Intelligent management method and system based on deep learning technology

Abstract

The invention discloses an intelligent management method and system based on a deep learning technology. The method comprises the following steps: receiving loan purpose information; determining a loan purpose feature corresponding to the loan purpose information, and determining a risk assessment index corresponding to the loan object; under the condition that the loan purpose feature is the production purpose, determining first loan limit information; under the condition that the loan purpose feature is the infrastructure construction purpose, determining second loan limit information; and determining loan contract information according to the loan limit information, and sending the loan contract information to the target device. The technical problem that the promotion efficiency of the whole loan process is relatively low due to the fact that different loan limits cannot be determined for different loan purposes of loan objects of village and town banks in combination with an artificial intelligence algorithm in related technologies is solved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and more specifically, to an intelligent management method and system based on deep learning technology. Background Art

[0002] In small financial institutions such as village banks, the loan approval process is often subject to the following technical limitations, which makes it impossible to combine artificial intelligence algorithms to determine different loan amounts for different borrowing purposes of borrowers, resulting in low efficiency in the overall loan process. Specifically, the loan approval process of village banks is often highly dependent on manual review, including key links such as judgment of the purpose of the loan, credit assessment of the borrower, and determination of the loan amount. This process is prone to bottlenecks when faced with a large number of applications, low approval efficiency, and may affect the rationality of the loan amount due to the subjectivity and inconsistency of manual judgment.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The present application provides an intelligent management method and system based on deep learning technology, so as to at least solve the technical problem that the overall loan process is inefficient because the relevant technology cannot combine artificial intelligence algorithms to determine different loan amounts according to the different borrowing purposes of the borrowers of village banks, resulting in low efficiency in promoting the overall loan process.

[0005] According to one aspect of the present application, an intelligent management method based on deep learning technology is provided, including: receiving information on the purpose of borrowing, wherein the information on the purpose of borrowing includes: a borrowing object; using a classification model to determine the borrowing purpose characteristics corresponding to the borrowing purpose information, and using a machine learning model to determine the risk assessment index corresponding to the borrowing object; when the borrowing purpose characteristics are production purposes, determining the first loan amount information according to the market price information of the means of production in the borrowing purpose information and the risk assessment index; when the borrowing purpose characteristics are infrastructure construction purposes, determining the second loan amount information according to the construction cost, expected service life, expected future income and risk assessment index of the infrastructure; determining the loan contract information according to the loan amount information, and sending the loan contract information to the target device, wherein the loan amount information includes: the first loan amount information and the second loan amount information.

[0006] Optionally, the classification model is obtained by training in the following method: collecting text data containing information on the purpose of borrowing from multiple data sources, wherein the data sources include: loan application documents, credit business records, customer questionnaire results, and communication records between customers and financial institution staff; the text data includes natural language text used to describe the characteristics of the purpose of borrowing; performing data cleaning on the text data, and performing word segmentation processing on the cleaned data to obtain word segmentation results, wherein data cleaning includes: removing preset special characters, correcting spelling errors, and unifying uppercase and lowercase formats; performing lexical analysis on the word segmentation results to extract basic vocabulary units in the text data to obtain target text data, wherein lexical analysis includes: part-of-speech tagging, stem extraction, and word form restoration operations; using natural language processing technology to convert the target text data into a semantic feature vector; using the semantic feature vector and the target label corresponding to the semantic feature vector to train the classification model, and completing the training of the classification model when the loss function of the classification model meets the preset convergence condition, wherein the target label includes: production purpose, infrastructure construction purpose, educational purpose, and medical purpose.

[0007] Optionally, the first loan amount information is determined based on the market price information of the means of production in the loan purpose information and the risk assessment index, including: obtaining multiple market price information of the means of production in the loan purpose information through different data sources, wherein the different data sources include: a market price database and an industry information platform; performing weighted summation on the multiple market price information to obtain the target market price information; calculating the total value of the means of production based on the quantity information of the means of production in the loan purpose information and the target market price information; when the risk assessment index is greater than a first preset threshold, multiplying the total value of the means of production by a first coefficient to obtain the loan amount information, wherein the first coefficient is less than 1; when the risk assessment index is less than the first preset threshold and greater than the second preset threshold, multiplying the total value of the means of production by a second coefficient to obtain the loan amount information, wherein the second preset threshold is less than the first preset threshold and the second coefficient is equal to 1; when the risk assessment index is less than the first preset threshold and greater than the second preset threshold, When the risk assessment index is greater than the third preset threshold, the target value is multiplied by the fourth coefficient to obtain the loan amount information, wherein the fourth coefficient is less than 1; when the risk assessment index is less than the third preset threshold and greater than the fourth preset threshold, the target value is multiplied by the fifth coefficient to obtain the loan amount information, wherein the fourth preset threshold is less than the third preset threshold and the fifth coefficient is equal to 1; when the risk assessment index is less than the fourth preset threshold, the target value is multiplied by the sixth coefficient to obtain the second loan amount information, wherein the sixth coefficient is greater than 1.

