Credit card risk control method, device, medium and system

Abstract

The invention provides a credit card risk control method, device, medium and system, and the method comprises the steps: constructing and updating a multi-dimensional user portrait, the multi-dimensional user portrait comprises a risk type of a credit card holder determined on the basis of real-time transaction flow data, multi-source credit investigation data, behavior biological characteristic data and social graph data under the condition of user authorization; generating a risk user management list according to the risk type; and under the condition that the repayment event is detected, adjusting the credit line of the card holder of the credit card according to the information in the risk user management list. The problem that risk intervention of the credit card depends on manual judgment in the prior art, so that the risk control effect is difficult to meet the expectation is solved.

Description

Technical Field

[0001] This application relates to the field of financial management technology, and more specifically, to a risk control method for credit cards, a risk control device for credit cards, a computer-readable storage medium, and a risk control system for credit cards. Background Technology

[0002] Credit card business faces multiple challenges such as credit risk, fraud risk, and misapproval risk. Traditional credit card risk control systems suffer from a lack of strategy, specifically in that risk intervention relies on manual judgment, credit limit reduction operations lack differentiated configuration, and high-quality customers are easily harmed. Summary of the Invention

[0003] The main objective of this application is to provide a risk control method, a risk control device, a computer-readable storage medium, and a risk control system for credit cards, so as to at least solve the problem that in related technologies, credit card risk intervention relies on manual judgment, which leads to the risk control effect being difficult to meet expectations.

[0004] To achieve the above objectives, according to one aspect of this application, a risk control method for credit cards is provided, comprising: constructing and updating a multi-dimensional user profile, wherein the multi-dimensional user profile includes a risk type of a credit card holder determined based on real-time transaction flow data, multi-source credit data, behavioral biometric data, and social graph data with the user's authorization; generating a risk user management list based on the risk type; and adjusting the credit limit of the credit card holder based on information in the risk user management list upon detecting a repayment event.

[0005] Optionally, the method further includes: during the process of constructing a multi-dimensional user profile, using a Kafka real-time message queue to analyze the transaction stream data to determine whether the cardholder's transaction behavior is abnormal; and using knowledge graph technology to analyze the cardholder's social network relationship data to determine the social graph data, wherein the social network relationship data includes the fund transfers between the cardholder and blacklisted accounts and the proportion of blacklisted accounts in the cardholder's contact list.

[0006] Optionally, constructing a multi-dimensional user profile includes: using a dynamic scoring model based on at least XGBoost and Transformer architectures to process real-time transaction stream data, multi-source credit data, behavioral biometric data, and social graph data to construct the multi-dimensional user profile.

[0007] Optionally, at least a dynamic scoring model based on the XGBoost and Transformer architecture is used to process real-time transaction stream data, multi-source credit data, behavioral biometric data, and social graph data to construct the multi-dimensional user profile. This includes: constructing an expert rule base, where data in the expert rule base is used to restrict feature selection and weight allocation of the dynamic scoring model; and using the dynamic scoring model based on the XGBoost and Transformer architecture and the expert rule base to process real-time transaction stream data, multi-source credit data, behavioral biometric data, and social graph data to construct the multi-dimensional user profile.

[0008] Optionally, adjusting the credit limit of a credit card holder based on information in the risk user management list includes: processing the information in the risk user management list using a neural network model to obtain a target credit limit; and adjusting the cardholder's credit limit to the target credit limit.

[0009] Optionally, adjusting the credit limit of a credit card holder based on information in the risk user management list includes: determining a target credit limit from a preset template library based on information in the risk user management list, wherein the preset template library stores credit limits corresponding to multiple preset templates; and adjusting the cardholder's credit limit to the target credit limit.

[0010] Optionally, the method further includes: if the cardholder is detected to have a target card usage behavior, adjusting the cardholder's card usage eligibility status to unqualified to use the card, wherein the target card usage behavior includes overdue payments.

