A bank note intelligent decision-making system and method based on risk quantification model

Through the intelligent decision-making system of bank bills based on risk quantitative model, combining default risk, risk quantification and financial risk prediction, the risk prediction problem during bank bills is solved, achieving more accurate risk management and decision-making, and reducing default risk .

CN118628222BActive Publication Date: 2024-12-03ZHEJIANG YOUCAI CLOUD CHAIN TECH CO LTD
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202410666294.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-28
Publication Date
2024-12-03
Estimated Expiration
2044-05-28

AI Technical Summary

Technical Problem

The prior art has failed to effectively solve the risk prediction problem of bank notes when issuing or discounting, especially in terms of default risk, credit risk and liquidity risk management.

Method used

Adopt a bank bill intelligent decision-making system based on risk quantification model, and obtain enterprise feature data from historical databases, perform feature processing and model training, predict default risk probability and risk level, combine enterprise financial risk prediction, intelligent decision-making, and through life cycle Data tracking optimization decisions.

Benefits of technology

It realizes risk prediction and management of bank notes when issuing or discounting, reduces default risks, improves the accuracy and efficiency of decision-making, and enhances the ability to predict corporate financial risks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118628222B_ABST
    Figure CN118628222B_ABST
Patent Text Reader

Abstract

The present invention discloses a bank bill intelligent decision-making system and method based on a risk quantification model, which relates to the field of intelligent decision-making technology, including: obtaining company characteristic data of historical applicant companies from a historical applicant company database and performing characteristic processing to obtain a pre-characteristic data set; obtaining bank bill default information of historical applicant companies to obtain a post-characteristic data set; inputting the company characteristic data of the current applicant company into a bank bill default risk prediction model to predict and obtain a default risk probability attribute; inputting the data into a bank bill risk quantification model to obtain a first risk level; inputting the data into an enterprise financial risk prediction model to obtain a second risk level; based on the default risk probability attribute, the first risk level and the second risk level, performing bank bill intelligent decision-making to complete bank bill intelligent decision-making behavior; the present application can perform risk prediction when applying for issuance or discounting of bank bills, thereby reducing the bank bill default risk.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of intelligent decision-making technology, and in particular to a bank note intelligent decision-making system and method based on a risk quantification model. Background Art

[0002] Bank acceptance bill is a type of bank bill. It is issued by the payee or the acceptance applicant, and the acceptance applicant applies to the account opening bank, and the bank seals and acknowledges the payment on maturity. According to the agreement between the two parties, if the bank acceptance bill is issued by the payee, it should be submitted to the account opening bank for acceptance; if it is issued by the acceptance applicant, it should be accepted by the account opening bank. As a common type of commercial bill, bank acceptance bill is widely used in commercial activities and provides enterprises with a flexible and efficient financing channel.

[0003] At the same time, bank acceptance bills also have certain risks. Enterprises that issue bank acceptance bills may not be able to pay the bills in full and on time due to poor management or tight liquidity, resulting in the risk of bill default. In the process of bill circulation, bill fraud such as fraud and forgery may occur, causing losses to the holder. In addition, changes in the macroeconomic situation, such as inflation and exchange rate fluctuations, will also indirectly affect the repayment ability of enterprises, making bank acceptance bills face more uncertainty. Therefore, when banks carry out bank acceptance bill business, they need to attach great importance to the prevention and control of credit risk, operational risk and liquidity risk, establish a sound bill risk management mechanism, and conduct dynamic monitoring of bill acceptance throughout the life cycle to safeguard the healthy development of themselves and their client companies.

[0004] The Chinese patent application with the publication number CN111695908B discloses a method and device for predicting bill business risk, including: obtaining the bill business risk prediction data of the enterprise to be predicted; performing feature processing on the bill business risk prediction data of the enterprise to be predicted to obtain the bill business risk feature data of the enterprise to be predicted; obtaining the risk prediction result of the enterprise to be predicted based on the bill business risk feature data of the enterprise to be predicted and the abnormal bill transaction prediction model; wherein the abnormal bill transaction prediction model is obtained after training based on the bill business risk prediction sample data and the predetermined risk label. This method improves the prediction efficiency of the bill business risk of the enterprise, but fails to solve the problem of risk prediction of bank bills when issuing or discounting.

[0005] To this end, the present invention proposes a bank bill intelligent decision-making system and method based on a risk quantification model. Summary of the invention

[0006] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a bank bill intelligent decision-making system and method based on a risk quantification model, which realizes the risk prediction of bank bills when issuing or discounting, and reduces the risk of bank bill default.

[0007] The present invention aims at solving the technical problems existing in the prior art and provides a bank note intelligent decision-making method based on a risk quantification model. The specific steps are as follows:

[0008] S1. Obtain company characteristic data of historical applicants from a historical applicant database and perform characteristic processing to obtain a pre-characteristic data set; obtain bank bill default information of historical applicants, receive bank bill default status information of the current applicant, and obtain a post-characteristic data set based on the pre-characteristic data set; the bank bill refers to a bank acceptance bill;

[0009] S2. Inputting the company characteristic data of the current applicant enterprise into a bank bill default risk prediction model to predict the default risk probability attribute; the bank bill default risk prediction model is obtained by training the post-event characteristic data set;

[0010] S3, inputting the company characteristic data of the current applicant enterprise into a bank bill risk quantification model to obtain a first risk level; the bank bill risk quantification model is obtained by analyzing the prior characteristic data set;

[0011] S4, inputting the company information data of the current applicant enterprise into the enterprise financial risk prediction model to obtain a second risk level; the enterprise financial risk prediction model is obtained by training the prior feature data set;

[0012] S5. Based on the default risk probability attribute, the first risk level and the second risk level, a bank bill intelligent decision is made to complete a bank bill intelligent decision-making behavior;

[0013] S6. Obtain the bank bill intelligent decision-making behavior and conduct life cycle data tracking to obtain the bank bill default status information of the current applicant enterprise and feed it back to the post-event feature data set.

