A dynamic, multi-dimensional, customized enterprise credit assessment system and method

Abstract

This invention relates to the field of computer and data processing technology, and discloses a dynamic, multi-dimensional, customized enterprise credit assessment system and method, including an API interface, a computing server, and a set of industry model modules. The API interface receives the target enterprise's original dataset and a weight configuration dictionary. The computing server extracts matching industry-specific models from the set of industry model modules based on the industry attributes of the dataset. The computing server executes a feature mapping method to convert the original data items into a lower-level fourth-level indicator set. Subsequently, the computing server calls a hierarchical aggregation calculation method, using the externally input weight configuration dictionary to perform layer-by-layer weighted summation operations from bottom to top, sequentially generating third-level, second-level, and first-level indicator sets, and outputting the final assessment score. The system configures a dynamic simulation module to generate a feature value evolution sequence and outputs a dynamic credit prediction result. This invention decouples the computing engine from the assessment logic, improves the system's assessment flexibility, and provides prediction dimensions.

Description

Technical Field

[0001] This invention relates to the field of computer and data processing technology, specifically to a dynamic, multi-dimensional, customized enterprise credit assessment system and method. Background Technology

[0002] Corporate credit assessment is a fundamental step in financial risk control and business cooperation decisions. Existing corporate credit assessment systems typically employ an architecture that tightly binds the assessment business logic to the underlying computing engine. When faced with assessment needs from different industrial sectors, developers often need to refactor the core computing code to adapt to the characteristic variables and business rules of different industries. This tightly coupled system architecture leads to high code maintenance costs and makes it difficult to horizontally scale the system to new industrial sectors at low cost.

[0003] Traditional credit assessment models generally rely on internally pre-defined, hard-coded weight parameters. Once the assessment system is deployed, the calculation weights of different assessment dimensions are fixed. This fixed parameter configuration prevents the assessment system from dynamically adjusting the weighting of specific risk factors based on specific business scenarios. External systems cannot flexibly intervene and adjust the calculation weights without altering the underlying code structure, thus limiting the flexibility of the credit assessment process.

[0004] Current credit assessment mechanisms primarily focus on calculating a company's historical raw data and current static indicators. Existing assessment platforms lack simulation mechanisms to account for the evolution of basic data items over time. The scores output by the system can only passively reflect a company's static credit level at a specific point in time, and cannot provide dynamic predictions of the company's future credit trajectory by incorporating a time dimension. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a dynamic, multi-dimensional, and customized enterprise credit assessment system and method. It solves the problems of existing credit assessment systems, such as difficulties in industry expansion due to the tight coupling between assessment business logic and underlying computing engine, the inability to dynamically adjust the calculation ratio of specific business risk factors due to the reliance on hard-coded weight parameters in the assessment model, and the inability to output prediction results of future credit evolution trajectory due to the limitation of credit assessment mechanism to static historical data.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The first aspect of this invention provides a dynamic, multi-dimensional, customized enterprise credit assessment system.

[0008] The dynamic, multi-dimensional, customized enterprise credit assessment system comprises an API interface, a computing server, and a collection of industry model modules. The API interface is configured to receive the target enterprise's original dataset and weight configuration dictionary from external requests, and then transmit these to the computing server. The weight configuration dictionary contains the weight values ​​used in aggregation calculations for each level of indicators. The computing server establishes a data communication connection with the API interface. The collection of industry model modules connects to the computing server. The industry model module collection stores multiple industry-specific models corresponding to different industrial sectors. Each industry-specific model defines a four-level indicator hierarchy, from fine-grained features to a comprehensive score.

[0009] The system decouples the evaluation business logic from the underlying computing engine. The industry model module set includes a system base class and industry-specific subclasses that inherit from it. These subclasses constitute industry-specific models for their respective industrial sectors. Different industry-specific models define independent feature variable mapping rules and hierarchical aggregation operation logic. Through a modular class inheritance structure, the system can dynamically extract matching industry-specific models and perform calculations based on the industry attributes of the target company's original dataset.