[0011] According to another aspect of this application, a risk control device for a credit card is provided, comprising: a construction unit for constructing and updating a multi-dimensional user profile, the multi-dimensional user profile including a risk type of the credit card holder determined based on real-time transaction flow data, multi-source credit data, behavioral biometric data, and social graph data with user authorization; a first processing unit for generating a risk user management list based on the risk type; and a second processing unit for adjusting the credit limit of the credit card holder based on information in the risk user management list upon detection of a repayment event.

[0012] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform any of the methods described.

[0013] According to another aspect of this application, a credit card risk control system is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including methods for performing any one of the methods described.

[0014] Applying the technical solution of this application, the user profile is not only based on traditional credit data, but also combines real-time transaction flow data, behavioral biometric data, and social graph data, making the user profile more comprehensive and accurate. Real-time transaction flow data can capture users' immediate consumption behavior, such as abnormal large transactions and sudden changes in merchant type; behavioral biometric data provides insights into users' operating habits, such as rapid clicks to submit and abnormal swiping patterns, which may indicate potential risky behaviors; social graph data helps identify implicit connections between users, such as the proportion of blacklisted contacts and closed-loop paths of fund flows. These data points constitute a three-dimensional user risk profile. By regularly updating and automatically analyzing these profiles, the system can accurately determine the risk status of cardholders without requiring manual review of large amounts of data, greatly improving the efficiency and accuracy of risk identification. Based on multi-dimensional user profiles, the system can automatically generate a list of risky users for management, eliminating the subjectivity and misjudgment that may result from manual judgment. The list not only identifies high-risk users but also includes specific credit limit reduction strategies and target limits for different risk types. These strategies and target limits are dynamically adjusted by the system based on the latest data analysis, rather than fixed rules or a single standard. This means the system can make the most appropriate risk control decisions based on the cardholder's latest situation, avoiding delays and inconsistencies that may occur in manual decision-making. When key transaction behaviors such as repayment events are detected, the system automatically checks the high-risk user management list to determine whether the cardholder's credit limit needs to be adjusted. This process is automated, eliminating the need for manual review, thus significantly shortening response time and improving processing efficiency. More importantly, this event-driven mechanism ensures the immediacy and effectiveness of the strategies, enabling action to be taken immediately upon the occurrence of risk, effectively preventing further escalation of risk, while also reducing unnecessary interference with normal cardholders. This solves the problem in related technologies where credit card risk intervention relies on manual judgment, leading to unsatisfactory risk control results. Attached Figure Description

[0015] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0016] Figure 1A flowchart illustrating a risk control method for a credit card according to an embodiment of this application is shown.

[0017] Figure 2 A structural block diagram of a credit card risk control device provided according to an embodiment of this application is shown. Detailed Implementation

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0019] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0020] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0021] As described in the background section, credit card business faces multiple challenges such as credit risk, fraud risk, and mis-approval risk. Traditional credit card risk control systems suffer from a lack of strategy, specifically in that risk intervention relies on manual judgment, credit limit reduction operations lack differentiated configuration, and high-quality customers are easily harmed. To address the problem that credit card risk intervention relies on manual judgment, resulting in unsatisfactory risk control effects, embodiments of this application provide a credit card risk control method, a credit card risk control device, a computer-readable storage medium, and a credit card risk control system.

[0022] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0023] This embodiment provides a risk control method for credit cards. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Also, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0024] Figure 1 This is a flowchart of a credit card risk control method according to an embodiment of this application. Figure 1 As shown, the method includes the following steps:

[0025] Step S101: Construct and update a multi-dimensional user profile, which includes the risk type of a credit card holder determined based on real-time transaction flow data, multi-source credit data, behavioral biometric data and social graph data with the user's authorization.

[0026] Step S102: Generate a risk user management list based on the above risk types;

[0027] Step S103: Upon detecting a repayment event, adjust the credit limit of the credit card holder based on the information in the aforementioned risk user management list.

