Enterprise credit evaluation system and method, electronic equipment and storage medium

Abstract

The invention provides an enterprise credit assessment system and method, electronic equipment and a storage medium. The system is connected with a preset overall database. The enterprise credit evaluation system comprises an input module, an acquisition module, a processing module and an analysis module. The input module is used for inputting an enterprise identifier of a target enterprise of the to-be-assessed medical creative type; the acquisition module is used for acquiring various to-be-evaluated data of the target enterprise from the overall database according to the enterprise identifier of the target enterprise; wherein the to-be-evaluated data comprises data under different dimensions, and the to-be-evaluated data covers the whole life cycle of the target enterprise; the processing module is used for processing the to-be-evaluated data to obtain sub-processing results of the to-be-evaluated data; and the analysis module is used for analyzing the sub-processing result to obtain an enterprise credit analysis result. According to the invention, full-life-cycle data analysis is carried out on the medical and invasive enterprises, and the problem that the credit rating and early warning of the enterprises cannot cover the industry pain points in the beginning period and the growth period is solved.

Description

Technical Field

[0001] The present application relates to the technical field of enterprise credit assessment, and in particular to a system, method, electronic device and storage medium for enterprise credit assessment. Background Art

[0002] Currently, credit reviews and risk management for technology companies by banks and non-bank institutions are primarily based on assessing credit ratings and monitoring credit fluctuations. The construction and monitoring of credit ratings are centered around historical default samples. These systems extract and quantify variables such as a company's financial information, the industry's competitive landscape, and the overall macroeconomic environment. Models are then used to estimate a company's debt repayment capacity and predict its potential default losses. In practice, companies' credit risk is typically assigned a rating based on their probability of default and default losses, such as AAA, AA, A, BBB, BB, or B, to differentiate risk levels and grant credit to technology companies accordingly.

[0003] Currently, there are the following problems with corporate credit assessment: 1) The most core financial data for evaluating corporate credit relies on the company's regular public disclosure. The regularly disclosed financial information has a lag effect and cannot reflect changes in the company's credit status in a timely manner; 2) The current credit analysis and early warning systems are mostly applicable to mature companies and cannot effectively analyze technology-based companies at different development stages, such as the start-up and growth stages, and cannot provide financial services covering the entire life cycle of technology-based companies; 3) The difficulty in credit assessment of small and micro enterprises lies in the fact that the data dimensions available to them are limited, and their privacy data is stored in non-public departments, which attach the greatest importance to privacy data protection. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a system, method, electronic device and storage medium for enterprise credit assessment to overcome the problems in the prior art.

[0005] In a first aspect, an embodiment of the present application provides a system for enterprise credit assessment, wherein the enterprise credit assessment system is connected to a preset overall database; the enterprise credit assessment system includes: an input module, an acquisition module, a processing module, and an analysis module;

[0006] The input module is used to input the corporate identity of the target scientific and technological innovation enterprise to be evaluated;

[0007] The acquisition module is configured to acquire, from the overall database, a plurality of data to be evaluated of the target enterprise based on the enterprise identification of the target enterprise; wherein the data to be evaluated includes data in different dimensions and covers the entire life cycle of the target enterprise;

[0008] The processing module is used to process the data to be evaluated to obtain sub-processing results of each data to be evaluated;

[0009] The analysis module is used to analyze the sub-processing results to obtain the enterprise credit analysis results of the target enterprise.

[0010] In some technical solutions of this application, the data to be evaluated includes: real public data and real private data; the overall database includes a public database and a private database;

[0011] When the acquisition module acquires the multiple data to be evaluated of the target enterprise from the overall database, the acquisition module includes:

[0012] Constructing a first privacy-preserving computing node in a public database corresponding to the real public data and constructing a second privacy-preserving computing node in a private database corresponding to the real private data;

[0013] Real public data is obtained through the first privacy computing node and real private data is obtained through the second privacy computing stage.

[0014] In some technical solutions of the present application, the acquisition module constructs the first privacy-preserving computing node and the second privacy-preserving computing node in the following manner:

[0015] Constructing the first privacy-preserving computing node in the first target database, and obtaining real public data in a first other database other than the first target database through the first privacy-preserving computing node;

[0016] The second privacy computing node is constructed in the second target database, and the real public data in the second other database other than the second target database in the privacy database is obtained through the second privacy computing node.

[0017] In some technical solutions of the present application, when constructing the first privacy-preserving computing node and the second privacy-preserving computing node, the acquisition module is further configured to:

[0018] If real public data is stored in the second privacy computing node, the second privacy computing node transmits the data to the first privacy computing node through oblivious transmission.

