Differentiated credit line calculation system and method
Through the differentiated credit line calculation system, using user portraits and weight decay factors and combined with risk assessment, personalized credit line configuration is achieved, solving the problem of flexibility and accuracy of financial institutions in credit line calculation, and increasing risks Control and customer satisfaction.
Patent Information
- Application Number
- CN202411039574.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-12-06
AI Technical Summary
Financial institutions find it difficult to accurately determine credit lines, and cannot meet the needs and risk preferences of different customers. The existing technology lacks flexibility and accuracy in credit lines calculations.
Using a differentiated credit limit calculation system, the user profile acquisition module, the timed update module, the risk level prediction module and the credit limit configuration module are used, and the weight decay factor and the credit limit coefficient are used to dynamically adjust the user profile and risk assessment to achieve personalized credit limit Configuration.
It improves the flexibility and accuracy of credit lines, enhances risk control and customer satisfaction, reduces the risk of financial crisis caused by long lending, and improves the competitiveness of financial institutions.
Smart Images

Figure CN119090620A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of computer information processing, and in particular to a differentiated credit limit calculation system and method. Background Art
[0002] Credit refers to the funds directly provided by financial institutions to non-financial institution users, or the guarantees made by financial institutions for the compensation and payment obligations that may arise from relevant economic activities, including on-balance sheet businesses such as loans, trade financing, bill financing, financial leasing, overdrafts, and various advances, as well as off-balance sheet businesses such as bill acceptance, issuance of letters of credit, guarantees, standby letters of credit, letter of credit confirmation, bond issuance guarantees, loan guarantees, asset sales with recourse, and unused irrevocable loan commitments. In simple terms, credit refers to the behavior of banks directly providing financial support to users, or guaranteeing the credit of users in relevant economic activities to third parties. The credit line refers to the maximum amount of money that a loan platform gives to individuals applying for loans within a period of time, and the borrower can use this money at any time during this period. The loan line refers to the amount actually borrowed by the borrower within the available amount given.
[0003] With the continuous development of the credit market and the intensification of market competition, financial institutions need to determine credit lines more accurately to meet the needs and risk preferences of different customers. Differentiated credit lines are designed to provide personalized loan lines to different customers by comprehensively considering various factors of customers, so as to improve the risk control level and customer satisfaction of credit business. Summary of the invention
[0004] The present invention aims to provide a differentiated credit limit calculation system and method, which effectively updates user portraits regularly through weight decay factors, obtains credit limits based on user portrait matching credit limit calculation models, configures differentiated credit limits more flexibly and quickly, and adjusts the credit limit coefficients obtained in combination with risk assessment to achieve credit limit control.
[0005] In order to achieve the above object, the present invention adopts the following technical scheme:
[0006] A differentiated credit limit calculation system, comprising:
[0007] User portrait acquisition module, used to obtain the user's credit application information and user portrait;
[0008] The timing update module is used to continuously correct and adjust the user profile based on the user behavior data using the weight attenuation factor; the calculation model of the weight attenuation factor is:
[0009] S t =S t0 *e-ki(t0-t)
[0010] Among them, S t is the current weight, S t0 is the original weight, ki is the cooling coefficient of label i, and t0-t is the time interval;
[0011] The risk level prediction module is used to retrieve the risk assessment model to evaluate and obtain the credit limit coefficient of the current user;
[0012] The credit limit configuration module is used to retrieve the user portrait and match the credit limit calculation model in the model configuration table, obtain the basic credit limit according to the credit limit calculation model, and correct the basic credit limit through the credit limit coefficient to obtain the credit limit.
[0013] The principles and advantages of this solution are: in actual application, the user's credit application information and user portrait are obtained as the basic information for credit limit calculation, and the user portrait is updated regularly to ensure the accuracy of credit; based on the weight decay factor, the calculation of each label model in the user portrait is gradually cooled down with the linear growth of time to reduce the influence of time factors and improve the practicality of each label in the user portrait; the user's risk assessment situation reflects the probability of user repayment to a certain extent, and the risk assessment model is evaluated to obtain the credit limit coefficient of the current user, and the credit limit coefficient is used as a reference factor for credit limit configuration to facilitate risk control; different types of users are suitable for different credit limit calculation models. Therefore, the credit limit calculation model is matched in the model configuration table based on the user portrait, and the basic credit limit is obtained according to the credit limit calculation model, and the basic credit limit is corrected by the credit limit coefficient to obtain the credit limit, and the credit limit calculation model is configured more flexibly to calculate the credit limit, so as to achieve differentiated control of the credit limit.
[0014] Preferably, as an improvement, the risk assessment model is a multi-loan box model, which boxes the user's behavior of applying for loans from multiple lending institutions within a preset time period to obtain a multi-model score, and obtains a credit limit coefficient based on the multi-model score.
[0015] Technical effect: Multiple borrowings can easily lead to debt accumulation problems. When the debt accumulates to a certain level, it becomes more difficult for borrowers to repay the loan, and may even lead to personal or corporate financial crisis. Using multiple borrowings as the basis for risk score assessment can help avoid financial crises for users.