[0014] In particular, the company feature data of historical application companies are obtained from the historical application company database and feature processed to obtain the prior feature data set, where:

[0015] The company characteristic data of the historical applicant enterprise includes the account opening data, credit data, financial data, trade data and guarantee data submitted by the enterprise when applying for bank bills;

[0016] The feature processing includes irrelevant data processing and missing value processing;

[0017] The missing value processing method is:

[0018] If the missing rate of company characteristic data of an applicant enterprise exceeds 30%, the company characteristic data of the applicant enterprise will be discarded;

[0019] If the missing rate of the company characteristic data of an applicant enterprise does not exceed 30%, the mean method will be used to fill in the missing value data in the company characteristic data of the applicant enterprise.

[0020] In particular, the bank note default risk prediction model is obtained by training the post-event feature data set, including:

[0021] Step S11: preprocessing the post-feature data set to obtain a feature data table;

[0022] Step S12: dividing the feature data table into a training set and a validation set, inputting the sets into the TabNet model for training, and obtaining a bank bill default risk prediction model;

[0023] Step S13: Based on the bank bill default risk prediction model, predict the default risk probability attribute;

[0024] The post-event feature data set includes the bank bill default information of the historical applicant enterprise and the bank bill default status information of the current applicant enterprise;

[0025] The preprocessing includes filling in missing values, encoding discrete variables, and standardizing continuous variables.

[0026] In particular, the bank note risk quantification model is obtained by analyzing the prior feature data set, including:

[0027] Step S21: using the analytic hierarchy process to classify the indicators of the prior feature data set and calculate the weights of the indicators;

[0028] Step S22: normalizing the numerical value of the indicator to obtain a normalized value of the indicator;

[0029] Step S23: multiplying the index normalization value by the weight of the corresponding index to obtain a comprehensive score;

[0030] Step S24: grading the comprehensive scores, establishing a first risk assessment system, and obtaining a bank bill risk quantification model;

[0031] Step S25: inputting the company characteristic data of the current applicant enterprise into the bank bill risk quantification model to obtain the first risk level;

[0032] The first risk assessment system is divided into two risk levels, namely:

[0033] Level A: The overall score is greater than or equal to the second threshold;

[0034] Grade B: The overall score is less than the second threshold.

[0035] In particular, the enterprise financial risk prediction model is obtained by training the prior feature data set, including:

[0036] Step S31: Calculate the enterprise financial F score based on the prior feature data set, and use the grey correlation analysis method to screen out the key features of financial forecasting;

[0037] Step S32: Establishing a second risk assessment system based on the enterprise financial F score;

[0038] Step S33: Obtain corporate news reports, corporate social media, corporate announcements, and industry research report texts of the applicant company, establish a corporate sentiment analysis dataset, divide the corporate sentiment analysis dataset into a training set and a validation set, input the dataset into the BETR model for training, obtain a corporate sentiment analysis model, and predict a corporate sentiment analysis result label based on the corporate sentiment analysis model;

[0039] Step S34: Based on the key features of financial forecasting, the enterprise financial F score and the enterprise sentiment analysis result label, establish an enterprise financial risk forecasting data set, divide the enterprise financial risk forecasting data set into a training set and a validation set, input them into the LSTM model for training, obtain an enterprise financial risk forecasting model, and obtain the enterprise financial forecast F score based on the enterprise financial risk forecasting model;

[0040] Step S35: Substituting the enterprise financial forecast F score into the second risk assessment system to obtain a second risk level;

[0041] The second risk assessment system is divided into two risk levels, namely:

[0042] Grade A: The enterprise financial forecast F score is less than or equal to 0.0274;

[0043] Grade B: The enterprise financial forecast F score is greater than 0.0274;

[0044] The grey correlation analysis method comprises:

[0045] Determine a reference sequence and a comparison sequence, wherein the reference sequence is the enterprise financial F score, and the comparison sequence is the financial sequence data obtained from the financial data;

[0046] Using range standardization to process the financial series data into dimensionless data;

[0047] For each comparison sequence, calculate its correlation coefficient with the reference sequence;

[0048] According to the correlation coefficient of each comparison series, the correlation degree of each financial series data is calculated;

[0049] The financial series data with the top 50% correlation are selected as the key features for financial prediction.

[0050] In particular, the bank bill intelligent decision-making is performed based on the default risk probability attribute, the first risk level and the second risk level, and the bank bill intelligent decision-making behavior is completed, including:

[0051] Step S41: Determine the default risk probability attribute of the current applicant enterprise. If the default risk probability attribute is greater than or equal to the first threshold, the bank bill intelligent decision-making executes a rejection behavior; if the default risk probability attribute is less than the first threshold, the bank bill intelligent decision-making executes an approval behavior;

[0052] Step S42: Performing a first risk level judgment on the current applicant enterprise that has been approved by the bank bill intelligent decision-making execution in step S11;

[0053] Step S43: Perform a second risk level judgment on the current applicant enterprise that has been approved by the bank bill intelligent decision-making execution in step S11;

[0054] Step S44: Determine the first risk level in step S22 and the second risk level in step S23,

[0055] If both the first risk level and the second risk level are A, the bank bill intelligent decision-making will execute the pending action and submit it to manual review; if at least one of the first risk level and the second risk level is not A, the bank bill intelligent decision-making will execute the approved action;

[0056] The bank bill intelligent decision-making function is used when the bank bill is issued or discounted;

[0057] The first threshold is manually set according to a specific task before the bank note intelligent decision is executed;

[0058] The specific tasks mentioned include bank bill issuance and discounting;

[0059] The first risk level is determined by a first risk assessment system;

[0060] The second risk level is determined by a second risk assessment system.