[0019] In some technical solutions of the present application, the above-mentioned method for analyzing the sub-processing results to obtain the enterprise credit analysis result of the target enterprise includes:

[0020] The preset federated learning model is trained using the preset test public data and test private data to obtain the trained target prediction model.

[0021] The sub-processing result is input into the target prediction model to obtain the enterprise credit analysis result of the target enterprise output by the target prediction model.

[0022] In some technical solutions of this application, the above-mentioned data to be evaluated include: public financial data, transaction data, invoice data and investment and financing data.

[0023] In some technical solutions of the present application, the processing module is used to process the data to be evaluated to obtain sub-processing results of each data to be evaluated, including:

[0024] Extracting the target enterprise's financial indicators from the public financial data, and establishing a benchmark status of the enterprise's credit based on the enterprise financial indicators;

[0025] Identify and classify the cash flow data and construct the true status of the enterprise's credit from the cash flow dimension;

[0026] Identify and classify the invoice data and construct the target enterprise's supply chain relationship network from a fiscal and tax perspective;

[0027] Based on the investment and financing data, the industry classification label, equity structure, investment and financing relationship network and valuation changes of the target enterprise are constructed from the investment and financing dimension.

[0028] In a second aspect, an embodiment of the present application provides a method for enterprise credit assessment, which is applied to the above-mentioned enterprise credit assessment system, and the method includes:

[0029] Obtain the corporate logo of the target scientific and technological innovation enterprise to be evaluated;

[0030] According to the enterprise identification of the target enterprise, a plurality of data to be evaluated of the target enterprise is obtained from the overall database; wherein the data to be evaluated includes data in different dimensions and covers the entire life cycle of the target enterprise;

[0031] Processing the data to be evaluated to obtain sub-processing results of each of the data to be evaluated;

[0032] The sub-processing results are analyzed to obtain the enterprise credit analysis result of the target enterprise. The enterprise credit analysis result module analyzes the sub-processing results to obtain the enterprise credit analysis result of the target enterprise.

[0033] In a third aspect, an embodiment of the present application provides an electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned enterprise credit assessment method when executing the computer program.

[0034] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned enterprise credit assessment method are executed.

[0035] The technical solutions provided by the embodiments of the present application may have the following beneficial effects:

[0036] The enterprise credit assessment system of the present application is connected to a preset overall database; the enterprise credit assessment system includes: an input module, an acquisition module, a processing module and an analysis module; the input module is used to input the corporate identity of a target scientific and technological innovation enterprise to be assessed; the acquisition module is used to obtain a variety of data to be assessed of the target enterprise from the overall database based on the corporate identity of the target enterprise; wherein the data to be assessed includes data under different dimensions, and the data to be assessed covers the entire life cycle of the target enterprise; the processing module is used to process the data to be assessed to obtain sub-processing results of each of the data to be assessed; the analysis module is used to analyze the sub-processing results to obtain the enterprise credit analysis results of the target enterprise.

[0037] This application conducts data analysis on the entire life cycle of science and technology-based enterprises to solve the industry pain point that their credit ratings and early warnings cannot cover the start-up and growth stages, and provides users with a credit analysis system for science and technology-based enterprises that is universal, flexible, and easy to operate.

[0038] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0040] Figure 1 A schematic diagram of a system for enterprise credit assessment provided by an embodiment of the present application is shown;

[0041] Figure 2 A flow chart of a method for evaluating corporate credit provided by an embodiment of the present application is shown;

[0042] Figure 3 A schematic diagram showing another enterprise credit assessment method provided by an embodiment of the present application is shown;

[0043] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of illustration and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowcharts can be implemented out of sequence, and steps without logical context can be reversed or implemented simultaneously. In addition, those skilled in the art, under the guidance of the contents of this application, can add one or more other operations to the flowchart, or remove one or more operations from the flowchart.

[0045] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present application.

[0046] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0047] Currently, credit reviews and risk management for technology companies by banks and non-bank institutions are primarily based on assessing credit ratings and monitoring credit fluctuations. The construction and monitoring of credit ratings are centered around historical default samples. These systems extract and quantify variables such as a company's financial information, the industry's competitive landscape, and the overall macroeconomic environment. Models are then used to estimate a company's debt repayment capacity and predict its potential default losses. In practice, companies' credit risk is typically assigned a rating based on their probability of default and default losses, such as AAA, AA, A, BBB, BB, or B, to differentiate risk levels and grant credit to technology companies accordingly.