[0008] Optionally, the machine learning model is trained by the following method: obtaining basic information data, credit history data and transaction flow data of the target object, wherein the basic information data at least includes: name, gender, age, the credit history data at least includes: loan information, the loan information includes: loan type, loan amount, loan term, loan issuance time, repayment information, the repayment information at least includes: repaid amount, unpaid amount, number of overdue payments, overdue duration, transaction flow data includes: timestamp of each transaction, transaction amount, transaction direction, transaction category and counterparty information; integrating the basic information data, credit history data and transaction flow data, A first feature vector related to risk assessment is selected from the integrated data, wherein the first feature includes: age group, credit history indicator, transaction behavior characteristics, credit history indicators include: average overdue duration, total overdue times, loan default rate, and transaction behavior characteristics include: average monthly transaction times, ratio of expenditure to income; the first feature vector is normalized to obtain a second feature vector; the second feature vector is divided into a training set, a validation set, and a test set; the machine learning model is trained using the training set, the parameters of the model are adjusted through an optimization algorithm, the model is validated using the validation set during the training process, the performance of the trained model is evaluated using the test set, and a machine learning model that has completed the training is obtained when the preset evaluation indicators meet the preset conditions.

[0009] Optionally, after determining the loan contract information based on the loan amount information and sending the loan contract information to the target device, the method also includes: recording and storing the identification information of the borrower, and collecting the repayment information of the borrower corresponding to the identification information based on a preset time interval or a preset trigger event; converting the repayment information of the borrower corresponding to the identification information into a target code suitable for the machine learning model, and adjusting the hyperparameters of the machine learning model based on the target code.

[0010] Optionally, the loan contract information is determined based on the loan amount information, including: when the loan amount information meets preset conditions, the loan approval form including the loan amount information is sent to the first device, wherein the preset conditions include at least one of the following: the loan amount exceeds a preset amount threshold, the number of overdue repayments by the borrower exceeds a preset number or the overdue repayment period exceeds a preset period, the means of production in the loan purpose characteristics are changed from preset low-risk means of production to preset high-risk means of production, and the price fluctuation of agricultural products in the target area exceeds a preset range; receiving the verification result of the loan approval form sent by the first device, and generating the loan contract information based on the verification result.

[0011] According to another aspect of the present application, there is also provided an intelligent management system based on deep learning technology, including: a receiving module for receiving information on the purpose of borrowing, wherein the information on the purpose of borrowing includes: a borrowing object; a first determination module for determining the borrowing purpose characteristics corresponding to the borrowing purpose information using a classification model, and determining the risk assessment index corresponding to the borrowing object using a machine learning model; a second determination module for determining the first loan amount information according to the market price information of the means of production in the borrowing purpose information and the risk assessment index when the borrowing purpose characteristics are production purposes; a third determination module for determining the second loan amount information according to the construction cost, expected service life, expected future income and risk assessment index of the infrastructure when the borrowing purpose characteristics are infrastructure construction purposes; a fourth determination module for determining the loan contract information according to the loan amount information, and sending the loan contract information to the target device, wherein the loan amount information includes: the first loan amount information and the second loan amount information.

[0012] According to another aspect of the present application, a non-volatile storage medium is also provided, which includes a stored program, wherein when the program is running, the device where the storage medium is located is controlled to execute the above-mentioned intelligent management method based on deep learning technology.

[0013] According to another aspect of the present application, an electronic device is provided, including: a memory and a processor, the processor being used to run a program stored in the memory, wherein the above-mentioned intelligent management method based on deep learning technology is executed when the program is running.

[0014] According to another aspect of the present application, a computer program is also provided, wherein when the computer program is executed by a processor, the above intelligent management method based on deep learning technology is implemented.

[0015] According to another aspect of the present application, a computer program product is also provided, which includes a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned intelligent management method based on deep learning technology is implemented.

