corresponding to each sub-server, perform weighted aggregation on the gradient encrypted data sent by each sub-server to obtain a corresponding aggregation result.

[0102] In this scenario example, the central server performs homomorphic encryption on the gradient encrypted data of each sub-server through the security aggregation module. During gradient aggregation, the gradient encrypted data can be weighted summed based on the second weight corresponding to each sub-server to obtain the aggregation result. For example, the following formula can be used to obtain the aggregation result:

[0103]

[0104] Among them, ɑi refers to the second weight corresponding to each sub-server, Refers to the gradient encrypted data corresponding to each sub-server, Refers to the aggregation result.

[0105] S903: Perform homomorphic decryption on the aggregation result to obtain a corresponding aggregation gradient.

[0106] Combined with the scenario example, the central server can homomorphically decrypt the aggregation result through the corresponding private key, and the obtained aggregation gradient can be recorded as .

[0107] S904: Update the learning model based on the aggregated gradient to obtain an updated new learning model.

[0108] Combined with the scenario example, finally based on the obtained The model parameters of the learning model are updated to obtain a new learning model.

[0109] Based on the method provided in this example, the central server aggregates the encrypted gradient encrypted data and then performs homomorphic decryption. The learning model is updated using the aggregated gradient after homomorphic decryption. Since the gradient encrypted data obtained by the central server is encrypted data, it can ensure that the data between the sub-servers will not be leaked.

[0110] Optionally, updating the learning model based on the aggregated gradient to obtain an updated new learning model includes:

[0111] Based on the aggregated gradient, the learning model is updated using a gradient descent method to obtain an updated new learning model.

[0112] Combined with the scenario examples, the learning model can be recorded as , and denote the new learning model as . Get the aggregate gradient and preset constants The product between them is used to update the learning model by the following formula to obtain a new learning model:

[0113]

[0114] Based on the method provided in this example, the gradient descent method is used to update the learning model, which can ensure the orderliness of the update process and thus ensure the correctness of the update of the learning model.

[0115] This embodiment obtains comprehensive credit data about the user by integrating the credit data of the sub-servers of various banks, and sends the comprehensive credit data to the target sub-server so that the target sub-server can determine the business processing strategy for the user based on the comprehensive credit data of the user. This solves the disadvantage that a single bank cannot provide accurate credit data, improves the correctness of the obtained credit data, and thus improves the accuracy of determining the business processing strategy.

[0116] Figure 10 This is a schematic diagram of the structure of a business processing device based on a credit processing system provided in this application. The business processing device based on a credit processing system is applied to the central server of the credit processing system, wherein the credit processing system includes a central server and multiple sub-servers, wherein the multiple sub-servers are respectively deployed in various banking institutions, including:

[0117] Receiving module 101, configured to receive a credit data processing request sent by any sub-server, wherein the credit data processing request is generated by any sub-server receiving a business processing request sent by a user terminal, obtaining user information of a target user in the business processing request, and generating the request based on the user information;

[0118] The processing module 102 is further configured to obtain user information from the credit data processing request, and query a local database based on the user information to obtain the credit data of the target user uploaded by each sub-server; wherein the credit data corresponding to each sub-server is determined by the sub-server inputting historical transaction data with the target user into a locally deployed credit risk model;

[0119] Processing module 102, configured to determine the historical prediction accuracy corresponding to each sub-server;

[0120] The processing module 102 is further configured to determine the comprehensive credit data corresponding to the target user based on the historical prediction accuracy and credit data corresponding to each sub-server;

[0121] The sending module 103 is used to send the comprehensive credit data corresponding to the target user to the target sub-server, so that the target sub-server determines the business processing strategy for the target user based on the comprehensive credit data, wherein the target sub-server is the sub-server that generates the credit data processing request.