[0048] Currently, there are the following problems with corporate credit assessment: 1) The most core financial data for evaluating corporate credit relies on the company's regular public disclosure. The regularly disclosed financial information has a lag effect and cannot reflect changes in the company's credit status in a timely manner; 2) The current credit analysis and early warning systems are mostly applicable to mature companies and cannot effectively analyze technology-based companies at different development stages, such as the start-up and growth stages, and cannot provide financial services covering the entire life cycle of technology-based companies; 3) The difficulty in credit assessment of small and micro enterprises lies in the fact that the data dimensions available to them are limited, and their private data is stored in non-public departments, which attach the greatest importance to privacy data protection.

[0049] Based on this, the embodiments of the present application provide a system, method, electronic device, and storage medium for enterprise credit assessment, which are described below through embodiments.

[0050] Figure 1 A schematic diagram of a system for enterprise credit assessment provided by an embodiment of the present application is shown, wherein the enterprise credit assessment system is connected to a preset overall database; the enterprise credit assessment system includes: an input module, an acquisition module, a processing module, and an analysis module;

[0051] An input module is used to input the corporate identity of the target scientific and technological innovation enterprise to be evaluated;

[0052] an acquisition module, configured to acquire, from the overall database, a plurality of data to be evaluated of the target enterprise according to the enterprise identification of the target enterprise; wherein the data to be evaluated includes data in different dimensions and covers the entire life cycle of the target enterprise;

[0053] A processing module, configured to process the data to be evaluated to obtain sub-processing results of each of the data to be evaluated;

[0054] The analysis module is used to analyze the sub-processing results to obtain the enterprise credit analysis results of the target enterprise.

[0055] This application conducts data analysis on the entire life cycle of science and technology-based enterprises to solve the industry pain point that their credit ratings and early warnings cannot cover the start-up and growth stages, and provides users with a credit analysis system for science and technology-based enterprises that is universal, flexible, and easy to operate.

[0056] The following describes some embodiments of the present application in detail. In the absence of conflict, the following embodiments and features in the embodiments may be combined with each other.

[0057] Technologically innovative enterprises, or technologically innovative enterprises, are defined as those that possess core technologies with independent intellectual property rights, renowned brands, strong innovation management and culture, and possess advanced overall technological capabilities compared to their peers. These enterprises possess competitive advantages and sustainable development capabilities. These enterprises typically focus on the research, development, production, and technical services of high-tech products and their products. They are the driving force behind the development of high-tech industries and a crucial component of the independent innovation system.

[0058] Given the different lifecycle stages of sci-tech enterprises, credit assessments of these enterprises require specific data. This specificity primarily requires that the data cover the entire lifecycle of the enterprise. Furthermore, the authenticity and integrity of the data must also be ensured.

[0059] The embodiment of the present application provides an enterprise credit assessment system, which is connected to a preset overall database, where the overall database is collected and organized by a specific organization. The authenticity of the data is guaranteed by direct acquisition. Specifically, the enterprise credit assessment system in the embodiment of the present application includes an acquisition module, which is used to obtain the data to be evaluated from the above-mentioned overall database in real time. The data to be evaluated here is for the target enterprise, and the target enterprise is a scientific and technological innovation enterprise that needs to be evaluated. The target enterprise is determined by the user. The specific determination method includes inputting the corporate identity of the target enterprise. The enterprise credit assessment system in the embodiment of the present application includes an input module, through which the user inputs the corporate identity. The corporate identity here can be the name of the enterprise, tax number, and other distinguishing characters, text, pictures, voice, etc. After the user enters the corporate identity, the enterprise credit assessment system in the embodiment of the present application obtains a variety of data to be evaluated of the target enterprise from the overall database based on the corporate identity.

[0060] To ensure the accuracy of the evaluation results, the various data to be evaluated are from different dimensions and cover the entire lifecycle of the target enterprise. Generally speaking, data directly obtained from the overall database is the initial data, which often has formatting or defects. After obtaining the initial data, it needs to be format-checked, integrity-checked, desensitized, and processed before being output as feature data (data to be evaluated) to ensure data security and usability.

[0061] During the actual acquisition process, the various data types to be evaluated include: real public data and real private data. During storage, real public data and real private data are stored in separate databases. Real public data is stored in the public database, while real private data is stored in the private database. The public and private databases are independent of each other and cannot communicate with each other. For example, real public data includes publicly available investment and financing data, public financial data, and connected invoice data, while real private data includes transaction data and undisclosed financial data.