[0028] First, data related to credit card users is collected from multiple data sources, including real-time transaction streams, multi-source credit data, behavioral biometric data, and social graph data. This data requires preprocessing, such as cleaning, formatting, and standardization, to ensure data quality and consistency. Next, feature engineering is performed to extract meaningful features from the collected data, such as merchant type mutation detection, debt ratio, abnormal transaction behavior, and the proportion of blacklisted contacts. These features are used to train machine learning models, such as those based on XGBoost and Transformer architectures, to assess users' risk levels. Simultaneously, expert rules are defined and optimized to capture risk signals that the models may miss. The trained models and expert rules are then used to identify and classify users based on risk. According to the output risk score and rules, users are labeled with risk tags, such as incorrectly approved customers, fraudulent customers, or cash-out customers. These tags form the basis of risk user management. Based on the risk classification results, the system automatically generates a risk user management list. Each entry includes an account number, risk type code, processing method code, target credit limit, reason for processing, and processing status code. Business personnel can adjust the specific parameters of each entry according to actual needs to customize the strategy. An event capture layer is established to monitor events such as changes in the status of misused cards. When an event is triggered, the decision logic layer automatically executes corresponding risk control measures, such as adjusting the fixed credit limit, based on the preset processing methods and target limits in the list. If multiple lists simultaneously affect a single account, the system adopts pre-defined conflict resolution principles (such as "lowest target limit priority") and resolves conflicts through a decision tree. Simultaneously, all decision-making processes are recorded, including the time of the credit limit reduction operation, the model version used, and the strategy configuration, ensuring traceability and compliance. The list generation and processing mechanism is dynamic and requires continuous optimization and iteration based on new data and feedback to improve the accuracy of risk identification, reduce misjudgments, and enhance user experience.

[0029] XGBoost (eXtreme Gradient Boosting) is a machine learning algorithm primarily used for regression and classification tasks, excelling particularly in handling large-scale datasets. It's an efficient implementation of the Gradient Boosting framework, generating a powerful predictive model by integrating multiple weak predictive models (typically decision trees). XGBoost's core advantages lie in its speed and performance. It accelerates feature value sorting using an approximate histogram method, reducing computational costs, and optimizes the traditional GBDT (Gradient Boosting Decision Trees) algorithm, allowing for parallel processing and further improving training efficiency.

[0030] The Transformer architecture is suitable for scenarios such as sequence translation and text understanding. Unlike traditional RNNs (Recurrent Neural Networks) and LSTMs (Long Short-Term Memory Networks), the Transformer completely abandons the recursive structure and instead relies on a self-attention mechanism to process different parts of the input sequence.

[0031] In the steps described above, this application not only bases user profiling on traditional credit data but also incorporates real-time transaction flow data, behavioral biometric data, and social graph data, making the user profiles more comprehensive and accurate. Real-time transaction flow data captures users' immediate consumption behavior, such as unusually large transactions or sudden changes in merchant type; behavioral biometric data provides insights into user operating habits, such as rapid clicks to submit or abnormal swiping patterns, which may indicate potential risky behaviors; social graph data helps identify implicit connections between users, such as the proportion of blacklisted contacts and closed-loop paths of fund flows. These data points constitute a three-dimensional user risk profile. By regularly updating and automatically analyzing these profiles, the system can accurately determine the risk status of cardholders without requiring manual review of large amounts of data, greatly improving the efficiency and accuracy of risk identification. Based on multi-dimensional user profiles, the system can automatically generate a list of risky users for management, eliminating the subjectivity and misjudgment that may result from manual judgment. The list not only identifies high-risk users but also includes specific credit limit reduction strategies and target limits for different risk types. These strategies and target limits are dynamically adjusted by the system based on the latest data analysis, rather than fixed rules or a single standard. This means the system can make the most appropriate risk control decisions based on the cardholder's latest situation, avoiding delays and inconsistencies that may occur in manual decision-making. When key transaction behaviors such as repayment events are detected, the system automatically checks the high-risk user management list to determine whether the cardholder's credit limit needs to be adjusted. This process is automated, eliminating the need for manual review, thus significantly shortening response time and improving processing efficiency. More importantly, this event-driven mechanism ensures the immediacy and effectiveness of the strategies, enabling action to be taken immediately upon the occurrence of risk, effectively preventing further escalation of risk, while also reducing unnecessary interference with normal cardholders. This solves the problem in related technologies where credit card risk intervention relies on manual judgment, leading to unsatisfactory risk control results.