[0016] Preferably, as an improvement, the cooling coefficient is dynamically adjusted according to the activity of the behavior data corresponding to the tag.
[0017] Technical effect: Each user has different performance in each tag. By dynamically adjusting the cooling coefficient, the accuracy of the user portrait can be improved.
[0018] Preferably, as an improvement, the model configuration table is a multidimensional data model, and the specified tags in the user portrait are used as various data dimensions.
[0019] Technical effect: User portraits usually contain multiple labels. Specifying labels as data dimensions facilitates dimensionality reduction to simplify the model configuration table.
[0020] Preferably, as an improvement, it also includes a credit limit upper and lower limit adjustment module for adjusting the upper and lower limits of the credit limit according to the channel source of the customer.
[0021] Technical effect: It is easy to improve the flexibility of credit line configuration, increase the experience of cooperative users, and reduce risks.
[0022] Preferably, as an improvement, a testing module is also included to randomly and evenly distribute users to several credit limit calculation models, evaluate post-loan performance, and optimize the model configuration table based on the post-loan performance.
[0023] Technical effect: It is easy to improve the accuracy of credit limit calculation model configuration.
[0024] Preferably, as an improvement, it also includes an emotion perception module, which is used to obtain the user's credit limit increase application and the user's credit limit increase application description, and obtain the user's demand keywords through natural language and semantic analysis technology;
[0025] The credit limit increase suggestion generation module determines whether the demand keywords touch upon sensitive scenarios. If so, it automatically generates a scenario-sensitive credit suggestion.
[0026] Technical effect: It makes it easier for decision makers to verify and modify credit limits based on credit recommendations, thereby improving user experience and helping to enhance the competitiveness of financial institutions.
[0027] Also included is a differentiated credit limit calculation method, which is applied to a differentiated credit limit calculation system. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 The figure is a schematic diagram of the structure of a differentiated credit limit calculation system according to an embodiment of the present invention. DETAILED DESCRIPTION
[0029] The following is further described in detail through specific implementation methods:
[0030] The embodiment is basically as shown in the attached Figure 1 As shown:
[0031] A differentiated credit limit calculation system, comprising:
[0032] The user portrait acquisition module is used to obtain the user's credit application information and user portrait; credit is triggered by the user's credit application; the user portrait summarizes and describes the user's behavioral data (i.e. personal information, interests, behavioral habits and other characteristics), which can help financial institutions understand and characterize their target user groups in a visual way. Using user portraits as basic information for credit limit calculation can help financial institutions grasp users more accurately, optimize credit limit configuration, and reduce decision-making risks.
[0033] The timing update module is used to continuously correct and adjust the user profile based on the user behavior data using the weight attenuation factor; the calculation model of the weight attenuation factor is:
[0034] S t =S t0 *e -ki(t0-t)
[0035] Among them, S t is the current weight, S t0 is the original weight, ki is the cooling coefficient of label i, and t0-t is the time interval. The heat of the label model in the user portrait will usually cool down with the linear growth of time, such as the user's income level, shopping consumption record, etc., and the reference value will decrease with the passage of time. The coefficients of the weight decay factor are different. Some labels are not even affected by time and are not considered during calculation. The weight decay factor facilitates the gradual cooling of the label models in the user portrait with the linear growth of time during calculation, so as to reduce the influence of the time factor and improve the practicality of each label in the user portrait. The cooling coefficient is dynamically adjusted according to the activity of the behavior data corresponding to the label. The user's activity in each label is determined by analyzing the data records of the user's social account, financial account, and e-commerce platform account. The cooling coefficient of each label is adjusted according to the activity. Specifically, the cooling coefficient can be found by comparing the preset cooling coefficient corresponding table with the statistical activity frequency, or the cooling coefficient can be obtained through a preset calculation model.
[0036] The risk level prediction module is used to retrieve the risk assessment model to evaluate the credit limit coefficient of the current user; the user risk assessment situation reflects the probability of the user's repayment to a certain extent, and the risk assessment model is used to evaluate the credit limit coefficient of the current user, and the credit limit coefficient is used as a reference factor for credit limit configuration to facilitate risk control; the risk assessment model is a multi-loan sub-box model,
[0037] Multiple borrowing means borrowing money from multiple channels, multiple financial institutions, and multiple methods at the same time. Multiple borrowing can easily lead to debt accumulation problems. When the debt accumulates to a certain extent, it becomes more difficult for the borrower to repay the loan, and even leads to a financial crisis for the individual or the company. Multiple borrowing often involves the situation of using new loans to repay old loans, or borrowing new loans for large-scale consumption due to insufficient current consumption capacity, with the principal and interest constantly accumulating. Using multiple borrowing as the basis for risk score assessment is convenient for avoiding the above-mentioned crises faced by users. The multiple borrowing binning model bins the behavior of users applying for loans from multiple lending institutions within a preset period of time to obtain a multiple model score. The corresponding relationship between the multiple model score and the credit coefficient is shown in Table 1:
[0038] Table 1
[0039] Long Model Credit factor A box model score (800 points or more) 2 B box model score (700 points - 800 points) 1.8 C box model score (600-700 points) 1.6 D box model score (500 points - 600 points) 1 E box model score (less than 500 points) 0.5
[0040] Different types of users are suitable for different credit limit calculation models. The credit limit configuration module is used to retrieve user portraits and match the credit limit calculation model in the model configuration table. The model configuration table is a multidimensional data model, which uses the specified tags in the user portrait as each data dimension to facilitate dimensionality reduction to simplify the model configuration table. The basic credit limit is obtained according to the credit limit calculation model, and the basic credit limit is corrected by the credit limit coefficient to obtain the credit limit. The credit limit calculation model is configured more flexibly to calculate the credit limit and realize differentiated control of the credit limit. The correction method is:
[0041] E fin =θ x *E ag
[0042] Among them, E fin is the credit limit, θ x is the credit factor, E ag The basic credit limit.
