Method and device for realizing joint supervision based on credit grading

Abstract

The invention discloses a method and a device for realizing joint supervision based on credit grading, and relates to the technical field of government affair service management. Comprising the steps of 1, constructing a credit assessment model; 2, creating a joint supervision trigger mechanism; 3, making a differential supervision strategy; 4, dynamically adjusting a credit rating according to supervision result feedback, and 5, executing joint supervision: generating a joint task containing a check list and an executor, and distributing the joint task to a joint supervision department. All supervision departments automatically summarize inspection results to form a unified supervision report; according to the invention, a cross-department joint supervision process is reconstructed through credit grading and an intelligent algorithm.

Description

Technical Field

[0001] The present invention discloses a method and device for realizing joint supervision based on credit grading, which relates to the technical field of government service management. Background Art

[0002] In the prior art, some systems simply classify through credit scoring, but do not deeply combine with joint supervision, and cannot achieve precise and dynamic cross-departmental collaboration. In the existing supervision mode, the supervision of market entities by supervision departments mainly relies on manual experience or fixed-cycle inspections, and there are some technical defects: (1) Most existing credit evaluation systems adopt a static scoring mechanism for simple classification, lacking the tracking and analysis of dynamic behavior data.

[0003] (2) Extensive joint supervision: Cross-departmental joint inspections rely on manual experience judgment, with strong subjectivity and prone to missing high-risk objects.

[0004] (3) Mismatch between resource allocation and risk: High-credit enterprises and low-credit enterprises adopt the same inspection frequency, which is likely to cause waste of supervision resources on compliant entities, while the inspection intensity of real high-risk entities is insufficient.

[0005] (4) Data islands and low collaboration efficiency: Credit data of each department is scattered in independent systems, making it difficult to form a unified risk view. Summary of the Invention

[0006] In view of the problems of the prior art, the present invention provides a method and device for realizing joint supervision based on credit grading, and reconstructs the cross-departmental joint supervision process through credit grading and intelligent algorithms.

[0007] The specific solution proposed by the present invention is: The present invention provides a method for realizing joint supervision based on credit grading, including: Step 1: Construct a credit evaluation model: Collect supervision behavior data to construct a credit evaluation database, and use machine learning algorithms to generate a credit scoring model for dynamically outputting credit levels. Step 2: Create a joint supervision trigger mechanism: Initiate a supervision plan, formulate inspection tasks, automatically call the credit level data of supervision entities, generate joint supervision task suggestions, and push them to the matching supervision departments. Step 3: Formulate differential supervision strategies: According to the credit levels of supervision entities and joint supervision task suggestions, formulate supervision strategies for supervision entities at different levels. Step 4: Dynamically adjust the credit level according to the feedback of supervision results: Update the supervision results to the credit scoring model in real time, recalculate the credit level, and synchronize it to the supervision departments. Step 5: Implement joint supervision: Generate joint tasks including inspection checklists and implementing personnel and dispatch them to the joint supervision department. Each supervision department will automatically summarize the inspection results to form a unified supervision report.

[0008] Furthermore, the supervision behavior data collected in Step 1 of the method for realizing joint supervision based on credit grading includes penalty records, compliance operation data, complaint information, and industry risk characteristics. Use the random forest or XGBoost in the machine learning algorithm to generate a credit scoring model, and set the grading rules for credit levels as dividing credit levels by setting thresholds, where level A ≥ 90 points, level B is 70 - 89 points, level C is 50 - 69 points, and level D < 50 points.

[0009] Furthermore, in Step 2 of the method for realizing joint supervision based on credit grading, conduct a credit risk analysis on the supervision subject. When the credit level of the supervision subject is level C / D, trigger the joint supervision matching engine to screen associated supervision departments for generating joint supervision task suggestions.

[0010] Furthermore, in Step 4 of the method for realizing joint supervision based on credit grading, automatically send a risk warning to the associated supervision department for the supervision subject whose credit score drops sharply after recalculating the credit level.

[0011] Furthermore, in Step 5 of the method for realizing joint supervision based on credit grading, adjust the parameters of the credit scoring model based on the task completion rate and problem discovery rate in the supervision report to optimize the credit scoring model.