[0032] In one embodiment of this application, the method further includes: during the process of constructing a multi-dimensional user profile, using a Kafka real-time message queue to analyze the transaction stream data to determine whether there are any abnormalities in the cardholder's transaction behavior; and using knowledge graph technology to analyze the cardholder's social network relationship data to determine the social graph data, wherein the social network relationship data includes the fund transfers between the cardholder and blacklisted accounts and the proportion of blacklisted accounts in the cardholder's contact list.

[0033] Specifically, Kafka is a distributed, publish-subscribe-based message queue service. Kafka's real-time message queue processes transaction stream data: as a real-time message queue, Kafka is responsible for quickly and reliably collecting and transmitting all transaction data from the credit card transaction system, including transaction time, amount, merchant type, and geographical location, ensuring data real-time performance and integrity. On the Kafka consumer side, stream processing frameworks (such as Apache Flink or Spark Streaming) are used to process the transaction stream data in real time, calculating various risk characteristics, such as high-frequency transactions, large-amount transactions, transactions at abnormal times, and sudden changes in merchant type. These characteristics help identify whether cardholder transaction behavior is abnormal. Based on the real-time calculated transaction characteristics, they are compared with preset risk models or rules to quickly detect any abnormal transaction behavior and immediately feed the abnormal information back to the risk control system, triggering further risk control measures.

[0034] Knowledge graph technology analyzes social network relationship data: It integrates cardholders' social network relationship data, including transaction records and contact lists, to construct a graph network containing cardholders, contacts, and blacklisted accounts. Nodes in the graph represent various types of accounts, and edges represent fund flows or connections. Through graph analysis techniques, such as graph algorithms (e.g., PageRank, label propagation algorithms), the extent of fund transactions between cardholders and blacklisted accounts, as well as the proportion of blacklisted accounts in their contact lists, can be assessed, thereby calculating the potential risk score of the cardholder's social graph. Knowledge graphs can also help uncover risk transmission paths, such as closed-loop fund paths, which contributes to a deeper understanding of risk propagation mechanisms and allows for timely measures to prevent risk spread.

[0035] Kafka ensures real-time processing of transaction data, while knowledge graph technology can uncover hidden risks from social network relationships. The combination of the two makes risk identification more timely and accurate. Through comprehensive analysis of transaction flow data and social graph data, multi-layered risks can be identified, including fraud risk, cash-out risk, and potential credit risk, providing a comprehensive risk assessment perspective. Knowledge graph technology can clearly demonstrate the risk transmission path, enhancing the transparency and interpretability of system decisions, while ensuring that risk control measures comply with regulatory requirements. Accurate risk identification can avoid misjudging high-quality customers, reduce unnecessary risk control interventions, thereby optimizing the overall customer experience while maintaining the security of bank funds. User profile building and updating is a dynamic process; Kafka and knowledge graph technologies support real-time data updates, enabling the system to quickly adapt to environmental changes and improve the flexibility and adaptability of risk control.

[0036] In one embodiment of this application, constructing a multi-dimensional user profile includes: using a dynamic scoring model based on at least XGBoost and Transformer architectures to process real-time transaction stream data, multi-source credit data, behavioral biometric data, and social graph data to construct the aforementioned multi-dimensional user profile.