[0043] It also includes a credit limit upper and lower limit adjustment module, which is used to adjust the upper and lower limits of the credit limit according to the customer's channel source. For example, when the proportion of high-quality customers corresponding to the channel source is large, the upper credit limit will be appropriately increased. When the proportion of high-quality customers corresponding to the channel source is small, the upper and lower credit limits of the customer will be changed. By adjusting the upper and lower credit limits through channel sources, it is convenient to improve the flexibility of credit limit configuration, increase the experience of cooperative users, and reduce risks at the same time.
[0044] It also includes a test module, which randomly and evenly distributes users to several credit limit calculation models, evaluates post-loan performance, and optimizes the model configuration table based on post-loan performance to improve the accuracy of credit limit calculation model configuration. Specifically, it evaluates the overdue situation of not repaying within 30 days after the loan is issued and lasting for more than 4 months, suspends the allocation of credit limit calculation models with overdue situations exceeding the threshold, and replaces the current credit limit calculation model in the model configuration table with the nearest credit limit calculation model.
[0045] In order to enhance the user experience and help improve the competitiveness of financial institutions, this application also takes into account the needs of users in specific situations, such as emergency medical expenses, education investment, and entrepreneurial funds. It also includes an emotional perception module for obtaining user credit limit applications and descriptions, and obtaining user demand keywords through natural language and semantic analysis technology; a credit limit recommendation generation module determines whether the demand keywords touch on sensitive scenarios, and automatically generates scenario-sensitive credit recommendations when it is judged to be so. It is convenient for decision makers to verify and modify credit limits based on credit recommendations combined with real-time market data and economic indicators.
[0046] Also included is a differentiated credit limit calculation method, which is applied to a differentiated credit limit calculation system.
[0047] The above is only an embodiment of the present invention, and the common knowledge such as the known specific technical solutions and / or characteristics in the solution is not described in detail here. It should be pointed out that for those skilled in the art, without departing from the technical solution of the present invention, several modifications and improvements can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.
Claims
1. A differentiated credit limit calculation system, characterized in that: include: User portrait acquisition module, used to obtain the user's credit application information and user portrait; The timing update module is used to regularly correct and adjust the user profile based on the user behavior data using the weight attenuation factor; the calculation model of the weight attenuation factor is: S t =S t0 *e -ki(t0-t) Among them, S t is the current weight, S t0 is the original weight, ki is the cooling coefficient of label i, and t0-t is the time interval; The risk level prediction module is used to retrieve the risk assessment model to evaluate and obtain the credit limit coefficient of the current user; The credit limit configuration module is used to retrieve the user portrait and match the credit limit calculation model in the model configuration table, obtain the basic credit limit according to the credit limit calculation model, and correct the basic credit limit through the credit limit coefficient to obtain the credit limit.
2. A differentiated credit limit calculation system according to claim 1, characterized in that: The risk assessment model is a multi-loan box model, which boxes the user's behavior of applying for loans from multiple lending institutions within a preset period of time, obtains a multi-model score, and obtains a credit limit coefficient based on the multi-model score.
3. A differentiated credit limit calculation system according to claim 1, characterized in that: The cooling coefficient is dynamically adjusted according to the activity of the behavior data corresponding to the tag.
4. A differentiated credit limit calculation system according to claim 1, characterized in that: The model configuration table is a multidimensional data model, which uses the specified tags in the user portrait as each data dimension.
5. A differentiated credit limit calculation system according to claim 1, characterized in that: It also includes a credit limit upper and lower limit adjustment module, which is used to adjust the upper and lower limits of the credit limit based on the customer's channel source.
6. A differentiated credit limit calculation system according to claim 1, characterized in that: It also includes a testing module, which randomly and evenly distributes users to several credit limit calculation models, evaluates post-loan performance, and optimizes the model configuration table based on the post-loan performance.
7. A differentiated credit limit calculation system according to claim 1, characterized in that: It also includes an emotion perception module, which is used to obtain the user's credit limit increase application and the user's credit limit increase application description, and obtain the user's demand keywords through natural language and semantic analysis technology; The credit limit increase suggestion generation module determines whether the demand keywords touch upon sensitive scenarios. If so, it automatically generates a scenario-sensitive credit suggestion.
8. A differentiated credit limit calculation method, characterized by: Applied to a differentiated credit limit calculation system.