[0016] In the present application, a method of receiving loan purpose information is adopted, wherein the loan purpose information includes: the borrower; a classification model is used to determine the loan purpose characteristics corresponding to the loan purpose information, and a machine learning model is used to determine the risk assessment index corresponding to the borrower; when the loan purpose characteristics are production purposes, the first loan amount information is determined according to the market price information of the means of production in the loan purpose information and the risk assessment index; when the loan purpose characteristics are infrastructure construction purposes, the second loan amount information is determined according to the construction cost, expected service life, expected future income and risk assessment index of the infrastructure; according to the loan amount information, the loan contract information is determined, and the loan contract information is sent to the target device, wherein the loan amount information includes: the first loan amount information and the second loan amount information, so as to achieve the purpose of combining artificial intelligence algorithms to determine different loan amounts for different borrowing purposes of the borrowers of village banks, thereby achieving the technical effect of improving the promotion efficiency of the overall loan process, and then solving the technical problem that the promotion efficiency of the overall loan process is low due to the inability of related technologies to combine artificial intelligence algorithms to determine different loan amounts for different borrowing purposes of the borrowers of village banks. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 is a flow chart of an intelligent management method based on deep learning technology according to an embodiment of the present application;

[0019] Figure 2 is a structural diagram of an intelligent management device based on deep learning technology according to an embodiment of the present application;

[0020] Figure 3 This is a hardware structure block diagram of a computer terminal according to an intelligent management method based on deep learning technology in an embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution 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. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0022] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0023] According to an embodiment of the present application, a method embodiment of an intelligent management method based on deep learning technology is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0024] Figure 1 is a flow chart of an intelligent management method based on deep learning technology according to an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:

[0025] Step S101, receiving loan purpose information, wherein the loan purpose information includes: a borrower.

[0026] The receipt of loan purpose information can be achieved in a variety of ways, including but not limited to online application, telephone communication, face-to-face consultation, etc. When receiving information, a special electronic form or application can be designed to ensure the completeness and accuracy of the information. The loan purpose information should include the identity information of the borrower, a detailed description of the loan purpose, the project or field in which the funds are expected to be used, etc., to facilitate subsequent classification and risk assessment.

[0027] Step S102, using a classification model to determine the loan purpose characteristics corresponding to the loan purpose information, and using a machine learning model to determine the risk assessment index corresponding to the borrower.

[0028] Collect a large amount of historical data with loan purpose labels, including borrower descriptions, supporting documents, internal bank records, etc., to build a training data set. Use natural language processing technology to extract text features, such as keywords, phrases, word frequencies, etc., and then train classification models, such as support vector machines, random forests, etc., to identify and classify loan purpose information. On the other hand, a risk assessment model, such as logistic regression, gradient boosting tree or neural network, can be trained based on the basic information of the borrower (such as income, employment status), credit record, financial data and borrowing history to predict the risk level of the borrower. The training data of the model should contain borrowing cases with known risk assessment results.

[0029] In step S102, the machine learning model can be loaded into the memory. For example, the raw data of the machine learning model can be loaded from the non-volatile memory into the volatile memory so that the processor can run the machine learning model. The raw data of the machine learning model refers to unprocessed data, which usually includes parameters and structural data of the machine learning model. The structural data can be a calculation relationship based on the parameters, such as the forward propagation calculation relationship between the intermediate layers and the neurons. Specifically, the structural data can include code related to the structure of the machine learning model, such as code for performing related calculations between the intermediate layers and the neurons. In one embodiment, an area for loading the machine learning model can be divided in the memory, which can include a structural data storage area and a parameter storage area. The structural data storage area is used to store structure-related codes, and the parameters referenced by it can point to the address of specific parameters in the parameter storage area through a pointer. During the training process of the machine learning model, it may be necessary to update the parameters frequently, and the parameter values ​​in the parameter storage area can be updated.

[0030] Step S103, when the borrowing purpose characteristic is production purpose, determine the first loan amount information according to the market price information of the means of production and the risk assessment index in the borrowing purpose information.

[0031] After the classification model identifies the loan purpose as production purpose, the market price information of the means of production is obtained, and the prices of raw materials, equipment, etc. can be obtained from market data sources (such as commodity trading platforms) in real time or regularly. Combined with the risk level predicted by the risk assessment model, different loan limit limits are set for loans for production purposes, while taking into account the borrower's financial status and debt repayment ability.

[0032] Step S104, when the borrowing purpose is characterized as infrastructure construction, determine the second loan amount information according to the construction cost, expected service life, expected future income and risk assessment index of the infrastructure.

[0033] For infrastructure construction purposes, you can collect construction cost data for related projects, including land costs, design and construction costs, equipment purchase costs, etc. Combined with the estimated service life and expected future benefits of the infrastructure, as well as risk assessment indicators, use financial models (such as net present value NPV or internal rate of return IRR) to evaluate the economic feasibility of the project, and then determine the second loan amount information.

[0034] Step S105, determining the loan contract information according to the loan amount information, and sending the loan contract information to the target device, wherein the loan amount information includes: first loan amount information and second loan amount information.