[0037] Specifically, an expert rule base is constructed, and the data in the expert rule base is used to restrict the feature selection and weight allocation of the dynamic scoring model. The dynamic scoring model based on XGBoost and Transformer architecture, along with the expert rule base, is used to process real-time transaction stream data, multi-source credit data, behavioral biometric data, and social graph data to construct the multi-dimensional user profile.

[0038] Expert rule base construction: Experience-based rules are provided by experts and analysts in the field of credit card risk control, such as transaction risks for specific merchant types, abnormal patterns of consecutive large transactions, and the relationship between sudden changes in geographical location and risk. The collected rules are converted into a computer-understandable form, for example: "If a cardholder has made more than M large transactions in the past N hours, and the transaction locations are in three different regions, the risk level is upgraded to high risk." The expert rule base should include rules for various risk scenarios to ensure its comprehensiveness and relevance. Furthermore, the rule base needs to be updated regularly to adapt to the ever-changing market environment.

[0039] Training and optimization of the dynamic scoring model: A dynamic scoring model using XGBoost and Transformer architectures is employed, capable of handling high-dimensional, non-linear, and time-series data. Real-time transaction stream data, multi-source credit data, behavioral biometric data, and social graph data are used as input features for preprocessing and feature selection.

[0040] Constraints in model training: During model training, data from the expert rule base is used to guide feature selection and weight allocation. For example, the model may learn that certain features contribute little to risk assessment but are marked as key features in the expert rule base. In such cases, the model will be forced to retain these features during training and adjust their weights to ensure consistency between the model's decisions and expert experience.

[0041] Model validation and iteration: Validate the accuracy and reliability of the model using historical data, and adjust the model parameters based on the validation results, including the degree of influence of expert rules, to optimize model performance.

[0042] Multi-dimensional user profile construction: Data processed by a dynamic scoring model is integrated with expert rule results to construct user profiles. These profiles should be updated in real time to reflect users' latest credit status and behavioral patterns, improving the timeliness and accuracy of risk control. The results of expert rules and model predictions are used to identify high-risk users, further categorized into different risk types, such as mis-approval risk, fraud risk, and cash-out risk.

[0043] The combination of expert rule bases and machine learning models ensures that model decisions are based not only on statistical analysis of data but also on in-depth insights from domain experts and business rules, improving the accuracy and reliability of risk assessment. By restricting the model's feature selection and weight allocation, the expert rule base makes the model's decision logic more transparent, facilitating business personnel's understanding of the reasons behind the model's decisions and enhancing the system's interpretability, which is crucial for regulatory compliance and customer dispute resolution. User profiles based on expert rules and model prediction results enable customized risk control strategies for different types of users, avoiding a one-size-fits-all approach, balancing risk control and customer experience, and helping to maintain good customer relationships. Real-time transaction stream data processing and automated response mechanisms significantly improve the response speed of risk control, optimizing it from hours to seconds, reducing the workload of manual review, and improving overall risk control efficiency.

[0044] In one embodiment of this application, adjusting the credit limit of a credit card holder based on information in the aforementioned risk user management list includes: processing the information in the aforementioned risk user management list using a neural network model to obtain a target credit limit; and adjusting the credit limit of the aforementioned cardholder to the aforementioned target credit limit.

[0045] Neural network models, by learning complex patterns in historical data, can capture the non-linear relationships between multi-dimensional information such as user behavior, transaction habits, and credit history, thereby more accurately predicting and assessing cardholders' creditworthiness. This method is more adaptable to data changes than traditional rule-based methods, improving the accuracy and rationality of credit limit adjustments. Neural network models can calculate personalized target credit limits based on each cardholder's specific circumstances. This means that even with the same risk label, different cardholders will have different credit limit adjustments depending on their specific credit performance and card usage behavior, thus providing more personalized financial services and reducing misjudgments of high-quality customers. Because neural network models can make predictions in real time with new data input, they can support dynamic adjustments to credit limits. This immediate feedback mechanism allows credit card risk control systems to quickly respond to market changes and changes in cardholder credit status, improving the overall system's flexibility and responsiveness. The automated processing function of neural network models reduces the need for manual review, lowers operating costs, and avoids biases and errors caused by human factors, improving efficiency and decision consistency.