[0010] The computing server is configured to execute feature mapping and hierarchical aggregation calculation methods within an industry-specific model. The feature mapping method transforms specific data items in the target company's original dataset into a standard fourth-level indicator set. The fourth-level indicator set consists of basic feature variables, constituting the basic feature variable set for this evaluation. The hierarchical aggregation calculation method includes third-level, second-level, and first-level indicator calculation methods. The third-level indicator calculation method retrieves the weight values ​​corresponding to the fourth-level indicator set from the weight configuration dictionary, performs a weighted sum operation on the fourth-level indicator set using these weight values, and generates the third-level indicator set. The third-level indicator set groups the fourth-level indicator set into evaluation sub-factors. The second-level indicator calculation method retrieves the weight values ​​corresponding to the third-level indicator set from the weight configuration dictionary, performs a weighted sum operation on the third-level indicator set using these weight values, and generates the second-level indicator set. The second-level indicator set represents the core evaluation dimension factors. The first-level indicator calculation method retrieves the weight values ​​corresponding to the second-level indicator set from the weight configuration dictionary, performs a weighted sum operation on the second-level indicator set using these weight values, and generates the first-level indicator set. The first-level indicator set includes the enterprise's own credit score and social credit score, serving as the final evaluation score corresponding to the target enterprise's original dataset. The hierarchical aggregation operation process follows the hierarchical relationship structure set by the static four-level tree structure, performing weighted summation calculations step by step from bottom to top.

[0011] In addition, the enterprise credit assessment system is equipped with a dynamic simulation module. This module includes a feature evolution simulation function. The computation server calls this function to calculate the numerical evolution of basic data items within a set simulation time period, generating a numerical sequence showing the changes in these basic data items over time. The enterprise credit assessment system uses this numerical sequence to perform model backtesting and scenario analysis. The dynamic simulation module inputs the numerical sequence into an industry-specific model to perform credit assessment calculations, and, combined with the set simulation time period, generates dynamic prediction results of the enterprise's credit trajectory. These dynamic prediction results provide a predictive dimension for credit assessment.

[0012] A second aspect of this invention provides a dynamic, multi-dimensional, and customized enterprise credit assessment method.

[0013] The specific steps involved in corporate credit assessment are as follows: In step S100, the API interface receives the target enterprise's original dataset and weight configuration dictionary sent by the external request, and then transmits the target enterprise's original dataset and weight configuration dictionary to the computing server. The weight configuration dictionary contains the weight values ​​used by each level of indicators in the aggregation calculation.

[0014] In step S200, the computing server extracts matching industry-specific models from the industry model module set. The industry-specific model is defined with a four-level indicator hierarchy structure, from fine-grained features to comprehensive scores.

[0015] In step S300, the computing server calls the feature mapping method within the industry-specific model to convert the specific data items in the target enterprise's original dataset into a standard fourth-level indicator set.

[0016] In step S400, the calculation server calls the third-level indicator calculation method within the industry-specific model, and uses the weight configuration dictionary to perform a weighted summation operation on the fourth-level indicator set to generate the third-level indicator set.

[0017] In step S500, the calculation server calls the second-level indicator calculation method within the industry-specific model, and uses the weight configuration dictionary to perform a weighted summation operation on the third-level indicator set to generate the second-level indicator set.

[0018] In step S600, the calculation server calls the first-level indicator calculation method within the industry-specific model, performs a weighted summation operation on the second-level indicator set using the weight configuration dictionary, and outputs the first-level indicator set as the final evaluation score.

[0019] This invention provides a dynamic, multi-dimensional, customized enterprise credit assessment system and method. It has the following beneficial effects: 1. This invention enhances the industry adaptability of the credit assessment system through a modular architecture. The computing server decouples the assessment business logic from the underlying computing engine, utilizing specific industry subclasses inherited from the system's base class within the industry model module set to configure independent feature variable mapping rules and hierarchical aggregation operation logic for different industrial sectors. When integrating credit assessment business from a new industrial sector, the system only needs to add the corresponding specific industry subclass, without modifying the core scheduling logic of the computing server, thus reducing system code maintenance costs.

[0020] 2. This invention enhances the flexibility of credit assessment based on a dynamic weight configuration dictionary and a four-level indicator hierarchy. The calculation server receives a weight configuration dictionary containing the weight values ​​of indicators at each level. The hierarchical aggregation calculation method uses the externally input weight configuration dictionary to perform weighted summation operations layer by layer from the fourth-level indicator set to the second-level indicator set, starting from the bottom up. The system can adjust the calculation proportion of different assessment dimensions by modifying the input values ​​in the weight configuration dictionary without changing the static four-level tree structure, thus meeting the emphasis requirements of different assessment scenarios for specific business risk factors.