[0035] Based on the first loan amount information and the second loan amount information, combined with the bank's loan policy and the borrower's specific conditions, loan contract information is generated, including loan amount, loan term, interest rate, repayment method, guarantee requirements, default clauses, etc.

[0036] By utilizing the electronic contract system of financial institutions, the loan contract information is sent to the borrower's target device (such as a smartphone or computer), and the contract signing process is completed through electronic signatures, thereby realizing the electronicization and automation of the loan approval process.

[0037] According to the above steps, the information on the purpose of borrowing is received, wherein the information on the purpose of borrowing includes: the borrower; the classification model is used to determine the borrowing purpose characteristics corresponding to the borrowing purpose information, and the machine learning model is used to determine the risk assessment indicators corresponding to the borrower; when the borrowing purpose characteristics are production purposes, the first loan amount information is determined according to the market price information of the means of production in the borrowing purpose information and the risk assessment indicators; when the borrowing purpose characteristics are infrastructure construction purposes, the second loan amount information is determined according to the construction cost, expected service life, expected future income and risk assessment indicators of the infrastructure; according to the loan amount information, the loan contract information is determined, and the loan contract information is sent to the target device, wherein the loan amount information includes: the first loan amount information and the second loan amount information, thereby achieving the purpose of combining artificial intelligence algorithms to determine different loan amounts for different borrowing purposes of the borrowers of village banks, thereby achieving the technical effect of improving the efficiency of the overall loan process.

[0038] The following Figure 1 The steps shown are exemplary and explanatory.

[0039] According to some optional embodiments of the present application, the classification model is obtained by training through the following method: collecting text data containing information on the purpose of borrowing from multiple data sources, wherein the data sources include: loan application documents, credit business records, customer questionnaire results, and communication records between customers and financial institution staff; the text data includes natural language text used to describe the characteristics of the purpose of borrowing; performing data cleaning on the text data, and performing word segmentation processing on the cleaned data to obtain word segmentation results, wherein data cleaning includes: removing preset special characters, correcting spelling errors, and unifying uppercase and lowercase formats; performing lexical analysis on the word segmentation results to extract basic vocabulary units in the text data to obtain target text data, wherein the lexical analysis includes: part-of-speech tagging, stem extraction, and word form restoration operations; using natural language processing technology to convert the target text data into a semantic feature vector; using the semantic feature vector and the target label corresponding to the semantic feature vector to train the classification model, and completing the training of the classification model when the loss function of the classification model meets the preset convergence condition, wherein the target label includes: production purpose, infrastructure construction purpose, educational purpose, and medical purpose.

[0040] In the above embodiment, text data containing information on the purpose of borrowing is first collected from various data sources, including loan application documents, credit business records, customer questionnaire results, and communication records between customers and financial institution staff. Ensure the diversity and comprehensiveness of the data so that the model can learn descriptions of the purpose of borrowing in different situations. Clean the collected raw text data, remove preset special characters (such as punctuation, numbers, HTML tags, etc.), correct spelling errors, unify the case format, and improve the quality and consistency of the text data. Use the natural language processing word segmentation tool to segment the cleaned text data, and divide the continuous text into independent vocabulary units to facilitate subsequent lexical analysis and feature extraction. For example, use the jieba word segmentation library to segment Chinese text.

[0041] Secondly, the word segmentation results are analyzed lexically, including part-of-speech tagging, stem extraction, and lemmatization. Part-of-speech tagging helps identify the role of words in sentences, while stem extraction and lemmatization help reduce vocabulary variants and improve feature consistency. The target text data is converted into semantic feature vectors. These vectors can capture the meaning of words in the text and their relationship with the context, providing richer information for subsequent model training.

[0042] Secondly, annotate the text data and assign a target label (such as production purpose, infrastructure construction purpose, education purpose, medical purpose) to each piece of loan purpose information to build a high-quality annotated data set. Select a machine learning or deep learning model suitable for text classification tasks, such as support vector machine, random forest, logistic regression, convolutional neural network, recurrent neural network or Transformer model. Initialize the model, such as setting the kernel function of SVM, the number of trees of RF, the learning rate and number of layers of the neural network and other parameters.

[0043] Finally, the classification model is trained using the labeled dataset. The semantic feature vector and its corresponding target label are input into the model, and the model is enabled to learn the ability to map semantic features to target labels by optimizing the loss function (such as cross entropy loss, mean square error, etc.). During the training process, the validation set is used to adjust the model's hyperparameters to prevent overfitting. The performance of the model is evaluated on the test set, using indicators such as accuracy, recall, and F1 score. If the model performance is poor, the model structure, feature extraction method, or data preprocessing process can be adjusted, and the model can be retrained until the model's loss function meets the preset convergence condition, that is, the model's training error no longer decreases significantly and reaches a stable state.