[0061] In particular, the bank bill intelligent decision-making behavior is obtained and life cycle data is tracked to obtain the bank bill default status information of the current applicant enterprise, and is fed back to the post-event feature data set, wherein:

[0062] The life cycle data refers to the status data of the bank note during its entire life cycle from issuance to acceptance;

[0063] The bank note default status information refers to the default status information of the bank note at the time of acceptance;

[0064] The feedback to the subsequent feature dataset is used to update the subsequent feature dataset.

[0065] Another aspect of the present application provides a bank note intelligent decision-making system based on a risk quantification model, the system comprising:

[0066] Data processing module: used to obtain the company characteristic data of the historical applicant companies from the historical applicant company database and perform characteristic processing to obtain the pre-event characteristic data set; obtain the bank bill default information of the historical applicant companies to obtain the post-event characteristic data set;

[0067] Data receiving module: used to receive the bank bill default status information of the current applicant enterprise and update the post-event feature data set;

[0068] Default prediction module: used to input the company characteristic data of the current applicant enterprise into the bank bill default risk prediction model to predict the default risk probability attribute;

[0069] Risk quantification module: used to input the company characteristic data of the current applicant enterprise into the bank bill risk quantification model to obtain the first risk level;

[0070] Financial forecasting module: used to input the company information data of the current applicant enterprise into the enterprise financial risk forecasting model to obtain the second risk level;

[0071] Intelligent decision-making module: used to make intelligent decisions on bank bills based on the default risk probability attribute, the first risk level and the second risk level, and complete the intelligent decision-making behavior of bank bills.

[0072] Another aspect of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the above-mentioned intelligent decision-making method for bank bills based on a risk quantification model are implemented.

[0073] Another aspect of the present application provides a readable storage medium, which stores a computer program, and the computer program is suitable for loading by a processor to implement the steps in the above-mentioned bank bill intelligent decision-making method based on a risk quantification model.

[0074] Beneficial technical effects of the present invention:

[0075] While establishing a differentiated risk assessment system, the present invention enhances the ability to predict corporate financial risks by combining corporate sentiment analysis, fully integrating text information such as corporate-related news reports, social media, announcements, and industry research reports, and extracting the company's emotional labels through a sentiment analysis model. Especially in the current complex and changeable Internet economic environment, the impact of emotional factors on the financial risks of companies, especially listed companies, is becoming increasingly prominent. Incorporating this factor into the prediction model can significantly improve the accuracy of financial risk prediction for companies, especially listed companies.

[0076] The present invention makes intelligent decisions on bank bills by comprehensively considering various risk factors, and adopts three sub-modules: default risk prediction, risk quantification, and financial risk prediction. Default risk prediction focuses on the default tendency of bank bills themselves, risk quantification analyzes its risk level based on the overall characteristics of the enterprise, and financial risk prediction focuses on the financial health of the enterprise, reflecting the healthy status of the enterprise's acceptance. Integrating the risk assessment results of the three dimensions solves the limitations of a single data indicator or model, so that the bank bill intelligent decision-making system can make more comprehensive intelligent decisions. At the same time, the present invention uses characteristic data from both the pre- and post-stages, covering the entire life cycle information of the enterprise from application to final default or normal acceptance. This dynamically iterative data can not only make the prediction model more accurate, but also optimize and improve the system through a feedback mechanism, continuously improve the accuracy of decision-making, and further reduce the risk of bank bill default. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 It is a flow chart of a bank bill intelligent decision-making method based on a risk quantification model in Example 1 of the present invention.

[0078] Figure 2 This is a step of predicting and obtaining a default risk probability attribute in Embodiment 1 of the present invention.

[0079] Figure 3 This is a step of obtaining a first risk level in embodiment 1 of the present invention.

[0080] Figure 4 This is a step of obtaining a second risk level in embodiment 1 of the present invention.

[0081] Figure 5 This is a flow chart of intelligent decision making for bank bills according to Embodiment 1 of the present invention.

[0082] Figure 6 It is a bank bill intelligent decision-making system based on a risk quantification model provided by Example 2 of the present invention.

[0083] Figure 7 This is a schematic diagram of the structure of an electronic device provided in Example 3 of the present invention.

[0084] Figure 8 It is a schematic diagram of the structure of a computer-readable storage medium provided in Example 4 of the present invention. DETAILED DESCRIPTION

[0085] In order to better understand the present application, a more detailed description will be made of various aspects of the present application with reference to the accompanying drawings. It should be understood that these detailed descriptions are only descriptions of exemplary embodiments of the present application, and are not intended to limit the scope of the present application in any way. Throughout the specification, the same reference numerals refer to the same elements. The expression "and / or" includes any and all combinations of one or more of the associated listed items.