[0021] 3. This invention utilizes a dynamic simulation module to provide a temporal predictive dimension for credit assessment. The dynamic simulation module is equipped with a feature evolution simulation function. The computation server calls this function to generate a numerical sequence of basic data items within the simulation time period. The system inputs the generated numerical sequence into a corresponding industry-specific model to perform weighted summation calculations. Combined with the simulation time period, it generates a dynamic prediction result of the enterprise's credit trajectory, realizing the system's transformation from static historical data evaluation to dynamic future credit trend prediction. Attached Figure Description

[0022] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a schematic diagram of the general structure of the four-level hierarchical index model of the present invention. Detailed Implementation

[0023] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Please see Figures 1-2 This invention provides a dynamic, multi-dimensional, customized enterprise credit assessment system and method.

[0025] To achieve dynamic, multi-dimensional credit assessment that supports deep industry customization, this invention provides an enterprise credit assessment system. Architecturally, the enterprise credit assessment system mainly consists of API interfaces, a computing server, and multiple industry model modules.

[0026] The API interface is configured to provide a network endpoint for connecting external systems to the compute server. The API interface receives data requests from external systems, which include the target enterprise dataset (list_data) and a dynamic weight configuration dictionary (formula_dict). The API interface then transmits the target enterprise dataset and the dynamic weight configuration dictionary to the compute server.

[0027] The computing server establishes a data communication connection with the API interface. The computing server obtains the target enterprise dataset and a dynamic weight configuration dictionary. Based on the industry attributes of the target enterprise dataset, the computing server retrieves the matching industry-specific model from the industry model module set. The computing server calls the built-in methods of the matching industry-specific model according to a preset sequence, controlling the multi-level credit assessment calculation process.

[0028] The industry model module collection stores multiple industry-specific models corresponding to different industrial sectors. Each industry-specific model encapsulates the business calculation rules and multi-level indicator definition data for that specific industrial sector. These industry-specific models inherit from the system's base class (LegalPerson). Different industry-specific models define independent feature variable mapping rules and hierarchical aggregation operation logic. The calculation server executes the invocation command, passing the dynamic weight configuration dictionary and the target enterprise dataset to the matched industry-specific model to derive the target credit score.

[0029] The corporate credit assessment calculation process specifically includes the following steps: In step S100, the API interface receives the target company's original dataset and weight configuration dictionary sent by the external request. The API interface then transmits the target company's original dataset and weight configuration dictionary to the computation server. The weight configuration dictionary contains the weight values ​​used by each level of indicators in the aggregation calculation.

[0030] In step S200, the computing server extracts matching industry-specific models from the industry model module set. The industry-specific model is defined with a four-level indicator hierarchy structure, from fine-grained features to comprehensive scores.

[0031] In step S300, the computing server calls the feature mapping method `getFeatures` within the industry-specific model. The `getFeatures` method transforms specific data items in the target company's original dataset into a standard set of fourth-level indicators. For example, `getFeatures` maps the input operating time data in the target company's original dataset to an internal feature variable of operating time within the fourth-level indicator set. This set of fourth-level indicators constitutes the basic set of feature variables for this evaluation.

[0032] In step S400, the calculation server calls the third-level indicator calculation method `thirdIndexes` within the industry-specific model. The third-level indicator calculation method `thirdIndexes` retrieves the weight values ​​corresponding to the fourth-level indicators from the weight configuration dictionary. The third-level indicator calculation method `thirdIndexes` performs a weighted summation operation on the fourth-level indicator set to generate the third-level indicator set. Taking the calculation of operating information in the third-level indicators as an example, the specific weighted summation calculation logic is as follows: the internal feature operating time variable is multiplied by the corresponding operating time weight value in the weight configuration dictionary, plus the internal feature operating address variable multiplied by the corresponding operating address weight value in the weight configuration dictionary, and so on, accumulating the results. Each indicator in the third-level indicator set constitutes an evaluation sub-factor.

[0033] In step S500, the calculation server calls the second-level indicator calculation method `secondIndexes` within the industry-specific model. The second-level indicator calculation method `secondIndexes` uses a weight configuration dictionary to perform a weighted summation operation on the third-level indicator set, generating the second-level indicator set. Taking the calculation of comprehensive quality among the second-level indicators as an example, the specific weighted summation calculation logic is as follows: the third-level indicator "operating information" is multiplied by the corresponding operating information weight value in the weight configuration dictionary, plus the third-level indicator "corporate assets" multiplied by the corresponding corporate asset weight value in the weight configuration dictionary, and so on, accumulating the results. Each indicator in the second-level indicator set reflects the core evaluation dimensions of the target enterprise.