[0046] In one embodiment of this application, adjusting the credit limit of a credit card holder based on the information in the aforementioned risk user management list includes: determining a target credit limit from a preset template library based on the information in the aforementioned risk user management list, wherein the preset template library stores credit limits corresponding to multiple preset templates; and adjusting the credit limit of the aforementioned cardholder to the aforementioned target credit limit.

[0047] The pre-set template library stores multiple templates that can be flexibly selected based on different risk types and levels, enabling personalized and refined management for different cardholders. This means that banks or financial institutions can quickly adjust credit limits to the most suitable level based on each cardholder's current risk profile, avoiding misjudgments and excessive restrictions that may result from a one-size-fits-all risk management strategy. The automated credit limit adjustment process reduces the need for manual intervention, significantly improving processing speed and efficiency. When a risk is detected, the system can instantly select the most suitable credit limit template from the library for a rapid response, avoiding the delays of traditional manual review and helping to prevent and control risks in a timely manner. Automated and template-based credit limit adjustments can significantly reduce the risk management costs for banks or financial institutions. The automated processing reduces reliance on manpower and lowers the additional costs caused by human error. At the same time, the use of pre-set templates makes the adjustment process more standardized and efficient. Adopting a multi-dimensional, customized management approach allows credit limits to be adjusted according to the cardholder's actual situation, ensuring the bank's risk control while avoiding excessive restrictions on high-quality customers, thus helping to maintain good customer relationships. For example, for customers who occasionally trigger risk rules but have good overall credit, a "conservative" template can be used to slightly reduce their credit limit instead of drastically cutting it, thus finding a balance between risk control and customer experience. The existence of a pre-set template library makes the credit limit adjustment process more transparent and traceable. Each template has clear adjustment criteria and standards, facilitating audits and compliance checks by regulatory agencies, while also providing cardholders with a clear explanation of the credit limit adjustment, increasing the transparency and rationality of the operation.

[0048] In one embodiment of this application, the method further includes: when the cardholder is detected to have a target card usage behavior, adjusting the cardholder's card usage eligibility status to unqualified to use the card, wherein the target card usage behavior includes overdue payments.

[0049] By monitoring and automatically adjusting card eligibility status in real time, banks can react immediately to potential credit risks and prevent them from escalating. For example, once a cardholder is detected to have overdue payments, the system can quickly set their card eligibility to "unqualified," preventing further overdrafts and thus effectively controlling the bank's bad debt risk. When a cardholder's eligibility is automatically adjusted to "unqualified," it typically triggers the bank's collection process. Since the system has already marked the account as high-risk, the collection department can prioritize these accounts, improving collection efficiency and recovery rates. Furthermore, this adjustment serves as an effective means of urging repayment, prompting cardholders to resolve overdue issues as soon as possible. This approach helps maintain a healthy credit card ecosystem, encouraging cardholders to repay on time and comply with credit rules. Automated adjustment of card eligibility status reduces the need for manual review and lowers operating costs. Manually checking each overdue payment and adjusting status for a large number of customers is extremely time-consuming. Automated processes can handle these situations efficiently, allowing employees to focus more on other complex issues or customer service. While this approach may negatively impact the user experience for cardholders with overdue payments in the short term, it can enhance the trust and stability of the entire credit card system in the long run. Users will perceive the system as fair and efficient, thus being more inclined to repay on time and avoid unnecessary restrictions, thereby improving the overall positive user experience.