[0044] According to some other optional embodiments of the present application, determining the first loan amount information based on the market price information of the means of production in the borrowing purpose information and the risk assessment index can be achieved by the following method: obtaining multiple market price information of the means of production in the borrowing purpose information through different data sources, wherein the different data sources include: a market price database and an industry information platform; performing weighted summation on multiple market price information to obtain target market price information; calculating the total value of the means of production based on the quantity information of the means of production in the borrowing purpose information and the target market price information; when the risk assessment index is greater than a first preset threshold, multiplying the total value of the means of production by a first coefficient to obtain the loan amount information, wherein the first coefficient is less than 1; when the risk assessment index is less than the first preset threshold and greater than the second preset threshold, multiplying the total value of the means of production by a second coefficient to obtain the loan amount information, wherein the second preset threshold is less than the first preset threshold and the second coefficient is equal to 1; when the risk assessment index is less than the second preset threshold, multiplying the total value of the means of production by a third coefficient to obtain the first loan amount information, wherein the third coefficient is greater than 1.

[0045] For example, the latest market price information of production materials can be obtained from the "Global Commodity Price Index" database and the "China Industrial Products Procurement Platform". The price data of various raw materials (such as steel, cement, wood), equipment (such as machine tools, transportation vehicles) and services (such as logistics, insurance) are captured from the database regularly. Considering that the data of the "China Industrial Products Procurement Platform" is closer to the local market, a higher weight (such as 0.7) is given, while the "Global Commodity Price Index" database, although covering the world, may not fully reflect the local reality, so a lower weight (such as 0.3) is given. Based on the production demand in the loan purpose information, the specific quantity of production materials required (such as 50 tons of steel, 10 machine tools) is obtained. For each production material, the target market price is multiplied by the required quantity, and then the sum is calculated. Assume that the risk assessment index ranges from 0 to 100, where 100 indicates the highest risk. The first preset threshold = 75, the second preset threshold = 50; the first coefficient = 0.8, the second coefficient = 1.0, and the third coefficient = 1.2. When the risk assessment index is greater than 75, the total value of the means of production is multiplied by 0.8 to obtain the loan amount. When the risk assessment index is less than 50, the total value of the means of production is multiplied by 1.2 to obtain the first loan amount.

[0046] Furthermore, determining the second loan amount information based on the construction cost, expected service life, expected future income and risk assessment index of the infrastructure can be achieved by the following method: determining the target value of the infrastructure based on the construction cost, expected service life and expected future income of the infrastructure; when the risk assessment index is greater than the third preset threshold, multiplying the target value by the fourth coefficient to obtain the loan amount information, wherein the fourth coefficient is less than 1; when the risk assessment index is less than the third preset threshold and greater than the fourth preset threshold, multiplying the target value by the fifth coefficient to obtain the loan amount information, wherein the fourth preset threshold is less than the third preset threshold and the fifth coefficient is equal to 1; when the risk assessment index is less than the fourth preset threshold, multiplying the target value by the sixth coefficient to obtain the second loan amount information, wherein the sixth coefficient is greater than 1.

[0047] For example, collect information on the total cost of infrastructure construction from project planning documents, engineering contracts, supplier quotations, etc. This includes land acquisition costs, design costs, material and equipment costs, labor costs, management costs, etc. Evaluate the expected service life of the infrastructure based on engineering design standards, industry experience, and consideration of physical and environmental factors. For example, the service life of a bridge may be 50 years, while a water treatment facility may be 30 years. Forecast the total benefits of the infrastructure over its expected future service life based on market demand analysis, charging or user fee standards, historical operating data, etc. Consider possible revenue growth or decline, changes in operating costs, maintenance and upgrade costs, etc. Combine construction costs, expected service life, and future expected benefits, and use appropriate financial models (such as discounted cash flow models) to calculate the target value of the infrastructure. Target value = discounted value of future expected benefits - construction cost.

[0048] Based on the borrower's credit score, financial status, industry risk, market volatility and other indicators, a machine learning model is used to predict the borrower's comprehensive risk assessment level. For example, the risk assessment index ranges from 0 to 100, where 100 represents the highest risk. The third preset threshold is set to 70 and the fourth preset threshold is set to 50, and the two thresholds are used to distinguish high risk, medium risk and low risk. When the risk assessment index is greater than the third preset threshold (70), it is determined to be high risk, and the target value is multiplied by the fourth coefficient, where the fourth coefficient is less than 1 (for example, 0.8) to reduce the loan amount and control the risk. When the risk assessment index is less than the third preset threshold (70) and greater than the fourth preset threshold (50), it is determined to be medium risk, and the target value is multiplied by the fifth coefficient, where the fifth coefficient is equal to 1, that is, the loan amount is not adjusted and the target value is maintained. When the risk assessment index is less than the fourth preset threshold (50), it is determined to be low risk, and the target value is multiplied by the sixth coefficient, where the sixth coefficient is greater than 1 (for example, 1.5) to increase the loan amount and support the financing needs of low-risk infrastructure projects.