[0062] Specifically, the publicly available investment and financing data fields include: financing time, financing round, financing amount, investor name, legal person shareholding ratio, and valuation amount. The publicly available financial data fields include: cash and cash equivalents, accounts receivable, inventory, fixed assets, intangible assets, short-term loans, accounts payable, long-term loans, operating income, operating costs, net profit, gross profit margin, net profit margin, cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, current ratio, quick ratio, debt-to-asset ratio, return on equity, equity, capital reserve, surplus reserve, and retained earnings. The publicly available invoice data fields include: invoice issuance time, counterparty, invoice amount, invoice tax rate, and invoice tax amount. Other publicly available data fields include: social security, provident fund, value-added tax, income tax, and basic industrial and commercial information. The transaction data fields include: loan amount, loan period, loan interest rate, number of active accounts, average daily deposits, operating income, salary expenditure, total capital outflow and total capital inflow, as well as financial data.

[0063] In order to ensure the security of data, when acquiring data, the embodiment of the present application constructs a privacy computing node through an acquisition module, specifically: constructing a first privacy computing node in the public database corresponding to the real public data and constructing a second privacy computing node in the private database corresponding to the real private data; acquiring the real public data through the first privacy computing node and acquiring the real private data through the second privacy computing stage.

[0064] In the case where there are multiple databases for real databases and privacy databases (for example, one database for investment and financing data, one database for public financial data, and one database for invoice data), when constructing a privacy computing node, you can select any data to construct a privacy computing node, and then use the privacy computing node to obtain data of the corresponding type in other databases. For example, the first privacy computing node is constructed in the first target database, and the first privacy computing node is used to obtain the real public data in the first other database in the public database other than the first target database; wherein the first target data is any database in the public database; the second privacy computing node is constructed in the second target database, and the second privacy computing node is used to obtain the real public data in the second other database in the privacy database other than the second target database, wherein the second target data is any database in the privacy database.

[0065] Considering the presence of financial indicators in the transactional data (private data includes public data, meaning the second privacy computing node stores real public data), this embodiment of the application requires transferring the real public data contained in the second privacy computing node to the first privacy computing node. To ensure data security during this transfer, this embodiment of the application uses an oblivious transfer method between the first and second privacy computing nodes.

[0066] After obtaining the aforementioned data to be evaluated, the processing module in the enterprise credit evaluation system processes the data to obtain the sub-processing results corresponding to each data to be evaluated. The analysis module in the enterprise credit evaluation system then analyzes each sub-processing result to obtain the enterprise credit analysis result of the target enterprise.

[0067] When analyzing the data to be evaluated, the analysis module includes: extracting the target enterprise's financial indicators from the publicly available financial data and constructing a baseline credit status for the enterprise based on these indicators; identifying and classifying the transaction flow data and constructing the true credit status of the enterprise from the perspective of cash flow; identifying and classifying the invoice data and constructing the target enterprise's supply chain relationship network from the perspective of finance and taxation; and constructing the target enterprise's industry classification label, equity structure, investment and financing relationship network, and valuation changes from the perspective of investment and financing based on the investment and financing data.

[0068] Specifically, the processing of public financial data involves extracting macroeconomic factors and corporate credit default factors based on public financial data. Macroeconomic factors used include the unemployment rate, inflation rate, the average return of the CSI 1000 Index over the past 12 months, and the one-year LPR. Corporate financial factors used include: DTD_level (the average of the distance to default over the past 12 months), DTD_trend (the difference between the distance to default for the current month and DTD_level), CASH / TA_level (the ratio of cash and short-term investments to total asset market value over the past 12 months), CASH / TA_trend (the difference between the CASH / TA value for the current month and CASH / TA_level), NI / TA_level (the average ratio of net income to total assets over the past 12 months), NI / TA_train (the difference between the NI / TA value for the current month and NI / TA_level), SIZE_level (the average size over the past 12 months, where SIZE is the logarithm of the ratio of the current company's market value to the average market value of the CSI 1000 companies), and SIZE_trend (the difference between the SIZE value for the current month and SIZE_level).

[0069] The processing of flow data is as follows:

[0070] 1. Integrate cash flow data into a unified standard format;

[0071] 2. Data verification: Check whether the transaction data is complete and reliable, and generate two indicators: reliability and completeness. (1) Reliability: Verify the degree of data processing (account balances should be continuous, interest should be received each quarter that matches the daily average balance, internal transfers between accounts should be matched and offset, etc.); (2) Completeness: Check whether the data of each account is complete and there are no missing items. In addition, check whether the company has submitted all the main accounts.