[0122] Optionally, the processing module 102 is further configured to, if there is no historical credit data processing request, determine the initial prediction accuracy rate corresponding to the credit risk model locally deployed on each sub-server as the historical prediction accuracy rate corresponding to each sub-server, wherein the initial prediction accuracy rate corresponding to the credit risk model is determined using a preset test set;

[0123] The processing module 102 is further specifically configured to, if there is a historical credit data processing request, query a local database based on the historical user information corresponding to the historical target user in the historical credit data processing request to obtain the historical credit data of the historical target user uploaded by each sub-server, and determine the historical comprehensive credit data corresponding to the historical target user. The historical prediction accuracy corresponding to each sub-server is determined through the historical comprehensive credit data corresponding to the historical target user and the historical credit data about the historical target user uploaded by each sub-server.

[0124] Optionally, the processing module 102 is further configured to determine the historical prediction accuracy corresponding to each sub-server as the first weight corresponding to each sub-server;

[0125] The processing module 102 is further configured to perform a weighted summation of the credit data corresponding to each sub-server based on the first weight corresponding to each sub-server to obtain a summation result;

[0126] The processing module 102 is further configured to determine a target number of credit data that meets a preset condition in the credit data corresponding to each sub-server;

[0127] The processing module 102 is further configured to determine the ratio between the sum result and the target number as the comprehensive credit data corresponding to the target user.

[0128] Optionally, the processing module 102 is also used to build a local database, wherein the local database includes user information corresponding to each user and credit data of each user uploaded by each sub-server; wherein the credit data of each user corresponding to each sub-server is determined after the sub-server inputs historical transaction data with each user into a locally deployed credit risk model.

[0129] Optionally, the processing module 102 is further configured to update the local database with the new credit data of each user uploaded by each sub-server in the current processing cycle according to a preset processing cycle.

[0130] Optionally, the processing module 102 is further configured to construct a learning model;

[0131] The processing module 102 is further configured to send the learning model to each sub-server so that each sub-server deploys the learning model, and input historical transaction data with the user into the learning model so that the learning model outputs the credit data corresponding to the user;

[0132] The processing module 102 is further configured to obtain encrypted gradient data sent by each sub-server, wherein the encrypted gradient data is obtained by each sub-server adding noise data to the gradient data of the model parameters in the learning model and performing encryption calculation using a preset encryption algorithm;

[0133] The processing module 102 is further configured to update the learning model based on the gradient encrypted data sent by each sub-server to obtain a new learning model;

[0134] The processing module 102 is further configured to repeatedly execute the steps of sending the learning model to each sub-server and updating the learning model based on the gradient encrypted data sent by each sub-server until the loss values ​​corresponding to the credit data output by the learning models deployed on each sub-server converge;

[0135] The processing module 102 is further configured to determine the most recently updated learning model as the credit scoring model, and send the credit scoring model to each sub-server so that each sub-server deploys the credit scoring model locally.

[0136] Optionally, the processing module 102 is further configured to determine a second weight corresponding to each sub-server;

[0137] The processing module 102 is further configured to perform weighted aggregation on the gradient encrypted data sent by each sub-server based on the second weight corresponding to each sub-server to obtain a corresponding aggregation result;

[0138] The processing module 102 is further configured to perform homomorphic decryption on the aggregation result to obtain a corresponding aggregation gradient;

[0139] The processing module 102 is further configured to update the learning model based on the aggregated gradient to obtain an updated new learning model.

[0140] Optionally, the processing module 102 is further configured to obtain, based on the aggregated gradient, an update of the learning model using a gradient descent method to obtain an updated new learning model.

[0141] The business processing device based on the credit processing system provided in this embodiment can execute the business processing method based on the credit processing system provided in the above method embodiment. Its implementation principle and technical effects are similar and will not be described in detail in this embodiment.

[0142] Figure 11 This is a schematic diagram of the structure of the electronic device provided in this application. Figure 11 As shown, the electronic device 50 provided in this embodiment includes: at least one processor 501 and a memory 502. Optionally, the electronic device 50 further includes a communication component 503. The processor 501, the memory 502 and the communication component 503 are connected via a bus.