[0049] If the risk assessment index is greater than 70, the loan amount information = target value * fourth coefficient (e.g. 0.8), ensuring that the loan amount is controlled below the target value to protect the lending institution from high-risk projects. If the risk assessment index is between 50 and 70, the loan amount information = target value * fifth coefficient (e.g. 1), that is, the loan amount is equal to the target value, which is suitable for infrastructure projects with medium risks. If the risk assessment index is less than 50, the loan amount information = target value * sixth coefficient (e.g. 1.5), which means that the lending institution is willing to provide loans exceeding the target value in low-risk projects to support the smooth progress of the project and possible additional needs.

[0050] In summary, by collecting information on infrastructure construction costs, expected service life and expected future returns, combined with borrower risk assessment indicators, village banks and other financial institutions can use the above implementation methods to determine the loan amount for infrastructure construction projects, achieve a balance between risk and return, and support infrastructure construction and development in rural and township areas. Under the premise of ensuring project feasibility and borrower's debt repayment ability, by adjusting the loan amount coefficient, it is possible to flexibly respond to infrastructure projects with different risk levels, optimize loan allocation, and improve the efficiency of capital use.

[0051] In some optional embodiments of the present application, the machine learning model is obtained by training through the following method: obtaining basic information data, credit history data and transaction flow data of the target object, wherein the basic information data at least includes: name, gender, age, the credit history data at least includes: loan information, the loan information includes: loan type, loan amount, loan term, loan issuance time, repayment information, the repayment information at least includes: repaid amount, unpaid amount, number of overdue payments, overdue duration, transaction flow data includes: timestamp of each transaction, transaction amount, transaction direction, transaction category and counterparty information; integrating the basic information data, credit history data and transaction flow data, A first feature vector related to risk assessment is selected from the integrated data, wherein the first feature includes: age group, credit history indicator, transaction behavior characteristics, credit history indicators include: average overdue duration, total overdue times, loan default rate, and transaction behavior characteristics include: average monthly transaction times, ratio of expenditure to income; the first feature vector is normalized to obtain a second feature vector; the second feature vector is divided into a training set, a validation set, and a test set; the machine learning model is trained using the training set, the parameters of the model are adjusted through an optimization algorithm, the model is validated using the validation set during the training process, the performance of the trained model is evaluated using the test set, and a machine learning model that has completed the training is obtained when the preset evaluation indicators meet the preset conditions.

[0052] Obtain the target object's basic information data from the bank's customer information system, such as name, gender, and age. For example, target object A, name: Zhang San, gender: male, age: 32 years old. Obtain the target object's loan information from the bank's credit management system, including loan type, loan amount, loan term, loan issuance time, and repayment information. For example, the target object A's credit history data is as follows: loan type: personal consumption loan, housing mortgage loan; loan amount: 100,000 yuan, 300,000 yuan; loan term: 2 years, 10 years; loan issuance time: October 1, 2018, January 1, 2020; repayment information: repaid amount: 90,000 yuan, 270,000 yuan; unpaid amount: 10,000 yuan, 30,000 yuan; overdue number of times: 2 times, 1 time; overdue duration: 30 days, 10 days. Collect transaction flow data from the target object's bank account transaction records. For example, the transaction flow data of target object A includes: transaction records for the most recent year, timestamp of each transaction, transaction amount, transaction direction (income / expenditure), transaction category (such as salary, shopping, transfer), and counterparty information.

[0053] Select features related to risk assessment from the integrated data to construct the first feature vector, including: Age group: Zhang San is 32 years old and belongs to the group "26-35 years old". Credit history indicators: average overdue time = (30 days + 10 days) / 2 = 20 days; total overdue times = 2 + 1 = 3 times; loan default rate = (10,000 + 30,000) / (100,000 + 300,000) = 0.1 (10%). Transaction behavior characteristics: average number of transactions per month = total number of transactions / 12 months; ratio of expenditure to income = total expenditure / total income.