[0072] 3. Data structuring, including labeling and classification of flow data; (1) Labeling of flow data: 1) By using the legal representative and shareholder information, we can determine whether the counterparty is our affiliated company. 2) By using the business scope, we can determine whether the company is a customer / supplier, which can help us establish industry indicators (such as the proportion of wholesaler-type expenditures to total revenue or total expenditures). 3) By using the registration date, we can classify companies into newly established, 1-2 years, 2-5 years, 5 years and above, etc., which can help identify suspicious companies that have close trade relations since their establishment. 4) By using the registered capital, we can classify companies into large, medium and small, etc.

[0073] (2) Label classification: first, distinguish by operating, fundraising and financing, and then further refine the classification into internal transfers, sales revenue, loan inflows, interbank borrowing, supplier procurement, salary expenditures, tax payments, water, electricity, rent, travel, etc.

[0074] Regarding invoice data processing, an invoice is the smallest element covering transactions between businesses. Based on the face of an invoice, over 40 basic elements can be directly extracted. After preliminary processing, over 300 basic indicators can be generated. Combined with industry chain, supply chain, and sectoral analysis, 3,000 indicators can be derived. For example, four elements can be extracted from an invoice: 1. Transaction time; 2. Transaction parties; 3. Transaction content; and 4. Transaction amount. 1. Transaction time allows for determining peak seasons, operating cycles, and invoicing intervals; 2. Transaction parties provides information on the buyer, seller, and transaction region; 3. Transaction content provides information on the product name and quantity; and 4. Transaction amount provides information on the invoice amount, tax rate, tax amount, and discounts. Furthermore, the transaction status allows for determining whether an invoice is normal, voided, red-checked, or out of control. The transaction type allows for determining whether the invoice is a special VAT invoice or a general VAT invoice.

[0075] In the supply chain scenario, invoices can be used to accurately identify upstream and downstream companies. The company's invoice data combined with the invoice data of upstream and downstream companies can further accurately identify the supply data of upstream suppliers, the order data of downstream distributors and other information.

[0076] Regarding investment and financing data, traditional corporate credit risk models are applicable to companies with mature business models and stable operating cash flow, as well as indirect financing models such as bank loans. Before reaching maturity, technology-based companies must first undergo the startup and growth phases, during which operating cash flow is unstable, and most companies even experience net losses. Therefore, before obtaining indirect financing such as bank loans, technology-based companies urgently need direct financing methods such as equity investment to secure funding for continued survival and growth. Therefore, investment and financing data is essential for a credit assessment system that covers the entire life cycle of technology companies.

[0077] Investment and financing data includes industry labels that are more accurate than the national industry classification, company overseas investment information, company financing information, shareholder information, etc.

[0078] Investment and financing data can be used to precisely identify a company's industry sector, financing rounds, shareholder background, and other key information. For example, the investment and financing rounds for a company are: Seed, Angel, A, B, C, and so on, leading to IPO. The earlier the round, the higher the overall credit risk. For example, using industry sector labels, investment and financing data can generate indicators related to investment trends, ranking industry hotspots based on investor investment in those sectors. The more financing a company receives in its industry, the greater the probability and amount of potential direct financing, and the lower the company's credit risk.

[0079] Taking shareholder background as an example, investors can be categorized into: PE, VC, secondary market funds, incubation funds, etc. Based on shareholder background information, we can generate indicators such as shareholder background classification indicators, total shareholder investment amount in the past 12 months, total shareholder investment amount in industry sectors, and total shareholder investment amount in competing products.

[0080] After obtaining the sub-processing results for each piece of data to be evaluated, a fusion analysis of these sub-processing results is required. Specifically, the results of processing public financial data, transaction data, invoice data, and investment and financing data are input into a preset target prediction model to obtain the corporate credit analysis results output by the target prediction model. The target prediction model may include a statistical learning model, a machine learning model, a federated learning model, or a combination of the three.

[0081] Statistical Model: Based on historical samples, we use logistic regression and traditional financial indicators to construct a benchmark corporate default probability model. The statistical model serves as a baseline model and serves as a comparison benchmark for subsequent machine learning models. This means that subsequent machine learning models must outperform statistical models in predicting corporate credit risk.

[0082] We use NLP models to semantically understand text within transaction data, invoice data, and investment and financing data. Machine learning is then used to label and categorize the data. This allows us to develop analytical metrics tailored to technology companies across different industries and lifecycles. These metrics cover scientific and technological innovation qualifications, awards and subsidies, industry competitiveness, awards for scientific and technological innovation, counterparties, R&D expenditures, and salary expenditures.