[0143] In a specific implementation process, at least one processor 501 executes the computer-executable instructions stored in the memory 502, so that the at least one processor 501 performs the above method.

[0144] The specific implementation process of the processor 501 can be found in the above method embodiment. Its implementation principle and technical effects are similar and will not be repeated here in this embodiment.

[0145] In the above embodiments, it should be understood that the processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the present invention may be directly executed by a hardware processor or by a combination of hardware and software modules in the processor.

[0146] The memory may include a high-speed memory (Random Access Memory, referred to as RAM), and may also include a non-volatile memory (NVM), such as at least one disk storage.

[0147] A bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Buses can be categorized as address buses, data buses, and control buses. For ease of presentation, the buses in the drawings of this application are not limited to just one bus or just one type of bus.

[0148] The present application also provides a computer program product, including a computer program, which implements the above method when executed by a processor.

[0149] The present application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the above method is implemented.

[0150] The readable storage medium may be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium may be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0151] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be an integral part of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist in a device as discrete components.

[0152] The division of units is merely a logical functional division; actual implementations may employ alternative divisions, such as combining or integrating multiple units or components into another system, or omitting or disabling certain features. Furthermore, any direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between devices or units, either through an interface, electrical, mechanical, or other means.

[0153] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0154] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0155] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the method of the present invention. The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0156] Those skilled in the art will appreciate that all or part of the steps in the above-described method embodiments can be implemented using hardware associated with program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0157] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all optional embodiments, and the actions and modules involved are not necessarily required by this application.

[0158] It should be understood that the above device embodiments are merely illustrative, and the device of the present application may also be implemented in other ways. For example, the division of units / modules in the above embodiments is merely a logical function division, and other division methods may be used in actual implementation.

[0159] In the above embodiments, the description of each embodiment has its own emphasis. For parts not described in detail in a particular embodiment, please refer to the relevant description of other embodiments. The technical features of the above embodiments can be combined in any way. To keep the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0160] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0161] It should be understood that the present application is not limited to the exact structure described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A business processing method based on a credit processing system, characterized in that: A central server applied to the credit processing system, wherein the credit processing system comprises a central server and a plurality of sub-servers, wherein the plurality of sub-servers are respectively deployed in various banking institutions, including: Receiving a credit data processing request sent by any sub-server, wherein the credit data processing request is generated by the sub-server receiving a business processing request sent by a user terminal, obtaining user information of a target user in the business processing request, and generating the request based on the user information; Obtaining user information from the credit data processing request, and querying a local database based on the user information to obtain the credit data of the target user uploaded by each sub-server; wherein the credit data corresponding to each sub-server is determined by the sub-server inputting historical transaction data with the target user into a locally deployed credit risk model; Determine the historical prediction accuracy corresponding to each sub-server; Determine the comprehensive credit data corresponding to the target user based on the historical prediction accuracy and credit data corresponding to each sub-server; The comprehensive credit data corresponding to the target user is sent to the target sub-server, so that the target sub-server determines the business handling strategy for the target user based on the comprehensive credit data, wherein the target sub-server is the sub-server that generates the credit data processing request.

2. The method according to claim 1, characterized in that Determining the historical prediction accuracy corresponding to each sub-server includes: If there is no historical credit data processing request, the initial prediction accuracy rate corresponding to the credit risk model locally deployed on each sub-server is determined as the historical prediction accuracy rate corresponding to each sub-server, wherein the initial prediction accuracy rate corresponding to the credit risk model is determined using a preset test set; If there is a historical credit data processing request, based on obtaining the historical user information corresponding to the historical target user in the historical credit data processing request, the local database is queried according to the historical user information to obtain the historical credit data of the historical target user uploaded by each sub-server, and the historical comprehensive credit data corresponding to the historical target user is determined. The historical prediction accuracy corresponding to each sub-server is determined through the historical comprehensive credit data corresponding to the historical target user and the historical credit data about the historical target user uploaded by each sub-server.