[0054] The processed second eigenvector is randomly divided into a training set (70%), a validation set (15%), and a test set (15%). Use the gradient boosting tree as the base model and train it using the training set data. During the training process, adjust the model parameters, such as the depth of the tree, the learning rate, etc., through the loss function (such as cross entropy loss or mean square error) and the optimization algorithm (such as gradient descent). Use the validation set data to verify the model during the training process, monitor the impact of the model parameter adjustment on the prediction effect, and prevent overfitting. Use the test set data to evaluate the performance of the trained model, and pay attention to the model's accuracy, recall rate, F1 value, AUC and other indicators.

[0055] In the above embodiment, the basic information, credit history and transaction flow data of the target object are first collected from the bank information system, and then a series of feature vectors related to risk assessment are constructed and normalized through data integration and feature engineering. Next, the feature vectors are divided into training set, validation set and test set, and a machine learning model is constructed using a gradient boosting tree, and the model parameters are optimized through the training and validation process. Finally, the model performance is evaluated using the test set to ensure that the model meets the conditions under the preset evaluation indicators, thereby obtaining a machine learning model that has completed training, which can be used to assess the risk level of the borrower and assist the bank in making credit decisions. Through this type of training, financial institutions such as village banks can more accurately identify credit risks and improve the efficiency and security of loan approval.

[0056] As some optional embodiments of the present application, after determining the loan contract information based on the loan amount information and sending the loan contract information to the target device, the following steps can also be performed: recording and storing the identification information of the borrower, and collecting the repayment information of the borrower corresponding to the identification information based on a preset time interval or a preset trigger event; converting the repayment information of the borrower corresponding to the identification information into a target code suitable for the machine learning model, and adjusting the hyperparameters of the machine learning model based on the target code.

[0057] For example, whenever a new customer applies for a loan, the system automatically generates a unique user ID and stores it in association with the customer's basic information (such as name, contact information, etc.). For example, the new customer Li Si is assigned user ID: 12345. After the customer successfully obtains the loan, the system will record and store the details of the loan, including the loan ID, associated with the user ID, the loan amount, the loan term, and the date the loan was issued. For example, user ID: 12345, loan ID: 54321, loan amount: 50,000 yuan, loan term: 24 months, loan issuance date: January 1, 2023.

[0058] Set the 1st of each month as the time point for collecting repayment information. For example, on February 1, 2023, the system automatically queries the repayment information of user ID: 12345, including the amount repaid, the amount not repaid, whether there is any overdue repayment record, etc. If the borrower is overdue for repayment, the system will immediately trigger an event to collect repayment information. For example, when the system detects that user ID: 12345 fails to repay on time on March 15, 2023, the system immediately updates its repayment information, recording the number of overdue days and the number of overdue times, etc.

[0059] The target code "0" (on-time repayment) can indicate a low credit risk, the target code "1" (overdue repayment) indicates a medium risk, and the target code "2" (multiple overdue payments) indicates a high credit risk. When the target code of the repayment information of the borrower is found to be "2", it means that the borrower has a high credit risk. At this point, the hyperparameters of the model can be adjusted, such as reducing the maximum depth to reduce model complexity, increasing the minimum number of sample splits to prevent overfitting, or increasing the number of trees to improve the prediction stability of the model, ensuring the accuracy and robustness of the model in predicting high-risk borrowers.

[0060] Specifically, assume that the hyperparameters of the machine learning model are initially set as follows: number of trees: 100, maximum depth: 15, minimum number of sample splits: 5. When the system detects that the target code of user ID: 12345 is "2", it indicates that there is a high credit risk in his repayment behavior. To this end, the system automatically adjusts the model hyperparameters, such as: number of trees: 150 (increase to improve the model's predictive ability), maximum depth: 10 (decrease to avoid overfitting), minimum number of sample splits: 10 (increase to improve the stability of the model).

[0061] In some optional embodiments of the present application, determining the loan contract information based on the loan amount information can be achieved by the following method: when the loan amount information meets the preset conditions, sending the loan approval form including the loan amount information to the first device, wherein the preset conditions include at least one of the following: the loan amount exceeds the preset amount threshold, the number of overdue repayments by the borrower exceeds a preset number or the overdue repayment period exceeds a preset period, the means of production in the loan purpose characteristics are changed from preset low-risk means of production to preset high-risk means of production, and the price fluctuation of agricultural products in the target area exceeds a preset range; receiving the verification result of the loan approval form sent by the first device, and generating the loan contract information based on the verification result.

[0062] Assuming that the execution object is a village bank focusing on agricultural loans, the machine learning model is used to evaluate loan applications, and a loan approval form is generated based on the evaluation results. When specific preset conditions are met, the approval form is sent to the bank's internal review device (first device) for further verification. After the review is completed, the loan contract information is generated based on the verification results.