[0012] The present invention also provides a device for realizing joint supervision based on credit grading, including a model management module, a supervision trigger module, a policy management module, an adjustment module, and an execution module. The model management module constructs a credit assessment model: Collect supervision behavior data to construct a credit assessment database, and use a machine learning algorithm to generate a credit scoring model for dynamically outputting credit levels. The supervision trigger module creates a joint supervision trigger mechanism: Initiate a supervision plan, formulate inspection tasks, automatically call the credit level data of the supervision subject, generate joint supervision task suggestions, and push them to the matching supervision department. The policy management module formulates differentiated supervision policies: According to the credit level of the supervision subject and the joint supervision task suggestions, formulate supervision policies for supervision subjects of different levels. The adjustment module dynamically adjusts the credit level according to the supervision result feedback: Update the supervision result to the credit scoring model in real time, recalculate the credit level, and synchronize it to the supervision department. The execution module performs joint supervision: generates joint tasks including inspection checklists and executors and distributes them to the joint supervision departments, and each supervision department automatically summarizes the inspection results to form a unified supervision report.

[0013] Furthermore, the regulatory behavior data collected by the model management module of the device for realizing joint supervision based on credit grading includes penalty records, compliance operation data, complaint information, and industry risk characteristics. Use the random forest or XGBoost in the machine learning algorithm to generate a credit scoring model, and set the grading rules of the credit level to divide the credit level by a set threshold, where level A ≥ 90 points, level B is 70 - 89 points, level C is 50 - 69 points, and level D < 50 points.

[0014] Furthermore, the supervision trigger module of the device for realizing joint supervision based on credit grading conducts credit risk analysis on the supervision subject. When the credit level of the supervision subject is level C / D, it triggers the joint supervision matching engine to screen associated supervision departments for generating joint supervision task suggestions.

[0015] Furthermore, the adjustment module of the device for realizing joint supervision based on credit grading automatically sends a risk warning to the associated supervision department for the supervision subject whose credit score drops sharply after recalculating the credit level.

[0016] Furthermore, the execution module of the device for realizing joint supervision based on credit grading adjusts the parameters of the credit scoring model based on the task completion rate and problem discovery rate in the supervision report to optimize the credit scoring model.

[0017] The advantages of the present invention are as follows: Realize accurate risk identification: construct a dynamic credit scoring model, fuse multi-source data in real time, and quantify the credit risk of market entities.

[0018] Intelligently trigger joint supervision: when the credit level reaches the preset risk threshold, such as level C / D, automatically match associated supervision departments to generate joint inspection task suggestions, reducing the manual coordination cost.

[0019] Differentiated resource allocation: implement different supervision strategies for high-credit and low-credit entities.

[0020] Optimize the closed-loop feedback: real-time feedback the joint inspection results to the credit scoring model, dynamically adjust the credit level, and form a "evaluation - execution - optimization" closed loop. Description of the Drawings

[0021] Figure 1 It is a schematic diagram of the application process of the method of the present invention. Detailed Embodiment

[0022] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, so that those skilled in the art can better understand the present invention and be able to implement it, but the exemplified embodiments are not intended to limit the present invention.

[0023] Embodiment 1: The present invention provides a method for realizing joint supervision based on credit grading, including: Step 1: Construct a credit assessment model: Collect regulatory behavior data to construct a credit assessment database, and use machine learning algorithms to generate a credit scoring model for dynamically outputting credit grades.

[0024] Among them, the collected regulatory behavior data includes penalty records, compliance operation data, complaint information, and industry risk characteristics. The random forest or XGBoost in machine learning algorithms can be used to generate a credit scoring model, and the grading rules for credit grades are set as setting thresholds to divide credit grades, where Grade A ≥ 90 points, Grade B 70 - 89 points, Grade C 50 - 69 points, and Grade D < 50 points.

[0025] Step 2: Create a joint supervision trigger mechanism: Initiate a supervision plan, formulate inspection tasks, automatically call the credit grade data of the supervised entity, generate joint supervision task suggestions, and push them to the matching supervision departments. Among them, conduct a credit risk analysis on the supervised entity. When the credit grade of the supervised entity is Grade C / D, trigger the joint supervision matching engine to screen associated supervision departments for generating joint supervision task suggestions.

[0026] Step 3: Formulate differentiated supervision strategies: According to the credit grades of the supervised entities and the joint supervision task suggestions, formulate supervision strategies for supervised entities at different levels.

[0027] For example, if the credit grade is Grade A, the supervision strategy is to simplify the process and conduct random annual inspections; if the credit grade is Grade B, the supervision strategy is to conduct regular inspections and reduce the frequency of cross-departmental inspections; if the credit grade is Grade C / D, the supervision strategy is to forcibly trigger multi-department joint inspections and increase the inspection intensity.

[0028] Step 4: Dynamically adjust the credit grade according to the feedback of the supervision results: Update the supervision results to the credit scoring model in real time, recalculate the credit grade, and synchronize it to the supervision departments. Among them, automatically send a risk warning to the associated supervision departments for the supervised entities whose credit scores drop significantly after recalculating the credit grade.

[0029] Step 5: Execute joint supervision: Generate a joint task including an inspection list and execution personnel and dispatch it to the joint supervision departments. Each supervision department automatically summarizes the inspection results to form a unified supervision report.