3. The method according to claim 1, characterized in that The determining of the comprehensive credit data corresponding to the target user based on the historical prediction accuracy and credit data corresponding to each sub-server includes: Determine the historical prediction accuracy corresponding to each sub-server as the first weight corresponding to each sub-server; Based on the first weight corresponding to each sub-server, performing a weighted summation on the credit data corresponding to each sub-server to obtain a summation result; Determining a target number of credit data that meets preset conditions in the credit data corresponding to each sub-server; The ratio between the sum result and the target number is determined as the comprehensive credit data corresponding to the target user.

4. The method according to claim 1, wherein Before querying the local database according to the user information, the method further includes: Construct a local database, wherein the local database includes user information corresponding to each user and credit data of each user uploaded by each sub-server; wherein the credit data of each user corresponding to each sub-server is determined after the sub-server inputs historical transaction data with each user into a locally deployed credit risk model.

5. The method according to claim 4, characterized in that After building the local database, it also includes: According to the preset processing cycle, the new credit data of each user uploaded by each sub-server in the current processing cycle is updated to the local database.

6. The method according to claim 1, wherein Before receiving the credit data processing request sent by any sub-server, the method further includes: Build learning models; Sending the learning model to each sub-server so that each sub-server deploys the learning model, and inputting historical transaction data with the user into the learning model so that the learning model outputs the credit data corresponding to the user; Obtaining gradient encrypted data sent by each sub-server, wherein the gradient encrypted data is obtained by each sub-server adding noise data to the gradient data of the model parameters in the learning model and performing encryption calculation using a preset encryption algorithm; Based on the gradient encrypted data sent by each sub-server, the learning model is updated to obtain a new learning model; Repeating the steps of sending the learning model to each sub-server and updating the learning model based on the gradient encrypted data sent by each sub-server until the loss values ​​corresponding to the credit data output by the learning models deployed on each sub-server converge; The most recently updated learning model is determined as the credit scoring model, and the credit scoring model is sent to each sub-server so that each sub-server deploys the credit scoring model locally.

7. The method according to claim 6, characterized in that The updating of the learning model based on the gradient encrypted data sent by each sub-server to obtain a new learning model includes: Determine a second weight corresponding to each sub-server; Based on the second weight corresponding to each sub-server, weighted aggregation is performed on the gradient encrypted data sent by each sub-server to obtain a corresponding aggregation result; Performing homomorphic decryption on the aggregation result to obtain the corresponding aggregation gradient; The learning model is updated based on the aggregated gradient to obtain an updated new learning model.

8. The method according to claim 7, characterized in that The updating of the learning model based on the aggregated gradient to obtain an updated new learning model includes: Based on the aggregated gradient, the learning model is updated using a gradient descent method to obtain an updated new learning model.

9. A business processing device based on a credit processing system, characterized in that: A central server applied to the credit processing system, wherein the credit processing system comprises a central server and a plurality of sub-servers, wherein the plurality of sub-servers are respectively deployed in various banking institutions, including: a receiving module, configured to receive a credit data processing request sent by any sub-server, wherein the credit data processing request is a business processing request received by any sub-server from a user terminal, obtain user information of a target user in the business processing request, and generate the request based on the user information; The processing module is further configured to obtain user information in the credit data processing request, and query a local database based on the user information to obtain the credit data of the target user uploaded by each sub-server; wherein the credit data corresponding to each sub-server is determined by the sub-server inputting historical transaction data with the target user into a locally deployed credit risk model; A processing module, used to determine the historical prediction accuracy corresponding to each sub-server; The processing module is further configured to determine the comprehensive credit data corresponding to the target user based on the historical prediction accuracy and credit data corresponding to each sub-server; The sending module is used to send the comprehensive credit data corresponding to the target user to the target sub-server, so that the target sub-server determines the business processing strategy for the target user based on the comprehensive credit data, wherein the target sub-server is the sub-server that generates the credit data processing request.

10. An electronic device, characterized in that: include: a processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 8 when executed by a processor.

12. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 8 when being executed by a processor.

Images