[0034] In step S600, the calculation server calls the first-level indicator calculation method `firstIndexes` within the industry-specific model. The first-level indicator calculation method `firstIndexes` performs a weighted summation operation on the second-level indicator set using a weight configuration dictionary, outputting the first-level indicator set. This first-level indicator set is the final evaluation score for the target company. The final evaluation score includes the company's own credit score and its social credit score.

[0035] In the above steps, the fourth-level indicators constitute the underlying structure of the evaluation model. The fourth-level indicators consist of basic feature variables. These basic feature variables correspond to the target company's original or pre-processed data points. The feature variables are expressed as directly quantifiable basic data items. The computing server converts the received external input target company's original dataset into internal feature variables within the fourth-level indicator system by calling the feature mapping method `getFeatures`. The computing server maps the operating time data in the input original dataset to internal feature variables related to operating time. The computing server also maps the company's credit record data with the tax authorities in the input original dataset to internal feature variables related to tax credit records.

[0036] The components of the fourth-level indicators are configured differently based on different industry-specific models. The computing server executes the feature mapping method `getFeatures`, mapping and assigning each element of the external raw data to the corresponding industry-specific model's preset fourth-level indicator variables. For example, in the civil engineering and construction industry model, the fourth-level indicator set includes the project manager's years of experience and the project management personnel's qualifications. In the computer manufacturing industry model, the fourth-level indicator set includes the equipment manufacturing enterprise's qualification level. The chemical industry model does not include the equipment manufacturing enterprise's qualification level. By configuring independent fourth-level indicator sets according to different industry-specific models, the underlying feature variables conform to the specific industrial sector's business evaluation standards.

[0037] The third-level indicators are calculated by weighted summation of the fourth-level indicators, grouping the fourth-level indicators into corresponding sub-factors. The third-level indicator calculation method `thirdIndexes` obtains the mapped set of fourth-level indicators and receives the weight configuration dictionary `wts` from an external source. The weight configuration dictionary `wts` contains the calculated weight values ​​for each fourth-level indicator. The third-level indicator calculation method `thirdIndexes` uses the weight configuration dictionary `wts` to perform a weighted summation operation on the scores of the called multiple fourth-level indicators. Taking the calculation of the third-level indicator `management_information` as an example, the third-level indicator calculation method `thirdIndexes` calls the internal feature operating time variable `operation_time` and the internal feature operating address variable `business_address`. The calculation method for the third-level indicator `management_information` is as follows: self.fourth['operation_time'] wts['four_weight_operation_time']+self.fourth['business_address'] The calculation server multiplies the internal feature operating time variable by the corresponding operating time weight value (four_weight_operation_time) in the weight configuration dictionary (wts), and adds it to the internal feature operating address variable multiplied by the corresponding operating address weight value (four_weight_business_address) in the weight configuration dictionary (wts). The accumulated result is the score of the sub-factor of operating information. Taking the civil engineering construction industry model as an example, the third-level indicator, staffing, is calculated. The third-level indicator calculation method, thirdIndexes, calls the scores of the fourth-level indicator, the project manager's years of service characteristic variable and the project management personnel's qualification characteristic variable. The third-level indicator calculation method, thirdIndexes, uses the weight configuration dictionary (wts) to perform a weighted summation operation on the scores of the project manager's years of service characteristic variable and the project management personnel's qualification characteristic variable. The numerical result obtained from the weighted summation operation is the comprehensive score of the third-level indicator, staffing.