[0050] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0051] This application also provides a credit card risk control device. It should be noted that the credit card risk control device of this application can be used to execute the credit card risk control method provided in this application. This device is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0052] The following describes the risk control device for credit cards provided in the embodiments of this application.

[0053] Figure 2 This is a schematic diagram of a credit card risk control device according to an embodiment of this application. Figure 2 As shown, the device includes:

[0054] The construction unit 21 is used to construct and update a multi-dimensional user profile, which includes the risk type of the credit card holder determined based on real-time transaction flow data, multi-source credit data, behavioral biometric data and social graph data with the user's authorization; the first processing unit 22 is used to generate a risk user management list based on the above risk type; the second processing unit 23 is used to adjust the credit limit of the credit card holder based on the information in the risk user management list when a repayment event is detected.

[0055] In one embodiment of this application, the above-mentioned apparatus further includes: a third processing unit used to use a Kafka real-time message queue to process the transaction stream data during the process of constructing a multi-dimensional user profile, in order to determine whether there are any abnormalities in the cardholder's transaction behavior; and a fourth processing unit used to use knowledge graph technology to analyze the cardholder's social network relationship data, and to determine the social graph data, wherein the social network relationship data includes the fund transactions between the cardholder and blacklisted accounts and the proportion of blacklisted accounts in the cardholder's contact list.

[0056] In one embodiment of this application, the construction unit includes: a fifth processing unit used to process real-time transaction stream data, multi-source credit data, behavioral biometric data and social graph data using a dynamic scoring model that employs at least XGBoost and Transformer architectures, in order to construct the aforementioned multi-dimensional user profile.

[0057] In one embodiment of this application, the fifth processing unit includes: a construction module for constructing an expert rule base, wherein the data in the expert rule base is used to restrict the feature selection and weight allocation of the dynamic scoring model; and a first processing module for using the dynamic scoring model based on the XGBoost and Transformer architecture and the expert rule base to process real-time transaction stream data, multi-source credit data, behavioral biometric data and social graph data to construct the multi-dimensional user profile.

[0058] In one embodiment of this application, the second processing unit includes: a second processing module for processing the information in the risk user management list using a neural network model to obtain a target credit limit; and a third processing module for adjusting the credit limit of the cardholder to the target credit limit.

[0059] In one embodiment of this application, the second processing unit includes: a fourth processing module for determining a target credit limit from a preset template library based on the information in the aforementioned risk user management list, wherein the preset template library stores credit limits corresponding to multiple preset templates; and a fifth processing module for adjusting the credit limit of the cardholder to the aforementioned target credit limit.

[0060] In one embodiment of this application, the above-mentioned device further includes: a sixth processing unit configured to adjust the cardholder's card eligibility status to uneligible when the target card usage behavior is detected, wherein the target card usage behavior includes overdue payment.

[0061] The aforementioned credit card risk control device includes a processor and a memory. The aforementioned building units, first processing unit, and second processing unit are all stored as program units in the memory. The processor executes these program units stored in the memory to achieve the corresponding functions. All of the aforementioned modules are located in the same processor; alternatively, the aforementioned modules may be located in different processors in any combination.

[0062] The processor contains a kernel, which retrieves the corresponding program unit from memory. One or more kernels can be configured, and adjusting kernel parameters can address the problem in related technologies where credit card risk intervention relies on manual judgment, leading to unsatisfactory risk control results.

[0063] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0064] This invention provides a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the credit card risk control method.

[0065] This invention provides a processor for running a program, wherein the program executes the risk control method for the credit card.

[0066] This invention provides a device including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs at least the following steps: constructing and updating a multi-dimensional user profile, wherein the multi-dimensional user profile includes, with user authorization, a risk type of a credit card holder determined based on real-time transaction flow data, multi-source credit data, behavioral biometric data, and social graph data; generating a risk user management list based on the risk type; and adjusting the credit limit of the credit card holder based on information in the risk user management list upon detecting a repayment event. The device described herein can be a server, PC, PAD, mobile phone, etc.