[0086] In the accompanying drawings, the size, dimensions and shapes of the elements have been slightly adjusted for ease of illustration. The drawings are for illustration only and are not strictly drawn to scale. As used herein, the terms "substantially", "approximately" and similar terms are used as terms of approximation, not as terms of degree, and are intended to illustrate the inherent deviations in measurements or calculations that will be recognized by those of ordinary skill in the art. In addition, in this application, the order in which the steps are described does not necessarily represent the order in which these processes occur in actual operation, unless otherwise specified or can be derived from the context.

[0087] It should also be understood that expressions such as "include", "including", "have", "contain" and / or "comprising" are open rather than closed expressions in this specification, which indicate the presence of the stated features, elements and / or components, but do not exclude the presence of one or more other features, elements, components and / or combinations thereof. In addition, when expressions such as "at least one of..." appear after a list of listed features, they modify the entire list of features rather than just the individual elements in the list. In addition, when describing embodiments of the present application, "may" is used to mean "one or more embodiments of the present application". And, the term "exemplary" is intended to refer to an example or illustration.

[0088] Unless otherwise defined, all words (including engineering terms and scientific and technological terms) used in this article have the same meaning as those commonly understood by ordinary technicians in the field to which this application belongs. It should also be understood that unless there is a clear explanation in this application, words defined in commonly used dictionaries should be interpreted as having the same meaning as their meaning in the context of the relevant technology, and should not be interpreted in an idealized or overly formal sense.

[0089] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0090] Example 1

[0091] like Figure 1 to Figure 5 As shown, this embodiment provides a bank bill intelligent decision-making method based on a risk quantification model;

[0092] like Figure 1 As shown, the steps of the bank note intelligent decision-making system and method based on the risk quantification model include:

[0093] S1. Obtain company characteristic data of historical applicants from a historical applicant database and perform characteristic processing to obtain a pre-characteristic data set; obtain bank bill default information of historical applicants, receive bank bill default status information of the current applicant, and obtain a post-characteristic data set based on the pre-characteristic data set; the bank bill refers to a bank acceptance bill;

[0094] In particular, the company feature data of historical application companies are obtained from the historical application company database and feature processed to obtain the prior feature data set, where:

[0095] The company characteristic data of the historical applicant enterprise includes the account opening data, credit data, financial data, trade data and guarantee data submitted by the enterprise when applying for bank bills;

[0096] Preferably, the company characteristic data of the historical applicant enterprise includes the data submitted by the same enterprise when applying for different bank bills at different times, and the data submitted each time will be stored in the historical applicant enterprise database, with the company name and application number as the primary key;

[0097] The feature processing includes irrelevant data processing and missing value processing;

[0098] The missing value processing method is:

[0099] If the missing rate of company characteristic data of an applicant enterprise exceeds 30%, the company characteristic data of the applicant enterprise will be discarded;

[0100] If the missing rate of the company characteristic data of an applicant enterprise does not exceed 30%, the mean method will be used to fill in the missing value data in the company characteristic data of the applicant enterprise.

[0101] S2. Inputting the company characteristic data of the current applicant enterprise into a bank bill default risk prediction model to predict the default risk probability attribute; the bank bill default risk prediction model is obtained by training the post-event characteristic data set;

[0102] In particular, see Figure 2 The bank note default risk prediction model is obtained by training the post-feature data set, and a step of predicting the default risk probability attribute is:

[0103] Step S11: preprocessing the post-feature data set to obtain a feature data table;

[0104] Step S12: dividing the feature data table into a training set and a validation set, inputting the sets into the TabNet model for training, and obtaining a bank bill default risk prediction model;

[0105] Step S13: Based on the bank bill default risk prediction model, predict the default risk probability attribute;

[0106] The post-event feature data set includes the bank bill default information of the historical applicant enterprise and the bank bill default status information of the current applicant enterprise;

[0107] The preprocessing includes filling in missing values, encoding discrete variables, and standardizing continuous variables.

[0108] S3, inputting the company characteristic data of the current applicant enterprise into a bank bill risk quantification model to obtain a first risk level; the bank bill risk quantification model is obtained by analyzing the prior characteristic data set;

[0109] In particular, see Figure 3 The bank note risk quantification model is obtained by analyzing the prior feature data set, and a step of obtaining the first risk level is:

[0110] Step S21: using the analytic hierarchy process to classify the indicators of the prior feature data set and calculate the weights of the indicators;

[0111] Step S22: normalizing the numerical value of the indicator to obtain a normalized value of the indicator;

[0112] Step S23: multiplying the index normalization value by the weight of the corresponding index to obtain a comprehensive score;

[0113] Step S24: grading the comprehensive scores, establishing a first risk assessment system, and obtaining a bank bill risk quantification model;

[0114] Step S25: inputting the company characteristic data of the current applicant enterprise into the bank bill risk quantification model to obtain the first risk level;

[0115] The first risk assessment system is divided into two risk levels, namely:

[0116] Level A: The overall score is greater than or equal to the second threshold;

[0117] Grade B: The overall score is less than the second threshold.