[0083] The model was established using a blending ensemble learning method. The ensemble model is constructed in two layers. The first layer utilizes high-fit models, including a logistic regression model (statistical model), XGBoost (machine learning model), and an ANN (machine learning model). Each model in the first layer uses different algorithms to extract features that best reflect the credit risk of technology companies from the aforementioned composite data. The models are trained from different perspectives, yielding differentiated results that are then output to the second layer. The second layer utilizes a logistic regression model with L2 regularization to ultimately determine the company's credit default probability. Based on the ranking of the company's default probability, the company's credit rating is assigned; based on changes in the company's default probability, a default warning signal is generated.

[0084] The training process of the federated learning model is as follows: for testing public data and testing private data;

[0085] (1) First, we use oblivious transmission to perform privacy intersection. The first and second privacy computing nodes use the social unified credit code of the enterprise as the association key to find the small and micro enterprises with the intersection. Before the privacy intersection, we use the second privacy computing node to screen the existing small and micro enterprises: small and micro enterprises that have been in the five-level classification of concern, secondary, suspicious, or loss in the past two years are marked as bad enterprises, and small and micro enterprises that have not been in the five-level classification of concern, secondary, suspicious, or loss in the past two years are randomly selected from 10 times the data and marked as good enterprises.

[0086] (2) The second step is data preprocessing. By configuring, we delete features with a missing value ratio of more than 80%, an outlier ratio of more than 10%, a feature IV value less than 0.02, a correlation coefficient greater than 0.8, a variance less than 0.01, or a unique value ratio greater than 0.95;

[0087] (3) Data division: According to time, enterprises in January 2025 and December 2024 are selected as the validation set, January 2023 to June 2024 as the training set, and July 2024 to November 2024 as the test set;

[0088] (4) The data set was screened again, and the PSI (Population Stability Index) of the training set, test set, and validation set data was calculated. Any variable with a PSI greater than 0.1 between the training set and the test set, or between the training set and the validation set, was deleted.

[0089] (5) Configure SecureBoost. The main parameters are:

[0090] {objective = 'binary:logistic', #loss function

[0091] eval_metric = 'logloss', #Evaluation metric

[0092] scale_pos_weight = np.sum(y_train == 0) / np.sum(y_train == 1), #Amplify the loss contribution of the minority class

[0093] tree_method = 'hist', # continuous feature discretization using histogram algorithm

[0094] learning_rate=0.05, #learning rate

[0095] max_depth=3, #depth of the tree

[0096] min_child_weight=10, #minimum weight of child nodes

[0097] subsample=0.7, #row sampling

[0098] colsample_bytree=0.7, # column sampling

[0099] Gamma=2, split threshold

[0100] n_estimators = 1000 #number of iterations

[0101] The encryption scheme used is: Paillier (homomorphic encryption);

[0102] Key length: 1024 bits

[0103] }

[0104] (6) Continue configuring the test module and verification module, and adjust the parameters in (5) according to the results until the test set auc is greater than 0.75 and ks is greater than 0.25;

[0105] (7) To configure a vertical federated DNN, we first need to perform Z-Score normalization on the three datasets and then configure the model parameters:

[0106] Input layer: dimension: [batch_size, 128];

[0107] Fully connected layer 1: number of units: 256, activation function: ReLU;

[0108] Batch Normalization layer: momentum: 0.9, epsilon: 1e-5;

[0109] Fully connected layer 2: number of units: 128, activation function: Sigmoid;

[0110] Output layer: number of units: 1, activation function: Sigmoid;

[0111] Optimizer: Privacy-preserving Adam (learning rate = 0.001, β1 = 0.9, β2 = 0.999);

[0112] Loss function: Binary Cross-Entropy;

[0113] Batch size: 256;

[0114] Iterations: 100

[0115] Regularization: L2 regularization (λ = 0.01) + Dropout (ratio = 0.2);

[0116] Differential privacy: Gaussian noise (σ=0.5)

[0117] }

[0118] (8) Continue configuring the test module and verification module, and adjust the parameters in (7) according to the results until the test set auc is greater than 0.75 and ks is greater than 0.25;

[0119] (9) The prediction results of vertical federation DNN and SecureBoost are:

[0120] prediction_result = α1*SecureBoost_prediction + α2*DNN_prediction, adjust α1 and α2 until the AUC and KS of the validation set are the highest;

[0121] (10) Transform the prediction_result into a score range of 500-1000, where:

[0122] ①[1000,900):AAA;

[0123] ②[900,800):AA;

[0124] ③[800,700):A;

[0125] ④[700,600):BBB,BB,B;

[0126] ⑤[600,550):BB;

[0127] ⑥[550,500):B.

[0128] The target prediction model input is obtained through the above method.