[0067] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program having at least the following method steps: constructing and updating a multi-dimensional user profile, wherein the multi-dimensional user profile includes a risk type of a credit card holder determined based on real-time transaction flow data, multi-source credit data, behavioral biometric data, and social graph data with the user's authorization; generating a risk user management list based on the aforementioned risk type; and adjusting the credit limit of the credit card holder based on information in the aforementioned risk user management list upon detecting a repayment event.

[0068] This application also provides a credit card risk control system, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include methods for performing any of the above-described methods.

[0069] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0070] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0071] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0072] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0073] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0074] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0075] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0076] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0077] The technical features of the above embodiments can be combined in any way. For the sake of brevity, 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.

[0078] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0079] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A risk control method for credit cards, characterized in that, include: Construct and update multi-dimensional user profiles, which include risk types of credit card holders determined based on real-time transaction flow data, multi-source credit data, behavioral biometric data, and social graph data with user authorization; Generate a risk user management list based on the risk type; Upon detection of a repayment event, the credit limit of the credit card holder is adjusted based on information from the risk user management list.

2. The method according to claim 1, characterized in that, The method further includes: In the process of building a multi-dimensional user profile, a Kafka real-time message queue is used to process the transaction stream data to determine whether there are any anomalies in the cardholder's transaction behavior. The social network relationship data of the cardholder is analyzed using knowledge graph technology to determine the social graph data, which includes the financial transactions between the cardholder and blacklisted accounts, as well as the proportion of blacklisted accounts in the cardholder's contact list.

3. The method according to claim 1, characterized in that, Construct multi-dimensional user profiles, including: At least a dynamic scoring model based on XGBoost and Transformer architecture is used to process real-time transaction stream data, multi-source credit data, behavioral biometric data, and social graph data to construct the multi-dimensional user profile.

4. The method according to claim 3, characterized in that, At least a dynamic scoring model based on XGBoost and Transformer architectures is used to process real-time transaction stream data, multi-source credit data, behavioral biometric data, and social graph data to construct the aforementioned multi-dimensional user profile, including: An expert rule base is constructed, and the data in the expert rule base is used to restrict the feature selection and weight allocation of the dynamic scoring model. A dynamic scoring model based on XGBoost and Transformer architecture, along with the expert rule base, is used to process real-time transaction stream data, multi-source credit data, behavioral biometric data, and social graph data to construct the multi-dimensional user profile.

5. The method according to claim 1, characterized in that, Based on the information in the aforementioned risk user management list, adjust the credit limit of the credit card holder, including: A neural network model is used to process the information in the risk user management list to obtain the target credit limit; Adjust the cardholder's credit limit to the target credit limit.

6. The method according to claim 1, characterized in that, Based on the information in the aforementioned risk user management list, adjust the credit limit of the credit card holder, including: Based on the information in the risk user management list, the target credit limit is determined from the preset template library, which stores credit limits corresponding to multiple preset templates; Adjust the cardholder's credit limit to the target credit limit.

7. The method according to claim 1, characterized in that, The method further includes: If the cardholder is found to have engaged in targeted card usage behavior, the cardholder's eligibility status will be adjusted to "unqualified to use"; the targeted card usage behavior includes overdue payments.

8. A risk control device for a credit card, characterized in that, include: The building unit is used to build and update a multi-dimensional user profile, which includes the risk type of a credit card holder determined based on real-time transaction flow data, multi-source credit data, behavioral biometric data and social graph data with the user's authorization. The first processing unit is used to generate a risk user management list based on the risk type. The second processing unit is used to adjust the credit limit of the credit card holder based on the information in the risk user management list when a repayment event is detected.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method according to any one of claims 1 to 7.

10. A risk control system for credit cards, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs comprising methods for performing any one of claims 1 to 7.

Images