[0118] Furthermore, the steps of AHP are:

[0119] (1) For each level of indicators, compare their relative importance pairwise, assign the comparison results using the 1-9 scaling method, and construct a judgment matrix A; for multiple indicators at the same level, the pairwise comparison judgment matrix can be obtained: , the values ​​in matrix A must satisfy ; The following table shows the scales and meanings of the judgment matrix 1-9;

[0120]

[0121] (2) Calculation of indicator weights :

[0122]

[0123]

[0124] Among them, Mi is the numerical product of each row of the judgment matrix, is the element in the judgment matrix;

[0125] (3) Perform consistency check, first calculate the consistency index :

[0126]

[0127]

[0128]

[0129] in, is the maximum eigenvalue of the judgment matrix;

[0130] Query the average random consistency index , as shown in the table:

[0131] Degree 1 2 3 4 5 6 7 8 9 #timg# 0.00 0.00 0.58 0.90 1.12 1.24 1.32 1.41 1.45

[0132] Finally, calculate the consistency ratio :

[0133]

[0134] Verification results:

[0135] like When , the judgment matrix has complete consistency;

[0136] like When , the judgment matrix has satisfactory consistency;

[0137] like , then the judgment matrix has unsatisfactory consistency and should be adjusted.

[0138] S4, inputting the company information data of the current applicant enterprise into the enterprise financial risk prediction model to obtain a second risk level; the enterprise financial risk prediction model is obtained by training the prior feature data set;

[0139] In particular, see Figure 4 The enterprise financial risk prediction model is obtained by training the prior feature data set, and a step of obtaining the second risk level is:

[0140] Step S31: Calculate the enterprise financial F score based on the prior feature data set, and use the grey correlation analysis method to screen out the key features of financial forecasting;

[0141] Furthermore, the corporate financial F score is calculated by the following formula:

[0142]

[0143] in, (Current assets at the end of the period - current liabilities at the end of the period) / total assets at the end of the period;

[0144] Ending retained earnings / ending total assets;

[0145] (net income after tax + depreciation) / average total debt;

[0146] Market value of shareholders' equity at the end of the period / total liabilities at the end of the period;

[0147] (Net income after tax + interest + depreciation) / average total assets.

[0148] Step S32: Establishing a second risk assessment system based on the enterprise financial F score;

[0149] Step S33: Obtain corporate news reports, corporate social media, corporate announcements, and industry research report texts of the applicant company, establish a corporate sentiment analysis dataset, divide the corporate sentiment analysis dataset into a training set and a validation set, input the dataset into the BETR model for training, obtain a corporate sentiment analysis model, and predict a corporate sentiment analysis result label based on the corporate sentiment analysis model;

[0150] Preferably, the acquired text data is subjected to preprocessing operations including but not limited to cleaning, word segmentation, and removal of stop words, and the preprocessed text data is annotated with sentiment tendencies (positive, negative, neutral) as labels to construct an enterprise sentiment analysis dataset.

[0151] Step S34: Based on the key features of financial forecasting, the enterprise financial F score and the enterprise sentiment analysis result label, establish an enterprise financial risk forecasting data set, divide the enterprise financial risk forecasting data set into a training set and a validation set, input them into the LSTM model for training, obtain an enterprise financial risk forecasting model, and obtain the enterprise financial forecast F score based on the enterprise financial risk forecasting model;

[0152] Preferably, when establishing an enterprise financial risk prediction dataset, more in-depth feature engineering processing can be performed on the key features of financial predictions and enterprise sentiment analysis result labels, including but not limited to normalization, standardization or principal component analysis of data values.

[0153] In the enterprise financial risk prediction dataset, the enterprise sentiment analysis result label is used as an independent input feature with the same function as the key feature of financial prediction. In the model training phase, the key features of financial prediction after feature engineering and the enterprise sentiment analysis result label are used as input, and the enterprise financial F score is used as the supervision target to train the financial risk prediction model.

[0154] Step S35: Substituting the enterprise financial forecast F score into the second risk assessment system to obtain a second risk level;

[0155] The second risk assessment system is divided into two risk levels, namely:

[0156] Grade A: The enterprise financial forecast F score is less than or equal to 0.0274;

[0157] Grade B: The enterprise financial forecast F score is greater than 0.0274;

[0158] The grey correlation analysis method comprises:

[0159] Determine a reference sequence and a comparison sequence, wherein the reference sequence is the enterprise financial F score, and the comparison sequence is the financial sequence data obtained from the financial data;

[0160] Using range standardization to process the financial series data into dimensionless data;

[0161] For each comparison sequence, calculate its correlation coefficient with the reference sequence;

[0162] According to the correlation coefficient of each comparison series, the correlation degree of each financial series data is calculated;

[0163] The financial series data with the top 50% correlation are selected as the key features for financial prediction.

[0164] Grey correlation analysis is a nonlinear correlation analysis method that can discover the inherent laws hidden in complex data. Grey correlation analysis can effectively screen out key financial indicators that have an important impact on corporate financial risks, provide better input data for the establishment of subsequent corporate financial risk prediction models, and thus improve the accuracy and robustness of the model.

[0165] In the current complex and ever-changing Internet economic environment, the impact of emotional factors on the financial risks of enterprises, especially listed companies, is becoming increasingly prominent. Using sentiment analysis models to extract enterprise sentiment analysis result labels and incorporating this factor into the prediction model can significantly improve the accuracy of financial risk prediction for enterprises, especially listed companies.