[0129] This application realizes credit assessment and monitoring covering the entire life cycle of technology-based enterprises; through the acquisition of multi-dimensional composite data, it realizes the true restoration of dimensions such as the enterprise's credit benchmark status, operating status, supply chain status, and investment and financing status; through an intelligent analysis system adapted to multi-dimensional composite data, it realizes the matching, verification, and dimensionality reduction of enterprise credit data, and the accurate output of enterprise credit analysis results.

[0130] Figure 2 A flow chart of an enterprise credit assessment method provided in an embodiment of the present application is shown, which is applied to an enterprise credit assessment system, wherein the enterprise credit assessment system is connected to a preset overall database; the enterprise credit assessment system includes: an input module, an acquisition module, a processing module, and an analysis module; wherein the method includes steps S101-S104; specifically:

[0131] S101. Obtain the corporate identity of the target scientific and technological innovation enterprise to be evaluated;

[0132] S102. Acquire, from the overall database, a plurality of data to be evaluated of the target enterprise based on the enterprise identifier of the target enterprise; wherein the data to be evaluated includes data in different dimensions and covers the entire life cycle of the target enterprise;

[0133] S103, processing the data to be evaluated to obtain sub-processing results of each data to be evaluated;

[0134] S104: Analyze the sub-processing results to obtain the enterprise credit analysis results of the target enterprise.

[0135] The data to be evaluated includes: real public data and real private data; the overall database includes a public database and a private database;

[0136] The step of obtaining the plurality of data to be evaluated of the target enterprise from the overall database includes:

[0137] Constructing a first privacy-preserving computing node in a public database corresponding to the real public data and constructing a second privacy-preserving computing node in a private database corresponding to the real private data;

[0138] Real public data is obtained through the first privacy computing node and real private data is obtained through the second privacy computing stage.

[0139] The first privacy-preserving computing node and the second privacy-preserving computing node are constructed in the following manner:

[0140] Constructing the first privacy-preserving computing node in a first target database, and obtaining real public data from a first database other than the first target database in the public database through the first privacy-preserving computing node; wherein the first target data is any database in the public database;

[0141] Construct the second privacy computing node in the second target database, and obtain the real public data in the second other database in the privacy database except the second target database through the second privacy computing node, wherein the second target data is any database in the privacy database.

[0142] The method further comprises:

[0143] If real public data is stored in the second privacy computing node, the second privacy computing node transmits the data to the first privacy computing node through oblivious transmission.

[0144] The analyzing of the sub-processing results to obtain the enterprise credit analysis result of the target enterprise includes:

[0145] The preset federated learning model is trained using the preset test public data and test private data to obtain the trained target prediction model.

[0146] The sub-processing result is input into the target prediction model to obtain the enterprise credit analysis result of the target enterprise output by the target prediction model.

[0147] The data to be evaluated include: public financial data, transaction data, invoice data and investment and financing data.

[0148] The processing of the data to be evaluated to obtain sub-processing results of each of the data to be evaluated includes:

[0149] Extracting the target enterprise's financial indicators from the public financial data, and establishing a benchmark status of the enterprise's credit based on the enterprise financial indicators;

[0150] Identify and classify the cash flow data and construct the true status of the enterprise's credit from the cash flow dimension;

[0151] Identify and classify the invoice data and construct the target enterprise's supply chain relationship network from a fiscal and tax perspective;

[0152] Based on the investment and financing data, the industry classification label, equity structure, investment and financing relationship network and valuation changes of the target enterprise are constructed from the investment and financing dimension.

[0153] When the embodiment of the present application is implemented, it can be Figure 3 The method shown is as follows: input the target enterprise, obtain the target enterprise's public (financial) data, enterprise bank flow data, enterprise invoice data, and enterprise investment and financing data. After processing each, the obtained verification data is comprehensively analyzed and the enterprise credit analysis results are output.

[0154] like Figure 4 As shown, an embodiment of the present application provides an electronic device for executing the enterprise credit assessment method in the present application, the device including a memory, a processor, a bus, and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the enterprise credit assessment method when executing the computer program.

[0155] Specifically, the above-mentioned memory and processor can be general-purpose memory and processor, which are not specifically limited here. When the processor runs the computer program stored in the memory, it can execute the above-mentioned enterprise credit assessment method.

[0156] Corresponding to the enterprise credit assessment method in the present application, an embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned enterprise credit assessment method are executed.

[0157] Specifically, the storage medium can be a general storage medium, such as a mobile disk, a hard disk, etc. When the computer program on the storage medium is run, the above-mentioned enterprise credit assessment method can be executed.

[0158] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. The system embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interface, indirect coupling or communication connection of the system or unit, which can be electrical, mechanical or other forms.