[0038] The second-level indicators are derived by weighted summation of the third-level indicators. The second-level indicators represent a broader dimension of corporate credit. The calculation server calls the secondIndexes method within the industry-specific model to obtain the set of third-level indicators after weighted aggregation. The secondIndexes method receives a weight configuration dictionary (wts) containing weight values ​​and uses it to perform a weighted summation operation on the third-level indicator set. Taking the calculation of the comprehensive quality indicator (comprehensive_quality) in the second-level indicators as an example, the secondIndexes method calls the management_information score and the enterprise_property score from the third-level indicators. The calculation method for the comprehensive quality indicator (comprehensive_quality) is as follows: self.third['management_information'] wts['third_weight_management_information']+self.third['enterprise_property'] The calculation server multiplies the third-level indicator's management information score by the corresponding management information weight value (third_weight_management_information) from the weight configuration dictionary (wts). It then multiplies the third-level indicator's enterprise property score by the corresponding enterprise property weight value (third_weight_enterprise_property) from the weight configuration dictionary (wts). The server then sums these products. The sum is used as the final score for the second-level indicator's comprehensive quality. The second-level indicators also include financial credit (financial_credit) and public credit (public_credit). Financial credit and public credit are calculated using the same weighted summation logic as comprehensive quality.

[0039] The first-level indicator is the final comprehensive score. It is calculated by weighted summation of the second-level indicators. The calculation server calls the first-level indicator calculation method `firstIndexes` within the industry-specific model to obtain the set of second-level indicators that have undergone weighted aggregation calculation. The first-level indicator calculation method `firstIndexes` performs a weighted summation operation on the second-level indicators using the corresponding weights from the weight configuration dictionary. Specifically, the calculation server multiplies the second-level indicator score by the corresponding weight value from the weight configuration dictionary and accumulates the product results. The accumulated result generates comprehensive scores for different main categories. The comprehensive score includes a comprehensive enterprise self-credit score and a social credit score. The comprehensive enterprise self-credit score is represented as `self.first['c']`, and the social credit score is represented as `self.first['s']`. The first-level indicator calculation method `firstIndexes` outputs the first-level credit score as the final evaluation result.

[0040] Regarding the API interface communication and parameter parsing mechanism, the API interface `calculationApplication.py` provides a secure HTTP endpoint ` / cal_py / compcal_py`. External systems interact with the computation engine through this endpoint. The API interface `calculationApplication.py` receives requests from external systems. These requests contain enterprise data `list_data` and dynamic weight configuration `formula_dict`. `calculationApplication.py` then transmits the enterprise data `list_data` and dynamic weight configuration `formula_dict` to the computation server. After receiving these documents, the computation server initiates the selection process for the corresponding industry-specific model and the multi-level credit assessment calculation process.

[0041] The Cal_Server, acting as the core processing engine, executes the system's scheduling mechanism. It receives data and weights from the API interface and selects the appropriate industry model. The Cal_Server coordinates the entire computation process by calling the methods of the selected industry model in a predetermined order.

[0042] The system's industry model module consists of Python classes that encapsulate specific industry business logic and indicator definitions. The system uses a base class, `LegalPerson`, which defines the general structure for calculations. Industry-specific subclasses in the industry model module inherit from `LegalPerson`. These subclasses implement unique sets of indicators, defining specific fourth-level features and aggregation logic for their respective industries. For example, the subclass `LegalPerson_HXYLZZ` corresponds to the chemical industry; `LegalPerson_JSJTXDZSBZZ` corresponds to the computer manufacturing industry; and `LegalPerson_TMGCJZGC` corresponds to the civil engineering and construction industry. This modular design allows the system to adapt to industry-specific risk factors by adding, deleting, or modifying indicators at any level.

[0043] The evaluation model for specific industries is configured with a structured indicator system. In the civil engineering industry example, the system is configured with a "Civil Engineering Indicator System.csv" file. This file uses a tabular format to define a standard four-level tree structure. This four-level tree structure forms the static foundation of the evaluation model. The four-level tree structure includes primary and secondary indicators. Primary indicators are set as the highest-level categories. Primary indicators include corporate credit and social credit. Secondary indicators are set as subdivisions of primary indicators. The corporate credit primary indicator includes comprehensive quality, project management capability, and engineering quality secondary indicators. The four-level tree structure includes tertiary and quaternary indicators. Tertiary indicators are set as subdivisions of secondary indicators. The project management capability secondary indicator includes staffing and management system tertiary indicators. Quaternary indicators are set as feature variables. Feature variables correspond to the most basic data items that can be directly quantified. The staffing tertiary indicator includes the project manager's years of experience and the project management personnel's qualifications quaternary indicators.