[0166] S5. Based on the default risk probability attribute, the first risk level and the second risk level, a bank bill intelligent decision is made to complete a bank bill intelligent decision-making behavior;

[0167] In particular, see Figure 5 , a process of intelligent decision-making for bank bills is:

[0168] Step S41: Determine the default risk probability attribute of the current applicant enterprise. If the default risk probability attribute is greater than or equal to the first threshold, the bank bill intelligent decision-making executes a rejection behavior; if the default risk probability attribute is less than the first threshold, the bank bill intelligent decision-making executes an approval behavior;

[0169] Step S42: Performing a first risk level judgment on the current applicant enterprise that has been approved by the bank bill intelligent decision-making execution in step S11;

[0170] Step S43: Perform a second risk level judgment on the current applicant enterprise that has been approved by the bank bill intelligent decision-making execution in step S11;

[0171] Step S44: Determine the first risk level in step S22 and the second risk level in step S23,

[0172] If both the first risk level and the second risk level are A, the bank bill intelligent decision-making will execute the pending action and submit it to manual review; if at least one of the first risk level and the second risk level is not A, the bank bill intelligent decision-making will execute the approved action;

[0173] The bank bill intelligent decision-making function is used when the bank bill is issued or discounted;

[0174] The first threshold is manually set according to a specific task before the bank note intelligent decision is executed;

[0175] The specific tasks mentioned include bank bill issuance and discounting;

[0176] The first risk level is determined by a first risk assessment system;

[0177] The second risk level is determined by a second risk assessment system;

[0178] The bank bill intelligent decision-making adopts a hierarchical intelligent decision-making mechanism. First, the first threshold is set according to the default risk probability attribute for preliminary risk screening, which can quickly exclude some applicants with clear high risks, thereby improving the overall decision-making efficiency. For enterprises that pass the initial screening, their first and second risk levels are further judged, and decisions on whether to agree or wait are made based on different combinations. This multi-level and differentiated decision-making logic achieves the effect of refined risk control by the bank bill intelligent decision-making system.

[0179] Intelligent bank bill decision-making not only sets up the logic of automated decision-making, but also retains the link of manual review. When the first risk level and the second risk level are both A, the system will hand over the application to manual for final review. Automated decision-making can improve efficiency, while manual review can make up for the limitations of machines in complex decision-making and ensure the accuracy and rationality of decisions.

[0180] The first threshold can be dynamically adjusted manually according to specific bank bill business tasks such as issuance or discounting. The flexible threshold setting mechanism allows banks to exercise fine-grained control over the decision-making behavior of the bank bill intelligent decision-making system based on different business scenarios and risk preferences.

[0181] S6. Obtain the bank bill intelligent decision-making behavior and conduct life cycle data tracking to obtain the bank bill default status information of the current applicant enterprise and feed it back to the post-event feature data set.

[0182] In particular, the life cycle data refers to the status data of the bank note during its entire life cycle from issuance to acceptance;

[0183] The bank note default status information refers to the default status information of the bank note at the time of acceptance;

[0184] The feedback to the subsequent feature dataset is used to update the subsequent feature dataset.

[0185] Feedback the bank bill default status information of the current applicant enterprise to the post-event feature data set to form a closed-loop data collection and feedback process, which can continuously update the feature data set, thereby optimizing and improving the prediction accuracy of the bank bill default risk prediction model. Based on the continuous learning mechanism of the full life cycle data, the bank bill intelligent decision-making system can further improve the level of intelligent decision-making.

[0186] Example 2

[0187] Figure 6 An embodiment of the present application provides a bank note intelligent decision-making system based on a risk quantification model. The system as a whole can be divided into:

[0188] Data processing module: used to obtain the company characteristic data of the historical applicant companies from the historical applicant company database and perform characteristic processing to obtain the pre-event characteristic data set; obtain the bank bill default information of the historical applicant companies to obtain the post-event characteristic data set;

[0189] Data receiving module: used to receive the bank bill default status information of the current applicant enterprise and update the post-event feature data set;

[0190] Default prediction module: used to input the company characteristic data of the current applicant enterprise into the bank bill default risk prediction model to predict the default risk probability attribute;

[0191] Risk quantification module: used to input the company characteristic data of the current applicant enterprise into the bank bill risk quantification model to obtain the first risk level;

[0192] Financial forecasting module: used to input the company information data of the current applicant enterprise into the enterprise financial risk forecasting model to obtain the second risk level;

[0193] Intelligent decision-making module: used to make intelligent decisions on bank bills based on the default risk probability attribute, the first risk level and the second risk level, and complete the intelligent decision-making behavior of bank bills.

[0194] Example 3

[0195] Figure 7 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Figure 7 As shown, according to another aspect of the present application, an electronic device 100 is also provided. The electronic device 100 may include one or more processors and one or more memories. The memories store computer readable codes, and when the computer readable codes are executed by one or more processors, a bank bill intelligent decision method based on a risk quantification model as described above may be executed.

[0196] The method or device according to the embodiment of the present application can also be used by Figure 7 The electronic device architecture shown in FIG. Figure 7As shown, the electronic device 100 may include a bus 101, one or more CPUs 102, a read-only memory (ROM) 103, a random access memory (RAM) 104, a communication port 105 connected to a network, an input / output component 106, a hard disk 107, etc. The storage device in the electronic device 100, such as the ROM 103 or the hard disk 107, may store a bank bill intelligent decision-making method based on a risk quantification model provided in the present application. A bank bill intelligent decision-making method based on a risk quantification model may, for example, include the following steps: obtaining company feature data of historical applicants from a historical applicant enterprise database and performing feature processing to obtain a prior feature data set; obtaining bank bill default information of historical applicants, receiving bank bill default status information of the current applicant enterprise, and obtaining a post-feature data set based on the prior feature data set; the bank bill refers to a bank acceptance bill; inputting the company feature data of the current applicant enterprise into a bank bill default risk prediction model to predict a default risk probability attribute; the bank bill default risk prediction model is obtained by training the post-feature data set; the company feature data of the current applicant enterprise is input ... The characteristic data is input into a bank bill risk quantification model to obtain a first risk level; the bank bill risk quantification model is obtained by analyzing the prior characteristic data set; the company information data of the current applicant enterprise is input into an enterprise financial risk prediction model to obtain a second risk level; the enterprise financial risk prediction model is obtained by training the prior characteristic data set; based on the default risk probability attribute, the first risk level and the second risk level, bank bill intelligent decision-making is performed to complete the bank bill intelligent decision-making behavior; the bank bill intelligent decision-making behavior is obtained and life cycle data tracking is performed to obtain the bank bill default status information of the current applicant enterprise, and feedback to the post-characteristic data set.