[0030] Adjust the parameters of the credit scoring model based on the task completion rate and problem discovery rate in the supervision report to optimize the credit scoring model.

[0031] The method of the present invention improves accuracy: through the credit scoring model, the accuracy of identifying high-risk entities is increased to 92%, compared with 65% of the existing model; the joint inspection coverage rate is increased from 30% to 85%, reducing the number of repeated inspections by about 60%.

[0032] Optimize resources: The inspection burden of high-credit enterprises (A / B grade) is reduced by 70%, and regulatory resources are concentrated on C / D grade enterprises, with the proportion increased from 20% to 65%.

[0033] Collaborative efficiency: The time to generate joint inspection tasks is shortened from 3 days to 10 minutes, and the department response rate can be increased from 40% to 90%; Through closed-loop management on mobile terminals, the time for summarizing inspection results can be shortened from 1 week to 2 hours.

[0034] Risk prevention and control: The incidence of major accidents for Class D enterprises decreased by 55%, and the credit repair cycle was shortened by an average of 30%.

[0035] When conducting specific supervision, Figure 1 , real-time aggregation of multi-source data such as data center, public opinion monitoring and enterprise declaration, after cleaning and feature extraction, dynamically calculates the enterprise credit score (0-100 points) based on the LightGBM algorithm, and divides it into A / B / C / D grades; when the regulatory department initiates an inspection, C / D-level enterprises automatically trigger cross-departmental joint supervision, and the rule engine matches related institutions according to industry risks. For example, catering companies link up with market supervision, fire protection, and health departments to generate a customized list of mandatory items and risk items, which are distributed to law enforcement personnel through mobile terminals for on-site scanning and entry. The results are transmitted back in real time and trigger dynamic adjustment of credit points, such as deductions for violations and extra points for rectification. The updated credit rating is simultaneously pushed to bidding, taxation and other systems, and at the same time, it feeds back model parameter optimization to form a "data collection-assessment-supervision-feedback" closed loop, realizing a self-evolving governance model with precise and strong supervision of high-risk enterprises and less interference with low-risk enterprises.

[0036] Embodiment 2: The present invention also provides a device for implementing joint supervision based on credit grading, comprising a model management module, a supervision trigger module, a policy management module, an adjustment module and an execution module. The model management module builds a credit assessment model: collects regulatory behavior data to build a credit assessment database, uses machine learning algorithms to generate a credit scoring model for dynamic output of credit ratings, The supervision trigger module creates a joint supervision trigger mechanism: initiates supervision plans, formulates inspection tasks, automatically calls the credit rating data of the supervision subject, generates joint supervision task suggestions, and pushes them to the matching supervision department. The strategy management module formulates differentiated supervision strategies: formulate supervision strategies for different levels of supervision entities according to the credit rating of the supervision entity and the joint supervision task suggestions. The adjustment module dynamically adjusts the credit rating based on the regulatory result feedback: updates the regulatory result to the credit scoring model in real time, recalculates the credit rating, and synchronizes it to the regulatory department. The execution module performs joint supervision: generates a joint task including an inspection checklist and execution personnel and dispatches it to the joint regulatory department, and each regulatory department automatically summarizes the inspection results to form a unified supervision report.

[0037] Regarding the information interaction and execution process among the above-mentioned modules in the device, since they are based on the same concept as the method embodiment of the present invention, the specific content can be referred to the description in the method embodiment of the present invention, and will not be elaborated here.

[0038] Similarly, the advantages of the device of the present invention are: Achieve accurate risk identification: construct a dynamic credit scoring model, fuse multi-source data in real time, and quantify the credit risk of market entities.

[0039] Intelligently trigger joint supervision: when the credit rating reaches the preset risk threshold, such as C / D level, automatically match and associate the regulatory department, generate a joint inspection task suggestion, and reduce the manual coordination cost.

[0040] Differentiated resource allocation: implement different supervision strategies for high-credit and low-credit entities.

[0041] Optimize the closed-loop feedback: real-time feedback the joint inspection results to the credit scoring model, dynamically adjust the credit rating, and form a closed loop of "evaluation - execution - optimization".

[0042] It should be noted that not all steps and modules in the above-mentioned processes and device structures are necessary, and some steps or modules can be ignored according to actual needs. The execution order of each step is not fixed and can be adjusted according to needs. The system structure described in the above-mentioned embodiments can be a physical structure or a logical structure, that is, some modules may be implemented by the same physical entity, or some modules may be implemented by multiple physical entities respectively, or some components in multiple independent devices can be jointly implemented.