[0159] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0160] In addition, each functional unit in the embodiments provided in the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0161] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0162] It should be noted that similar numbers and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description and are not to be understood as indicating or implying relative importance.

[0163] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The scope of protection of the present application is not limited thereto. Although the present application has been described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-described embodiments within the technical scope disclosed in the present application, or perform equivalent replacements for some of the technical features thereof. However, these modifications, changes, or replacements do not deviate from the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application. They should all be included in the scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the scope of protection of the claims.

Claims

1. An enterprise credit evaluation system, characterized in that: The enterprise credit evaluation system is connected to a preset overall database; the enterprise credit evaluation system includes: an input module, an acquisition module, a processing module and an analysis module; The input module is used to input the corporate identity of the target scientific and technological innovation enterprise to be evaluated; The acquisition module is configured to acquire, from the overall database, a plurality of data to be evaluated of the target enterprise based on the enterprise identification of the target enterprise; wherein the data to be evaluated includes data in different dimensions and covers the entire life cycle of the target enterprise; The processing module is used to process the data to be evaluated to obtain sub-processing results of each data to be evaluated; The analysis module is used to analyze the sub-processing results to obtain the enterprise credit analysis results of the target enterprise.

2. The enterprise credit evaluation system according to claim 1, characterized in that: The data to be evaluated includes: real public data and real private data; the overall database includes a public database and a private database; When the acquisition module acquires the multiple data to be evaluated of the target enterprise from the overall database, the acquisition module includes: Constructing a first privacy-preserving computing node in a public database corresponding to the real public data and constructing a second privacy-preserving computing node in a private database corresponding to the real private data; Real public data is obtained through the first privacy computing node and real private data is obtained through the second privacy computing stage.

3. The enterprise credit evaluation system according to claim 2, characterized in that: The acquisition module constructs the first privacy-preserving computing node and the second privacy-preserving computing node in the following manner: Constructing the first privacy-preserving computing node in a first target database, and obtaining real public data from a first database other than the first target database in the public database through the first privacy-preserving computing node; wherein the first target data is any database in the public database; Construct the second privacy computing node in the second target database, and obtain the real public data in the second other database in the privacy database except the second target database through the second privacy computing node, wherein the second target data is any database in the privacy database.

4. The enterprise credit evaluation system according to claim 3, characterized in that: When constructing the first privacy-preserving computing node and the second privacy-preserving computing node, the acquisition module is further configured to: If real public data is stored in the second privacy computing node, the second privacy computing node transmits the data to the first privacy computing node through oblivious transmission.

5. The enterprise credit evaluation system according to claim 1, characterized in that: The method for analyzing the sub-processing result to obtain the enterprise credit analysis result of the target enterprise includes: The preset federated learning model is trained using the preset test public data and test private data to obtain the trained target prediction model. The sub-processing result is input into the target prediction model to obtain the enterprise credit analysis result of the target enterprise output by the target prediction model.

6. The enterprise credit evaluation system according to claim 1, characterized in that: The data to be evaluated include: public financial data, transaction data, invoice data and investment and financing data.

7. The enterprise credit evaluation system according to claim 6, characterized in that: The processing module is used to process the data to be evaluated to obtain sub-processing results of each of the data to be evaluated, including: Extracting the target enterprise's financial indicators from the public financial data, and establishing a benchmark status of the enterprise's credit based on the enterprise financial indicators; Identify and classify the cash flow data and construct the true status of the enterprise's credit from the cash flow dimension; Identify and classify the invoice data and construct the target enterprise's supply chain relationship network from a fiscal and tax perspective; Based on the investment and financing data, the industry classification label, equity structure, investment and financing relationship network and valuation changes of the target enterprise are constructed from the investment and financing dimension.

8. A method for evaluating corporate credit, characterized in that: Applied to the enterprise credit evaluation system according to any one of claims 1 to 7, the method comprises: Obtain the corporate logo of the target scientific and technological innovation enterprise to be evaluated; According to the enterprise identification of the target enterprise, a plurality of data to be evaluated of the target enterprise is obtained from the overall database; wherein the data to be evaluated includes data in different dimensions and covers the entire life cycle of the target enterprise; Processing the data to be evaluated to obtain sub-processing results of each of the data to be evaluated; The sub-processing results are analyzed to obtain an enterprise credit analysis result of the target enterprise.

9. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor and the memory communicate via the bus. When the machine-readable instructions are executed by the processor, the steps of the enterprise credit assessment method as claimed in claim 8 are performed.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which, when executed by a processor, executes the steps of the enterprise credit assessment method according to claim 8.

Images