[0197] Furthermore, the electronic device 100 may further include a user interface 108. Figure 7 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 7 One or more components of an electronic device are shown.

[0198] Example 4

[0199] Figure 8 Schematic diagram of a computer-readable storage medium structure provided by an embodiment of the present application. Figure 8As shown, a computer-readable storage medium 200 according to an embodiment of the present application is shown. Computer-readable instructions are stored on the computer-readable storage medium 200. When the computer-readable instructions are executed by the processor, a bank note intelligent decision-making method based on a risk quantification model according to an embodiment of the present application described with reference to the above figures can be executed. The computer-readable storage medium 200 includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0200] In addition, according to the implementation of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, the non-transitory machine-readable storage medium stores machine-readable instructions, the machine-readable instructions can be run by a processor to execute instructions corresponding to the method steps provided by the present application, and when the computer program is executed by a central processing unit (CPU), the above functions defined in the method of the present application are executed.

[0201] The methods, apparatuses, and devices of the present application may be implemented in many ways. For example, the methods, apparatuses, and devices of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present application are not limited to the order specifically described above unless otherwise specified. In addition, in some embodiments, the present application may also be implemented as programs recorded in a recording medium, which include machine-readable instructions for implementing the method according to the present application. Therefore, the present application also covers recording media storing programs for executing the method according to the present application.

[0202] In addition, the parts of the above-mentioned technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0203] The specific implementation modes as described above further describe the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation mode of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0204] The above preset parameters or preset thresholds are all set by technicians in this field according to actual conditions or obtained through large-scale data simulation.

[0205] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A bank note intelligent decision-making method based on a risk quantification model, characterized in that: The following steps are involved: S1. Obtain company characteristic data of historical applicants from the historical applicant enterprise database and perform characteristic processing to obtain a pre-event characteristic data set; obtain bank bill default information of historical applicants, receive bank bill default status information of current applicants, and obtain a post-event characteristic data set based on the pre-event characteristic data set; The bank note referred to is a bank acceptance bill; S2. Input the company characteristic data of the current applicant enterprise into the bank bill default risk prediction model to predict the default risk probability attribute; The bank note default risk prediction model is obtained by training the post-event feature data set; S3. Input the company characteristic data of the current applicant enterprise into the bank bill risk quantification model to obtain the first risk level; The bank note risk quantification model is obtained by analyzing the prior feature data set; S4. Input the company information data of the current applicant enterprise into the enterprise financial risk prediction model to obtain the second risk level; The enterprise financial risk prediction model is obtained by training the prior feature data set; S5. Based on the default risk probability attribute, the first risk level and the second risk level, a bank bill intelligent decision is made to complete a bank bill intelligent decision-making behavior; S6. Obtain the bank bill intelligent decision-making behavior and perform life cycle data tracking to obtain the bank bill default status information of the current applicant enterprise, and feed it back to the post-event feature data set; The bank note default risk prediction model is obtained by training the post-event feature data set, and includes: Step S11: preprocessing the post-feature data set to obtain a feature data table; Step S12: dividing the feature data table into a training set and a validation set, inputting the sets into the TabNet model for training, and obtaining a bank bill default risk prediction model; Step S13: Based on the bank bill default risk prediction model, predict the default risk probability attribute; The post-event feature data set includes the bank bill default information of the historical applicant enterprise and the bank bill default status information of the current applicant enterprise; The preprocessing includes filling missing values, encoding discrete variables, and standardizing continuous variables; The bank note risk quantification model is obtained by analyzing the prior feature data set, and includes: Step S21: using the analytic hierarchy process to classify the indicators of the prior feature data set and calculate the weights of the indicators; Step S22: normalizing the numerical value of the indicator to obtain a normalized value of the indicator; Step S23: multiplying the index normalization value by the weight of the corresponding index to obtain a comprehensive score; Step S24: grading the comprehensive scores, establishing a first risk assessment system, and obtaining a bank bill risk quantification model; Step S25: inputting the company characteristic data of the current applicant enterprise into the bank bill risk quantification model to obtain the first risk level; The first risk assessment system is divided into two risk levels, namely: Level A: The overall score is greater than or equal to the second threshold; Grade B: The overall score is less than the second threshold; The enterprise financial risk prediction model is obtained by training the prior feature data set, and includes: Step S31: Calculate the enterprise financial F score based on the prior feature data set, and use the grey correlation analysis method to screen out the key features of financial forecasting; Step S32: Establishing a second risk assessment system based on the enterprise financial F score; Step S33: Obtain corporate news reports, corporate social media, corporate announcements, and industry research report texts of the applicant company, establish a corporate sentiment analysis dataset, divide the corporate sentiment analysis dataset into a training set and a validation set, input the dataset into the BETR model for training, obtain a corporate sentiment analysis model, and predict a corporate sentiment analysis result label based on the corporate sentiment analysis model; Step S34: Based on the key features of financial forecasting, the enterprise financial F score and the enterprise sentiment analysis result label, establish an enterprise financial risk forecasting data set, divide the enterprise financial risk forecasting data set into a training set and a validation set, input them into the LSTM model for training, obtain an enterprise financial risk forecasting model, and obtain the enterprise financial forecast F score based on the enterprise financial risk forecasting model; Step S35: Substituting the enterprise financial forecast F score into the second risk assessment system to obtain a second risk level; The second risk assessment system is divided into two risk levels, namely: Grade A: The enterprise financial forecast F score is less than or equal to 0.0274; Grade B: The enterprise financial forecast F score is greater than 0.0274; The grey correlation analysis method comprises: Determine a reference sequence and a comparison sequence, wherein the reference sequence is the enterprise financial F score, and the comparison sequence is financial sequence data obtained from financial data; Using range standardization to process the financial series data into dimensionless data; For each comparison sequence, calculate its correlation coefficient with the reference sequence; According to the correlation coefficient of each comparison series, the correlation degree of each financial series data is calculated; The top 50% of the financial series data in terms of correlation are selected as the key features for financial prediction.