[0043] The above-mentioned embodiments are only preferred embodiments given to fully illustrate the present invention, and the protection scope of the present invention is not limited thereto. Equivalent substitutions or transformations made by those skilled in the art on the basis of the present invention are within the protection scope of the present invention. The protection scope of the present invention is subject to the claims.

Claims

1. A method for achieving joint supervision based on credit grading, characterized by: include: Step 1: Build a credit assessment model: Collect regulatory behavior data to build a credit assessment database, and use machine learning algorithms to generate a credit scoring model for dynamic output of credit ratings. Step 2: Create a joint supervision trigger mechanism: initiate a supervision plan, formulate inspection tasks, automatically call the credit rating data of the supervision subject, generate joint supervision task suggestions, and push them to the matching supervision department. Step 3: Formulate differentiated supervision strategies: Formulate supervision strategies for different levels of supervisory entities based on their credit ratings and joint supervision task recommendations. Step 4: Dynamically adjust credit ratings based on regulatory feedback: Update regulatory results to the credit scoring model in real time, recalculate credit ratings, and synchronize with regulatory authorities. Step 5: Execute joint supervision: Generate a joint task including an inspection checklist and executors and distribute it to the joint supervision department. Each supervision department will automatically summarize the inspection results to form a unified supervision report.

2. The method for realizing joint supervision based on credit grading according to claim 1 is characterized in that The regulatory behavior data collected in step 1 include penalty records, compliance operation data, complaint information, and industry risk characteristics. The credit scoring model is generated by using random forest or XGBoost in machine learning algorithms, and the credit rating classification rules are set to divide the credit rating according to the set threshold, where grade A is ≥90 points, grade B is 70-89 points, grade C is 50-69 points, and grade D is <50 points.

3. The method for realizing joint supervision based on credit grading according to claim 1 is characterized in that In step 2, a credit risk analysis is conducted on the regulatory entity. When the credit rating of the regulatory entity is C / D, the joint regulatory matching engine is triggered to screen related regulatory departments to generate joint regulatory task recommendations.

4. The method for realizing joint supervision based on credit grading according to claim 1 is characterized in that In step 4, a risk warning is automatically sent to the relevant regulatory department for the regulatory entity whose credit score drops sharply after the credit rating is recalculated.

5. The method for realizing joint supervision based on credit grading according to claim 1 is characterized in that In step 5, the parameters of the credit scoring model are adjusted based on the task completion rate and problem discovery rate in the regulatory report to optimize the credit scoring model.

6. A device for realizing joint supervision based on credit grading, characterized in that It includes model management module, regulatory trigger module, strategy management module, adjustment module and execution module. The model management module builds a credit assessment model: collects regulatory behavior data to build a credit assessment database, uses machine learning algorithms to generate a credit scoring model for dynamic output of credit ratings, The supervision trigger module creates a joint supervision trigger mechanism: initiates supervision plans, formulates inspection tasks, automatically calls the credit rating data of the supervision subject, generates joint supervision task suggestions, and pushes them to the matching supervision department. The strategy management module formulates differentiated supervision strategies: formulate supervision strategies for different levels of supervision entities according to the credit rating of the supervision entity and the joint supervision task suggestions. The adjustment module dynamically adjusts the credit rating based on the feedback from the regulatory results: the regulatory results are updated to the credit scoring model in real time, the credit rating is recalculated, and synchronized to the regulatory department. The execution module performs joint supervision: a joint task including an inspection checklist and executors is generated and dispatched to the joint supervision department. Each supervision department automatically summarizes the inspection results to form a unified supervision report.

7. The device for realizing joint supervision based on credit grading according to claim 6, characterized in that The regulatory behavior data collected by the model management module includes penalty records, compliance operation data, complaint information, and industry risk characteristics. The credit scoring model is generated by using random forest or XGBoost in machine learning algorithms, and the credit rating classification rules are set to divide the credit rating according to the set threshold, where grade A is ≥90 points, grade B is 70-89 points, grade C is 50-69 points, and grade D is <50 points.

8. The device for realizing joint supervision based on credit grading according to claim 6, characterized in that The regulatory trigger module conducts credit risk analysis on the regulatory entity. When the credit rating of the regulatory entity is C / D, it triggers the joint regulatory matching engine to screen related regulatory departments to generate joint regulatory task recommendations.

9. The device for realizing joint supervision based on credit grading according to claim 6, characterized in that The adjustment module automatically sends risk warnings to related regulatory departments for regulatory entities whose credit scores drop sharply after recalculating their credit ratings.

10. The device for realizing joint supervision based on credit grading according to claim 6, characterized in that The execution module adjusts the parameters of the credit scoring model based on the task completion rate and problem discovery rate in the regulatory report to optimize the credit scoring model.

Images