[0044] The dynamic mapping and computation execution mechanism of the algorithm model is implemented through a computation engine. Taking the civil engineering construction industry as an example, the computation engine includes the file TMGCJZGC.py. The file TMGCJZGC.py internally configures the class LegalPerson_TMGCJZGC. The class LegalPerson_TMGCJZGC is responsible for executing the feature variable mapping logic and the multi-level weighted aggregation calculation logic. The class LegalPerson_TMGCJZGC executes the feature mapping method getFeatures. The feature mapping method getFeatures receives the original dataset of the target enterprise from the outside and maps each data item in the original dataset to the fourth-level indicator variable defined in the civil engineering indicator system. The computation server assigns the project leader's years of experience data from the input original dataset to the feature variable self.feature['project_leader_experience'].

[0045] The class `LegalPerson_TMGCJZGC` executes a hierarchical aggregation calculation method. This method includes three levels of indicator calculation methods: `thirdIndexes` (third-level indicator calculation), `secondIndexes` (second-level indicator calculation), and `firstIndexes` (first-level indicator calculation). Following the hierarchical structure defined in the indicator system, the hierarchical aggregation calculation method performs a bottom-up weighted aggregation operation. The `thirdIndexes` method calls the scores of the fourth-level indicator (project manager's years of experience) and the fourth-level indicator (project management personnel's qualifications).

[0046] The third-level indicator calculation method, thirdIndexes, obtains the externally input weight dictionary wts and uses the weight dictionary wts to perform a weighted summation operation on the fourth-level indicator score of the project leader's years of service and the fourth-level indicator score of the project management personnel's qualifications, thereby generating the third-level indicator score for personnel allocation.

[0047] The second-level indicator calculation method, `secondIndexes`, calls the generated third-level indicator scores to perform a weighted summation operation, generating the second-level indicator scores. The first-level indicator calculation method, `firstIndexes`, calls the generated second-level indicator scores to perform a weighted summation operation, yielding the top-level evaluation result. The target company's original dataset is first aligned with the underlying feature variables of the indicator system using the feature mapping method `getFeatures`. The data items corresponding to the target company's original dataset follow the structural path defined by the indicator system, using the weight dictionary `wts` to perform layer-by-layer weighted aggregation calculations. The static, structured indicator system data is combined with an object-oriented algorithm model to output the enterprise credit assessment result.

[0048] In addition, the system is equipped with a dynamic simulation module. This module includes the feature evolution simulation function `randFeature`. `randFeature` calculates the numerical evolution of feature variables over a set simulation time period. The computation server executes the dynamic simulation module, calling `randFeature` to generate feature variable data for the simulation time period. The generated feature variable data contains a numerical sequence of basic data items changing over time. The system uses the feature variable data output by `randFeature` to perform model backtesting and scenario analysis. The dynamic simulation module inputs the feature variable data into an industry-specific model to perform credit assessment calculations. The system combines the set simulation time period and the feature variable data to generate dynamic prediction results for the enterprise's credit trajectory. These dynamic prediction results provide predictive dimensions for credit assessment.

[0049] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A dynamic multi-dimensional customized enterprise credit assessment system, characterized in that, include: The API interface is configured to receive the target enterprise's original dataset and weight configuration dictionary sent by external requests, and to transmit the target enterprise's original dataset and the weight configuration dictionary to the computing server. The weight configuration dictionary contains the weight values ​​used by each level of indicators in the aggregation operation. The computing server establishes a data communication connection with the API interface; An industry model module set is connected to the computing server. The industry model module set stores multiple industry-specific models corresponding to different industrial fields. The industry-specific models define a four-level indicator hierarchy structure from fine-grained features to comprehensive scores. The computing server is configured to extract matching industry-specific models from the industry model module set based on the industry attributes of the target enterprise's original dataset; the computing server is configured to execute a feature mapping method within the industry-specific model to convert data items in the target enterprise's original dataset into a fourth-level indicator set; the computing server is configured to execute a hierarchical aggregation calculation method within the industry-specific model, using the weight configuration dictionary to perform a layer-by-layer weighted summation operation on the fourth-level indicator set, sequentially generating a third-level indicator set, a second-level indicator set, and a first-level indicator set, where the first-level indicator set is the final evaluation score corresponding to the target enterprise's original dataset.

2. The dynamic, multi-dimensional, customized enterprise credit assessment system according to claim 1, characterized in that, The industry model module set includes a system base class and specific industry subclasses that inherit from the system base class. The specific industry subclasses constitute industry-specific models corresponding to specific industrial fields. Different industry-specific models define independent feature variable mapping rules and hierarchical aggregation operation logic.