2. The bank note intelligent decision-making method based on risk quantification model according to claim 1 is characterized in that: The company characteristic data of the historical application enterprises are obtained from the historical application enterprise database and feature processed to obtain a pre-feature data set, wherein: The company characteristic data of the historical applicant enterprise includes the account opening data, credit data, financial data, trade data and guarantee data submitted by the enterprise when applying for bank bills; The feature processing includes irrelevant data processing and missing value processing; The missing value processing method is: If the missing rate of company characteristic data of an applicant enterprise exceeds 30%, the company characteristic data of the applicant enterprise will be discarded; If the missing rate of the company characteristic data of an applicant enterprise does not exceed 30%, the mean method will be used to fill in the missing value data in the company characteristic data of the applicant enterprise.

3. The bank note intelligent decision-making method based on risk quantification model according to claim 1 is characterized in that: The performing of bank bill intelligent decision-making based on the default risk probability attribute, the first risk level and the second risk level to complete the bank bill intelligent decision-making behavior includes: Step S41: Determine the default risk probability attribute of the current applicant enterprise. If the default risk probability attribute is greater than or equal to the first threshold, the bank bill intelligent decision-making executes a rejection behavior; if the default risk probability attribute is less than the first threshold, the bank bill intelligent decision-making executes an approval behavior; Step S42: Performing a first risk level judgment on the current applicant enterprise that has been approved by the bank bill intelligent decision-making execution in step S11; Step S43: Perform a second risk level judgment on the current applicant enterprise that has been approved by the bank bill intelligent decision-making execution in step S11; Step S44: Determine the first risk level in step S22 and the second risk level in step S23, If both the first risk level and the second risk level are A, the bank bill intelligent decision-making will execute the pending action and submit it to manual review; if at least one of the first risk level and the second risk level is not A, the bank bill intelligent decision-making will execute the approved action; The bank bill intelligent decision-making function is used when the bank bill is issued or discounted; The first threshold is manually set according to a specific task before the bank note intelligent decision is executed; The specific tasks mentioned include bank bill issuance and discounting; The first risk level is determined by a first risk assessment system; The second risk level is determined by a second risk assessment system.

4. The bank note intelligent decision-making method based on risk quantification model according to claim 3 is characterized in that: The bank bill intelligent decision-making behavior is obtained and life cycle data is tracked to obtain the bank bill default status information of the current applicant enterprise and feed it back to the post-event feature data set, wherein: The life cycle data refers to the status data of the bank note during its entire life cycle from issuance to acceptance; The bank note default status information refers to the default status information of the bank note at the time of acceptance; The feedback to the subsequent feature dataset is used to update the subsequent feature dataset.

5. A bank note intelligent decision system based on a risk quantification model, which is implemented based on a bank note intelligent decision method based on a risk quantification model as claimed in any one of claims 1 to 4, characterized in that: The system comprises: Data processing module: used to obtain the company characteristic data of the historical applicant companies from the historical applicant company database and perform characteristic processing to obtain the pre-event characteristic data set; obtain the bank bill default information of the historical applicant companies to obtain the post-event characteristic data set; Data receiving module: used to receive the bank bill default status information of the current applicant enterprise and update the post-event feature data set; Default prediction module: used to input the company characteristic data of the current applicant enterprise into the bank bill default risk prediction model to predict the default risk probability attribute; Risk quantification module: used to input the company characteristic data of the current applicant enterprise into the bank bill risk quantification model to obtain the first risk level; Financial forecasting module: used to input the company information data of the current applicant enterprise into the enterprise financial risk forecasting model to obtain the second risk level; Intelligent decision-making module: used to make intelligent decisions on bank bills based on the default risk probability attribute, the first risk level and the second risk level, and complete the intelligent decision-making behavior of bank bills.

6. An electronic device, characterized in that: It comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of a bank bill intelligent decision-making method based on a risk quantification model as described in any one of claims 1 to 4.

7. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor to execute the steps in a bank bill intelligent decision-making method based on a risk quantification model as described in any one of claims 1-4.

Citation Information

Patent Citations

  • A method and device for predicting bill business risk

    CN111695908B

  • Bill business risk prediction method and device

    CN111695908A

  • Data processing method, device and equipment and storage medium

    CN112070593A

  • Fund mobility risk early warning method and device and electronic equipment

    CN112801456A

  • Enterprise default risk prediction method, computer equipment and storage medium

    CN115081716A