3. The dynamic, multi-dimensional, customized enterprise credit assessment system according to claim 1, characterized in that, The fourth-level indicator set consists of basic feature variables, which correspond to the original data points in the target enterprise's original dataset. The computing server executes the feature mapping method to convert the specific data items in the target enterprise's original dataset into internal feature variables. The fourth-level indicator set constitutes the basic feature variable set for this evaluation.

4. The dynamic, multi-dimensional, customized enterprise credit assessment system according to claim 1, characterized in that, The hierarchical aggregation calculation method includes a third-level indicator calculation method. The third-level indicator calculation method obtains the mapped fourth-level indicator set and obtains the weight values ​​corresponding to the fourth-level indicator set from the weight configuration dictionary. The third-level indicator calculation method uses the weight values ​​to perform a weighted summation operation on the fourth-level indicator set to generate the third-level indicator set. The third-level indicator set groups the fourth-level indicator set into evaluation sub-factors.

5. The dynamic, multi-dimensional, customized enterprise credit assessment system according to claim 4, characterized in that, The hierarchical aggregation calculation method includes a second-level indicator calculation method. The second-level indicator calculation method obtains the third-level indicator set after weighted aggregation calculation. The second-level indicator calculation method uses the weight configuration dictionary to perform a weighted summation operation on the third-level indicator set to generate the second-level indicator set. Each indicator in the second-level indicator set reflects the core evaluation dimension factors of the target enterprise.

6. The dynamic, multi-dimensional, customized enterprise credit assessment system according to claim 5, characterized in that, The hierarchical aggregation calculation method includes a first-level indicator calculation method. The first-level indicator calculation method obtains a second-level indicator set after weighted aggregation calculation. The first-level indicator calculation method uses the weight configuration dictionary to perform a weighted summation operation on the second-level indicator set and outputs the first-level indicator set. The first-level indicator set includes the comprehensive enterprise's own credit score and social credit score.

7. The dynamic, multi-dimensional, customized enterprise credit assessment system according to claim 1, characterized in that, The enterprise credit assessment system is equipped with a dynamic simulation module, which includes a feature evolution simulation function. The calculation server calls the feature evolution simulation function to calculate the numerical evolution of basic data items within a set simulation time period to generate a numerical sequence.

8. The dynamic, multi-dimensional, customized enterprise credit assessment system according to claim 7, characterized in that, The enterprise credit assessment system uses the numerical sequence to perform model backtesting and scenario analysis. The dynamic simulation module inputs the numerical sequence into the industry-specific model to perform credit assessment calculations and generates dynamic prediction results of the enterprise credit trajectory based on the set simulation time period.

9. A dynamic, multi-dimensional, customized enterprise credit assessment system according to claim 1, characterized in that, The evaluation model for a specific industry is configured with a structured indicator system. The structured indicator system uses a standard four-level tree structure defined in tabular form. The four-level tree structure constitutes the static foundation of the evaluation model. The feature mapping method and the hierarchical aggregation calculation method perform bottom-up weighted aggregation operations in accordance with the hierarchical association structure set by the four-level tree structure.

10. A dynamic, multi-dimensional, customized enterprise credit assessment method, characterized in that, Includes the following steps: Step S100: The API interface receives the target enterprise's original dataset and weight configuration dictionary sent by the external request, and transmits the target enterprise's original dataset and the weight configuration dictionary to the computing server. The weight configuration dictionary contains the weight values ​​used by each level of indicators in the aggregation operation. Step S200: The computing server extracts matching industry-specific models from the industry model module set. The industry-specific models define a four-level indicator hierarchy structure from fine-grained features to comprehensive scores. Step S300: The computing server calls the feature mapping method within the industry-specific model to convert the specific data items in the original dataset of the target enterprise into a standard fourth-level indicator set. Step S400: The computing server calls the third-level indicator calculation method in the industry-specific model, and uses the weight configuration dictionary to perform a weighted summation operation on the fourth-level indicator set to generate the third-level indicator set. Step S500: The computing server calls the second-level indicator calculation method within the industry-specific model, and uses the weight configuration dictionary to perform a weighted summation operation on the third-level indicator set to generate the second-level indicator set; In step S600, the computing server calls the first-level indicator calculation method within the industry-specific model, uses the weight configuration dictionary to perform a weighted summation operation on the second-level indicator set, and outputs the first-level indicator set as the final evaluation score.

Images