[0063] Borrower: Li Hua, applying for a loan to purchase fertilizers and irrigation equipment. According to the model assessment, Li Hua's loan amount is 300,000 yuan, which is lower than the high-risk loan amount threshold we set (500,000 yuan), and his credit record is good, with no overdue repayments. The means of production in the loan purpose characteristics are fertilizers and irrigation equipment, both of which belong to our preset low-risk means of production list. The price fluctuations of agricultural products in the target area are within the normal range and do not exceed the preset range.

[0064] The loan amount exceeds the preset limit threshold: Not satisfied, Li Hua's loan amount is 300,000 yuan, which is lower than the preset limit threshold of 500,000 yuan. The number of overdue repayments of the borrower exceeds the preset number or the overdue repayment duration exceeds the preset duration: Not satisfied, Li Hua has no overdue repayment record. The means of production in the loan purpose characteristics are changed from the preset low-risk means of production to the preset high-risk means of production: Not satisfied, the means of production in the loan purpose all belong to the preset low-risk category. The price fluctuation of agricultural products in the target area exceeds the preset range: Not satisfied, the price fluctuation is within the expected range.

[0065] Since Li Hua's loan amount information and other conditions did not trigger the preset conditions, the loan approval form does not need to be sent to the first device for additional verification. However, suppose in another case, Zhang Lin applied for a loan amount of 600,000 yuan, which exceeded the preset amount threshold. According to our preset conditions, the system will automatically send the loan approval form to the first device for strict review.

[0066] After the first device receives the approval form, a dedicated credit approval staff will conduct a detailed check, including the rationality of the loan amount, the repayment ability assessment of the borrower, and the compliance check of the purpose of the loan. For Zhang Lin's loan approval form, the first device's check result shows that the loan amount is too high, and it is recommended to re-evaluate the repayment ability of the borrower and the risk of the purpose of the loan, and consider reducing the loan amount to 450,000 yuan. According to the check results, the loan contract information can be automatically generated, including: Loan amount: 450,000 yuan Repayment period: 36 months Annual interest rate: 6% Monthly repayment amount: calculated based on the loan amount, repayment period and annual interest rate Contract terms: The purpose of the loan, repayment method, breach of contract liability, etc. are listed in detail.

[0067] In the above embodiment, a mechanism is designed to automatically detect whether the loan approval form meets the preset conditions. If so, the loan approval form will be sent to the first device within the bank for strict review to ensure the security and compliance of the loan issuance. After the review is completed, the loan contract information is generated based on the verification results to ensure that the contract terms are consistent with the approval results, providing a clear legal basis for both parties. Through this process, rural banks can effectively manage loan risks while improving approval efficiency and customer satisfaction.

[0068] Figure 2 is a structural diagram of an intelligent management system based on deep learning technology according to an embodiment of the present application, such as Figure 2 As shown, the system includes:

[0069] The receiving module 21 is used to receive the borrowing purpose information, wherein the borrowing purpose information includes: the borrowing object.

[0070] The first determination module 22 is used to determine the loan purpose characteristics corresponding to the loan purpose information using a classification model, and to determine the risk assessment index corresponding to the loan object using a machine learning model.

[0071] The second determination module 23 is used to determine the first loan amount information according to the market price information of the means of production and the risk assessment index in the loan purpose information when the loan purpose characteristic is production purpose.

[0072] The third determination module 24 is used to determine the second loan amount information according to the construction cost, expected service life, future expected income and risk assessment index of the infrastructure when the loan purpose characteristic is infrastructure construction.

[0073] The fourth determination module 25 is used to determine the loan contract information according to the loan amount information, and send the loan contract information to the target device, wherein the loan amount information includes: first loan amount information and second loan amount information.

[0074] Optionally, the classification model is obtained by training in the following method: collecting text data containing information on the purpose of borrowing from multiple data sources, wherein the data sources include: loan application documents, credit business records, customer questionnaire results, and communication records between customers and financial institution staff; the text data includes natural language text used to describe the characteristics of the purpose of borrowing; performing data cleaning on the text data, and performing word segmentation processing on the cleaned data to obtain word segmentation results, wherein data cleaning includes: removing preset special characters, correcting spelling errors, and unifying uppercase and lowercase formats; performing lexical analysis on the word segmentation results to extract basic vocabulary units in the text data to obtain target text data, wherein lexical analysis includes: part-of-speech tagging, stem extraction, and word form restoration operations; using natural language processing technology to convert the target text data into a semantic feature vector; using the semantic feature vector and the target label corresponding to the semantic feature vector to train the classification model, and completing the training of the classification model when the loss function of the classification model meets the preset convergence condition, wherein the target label includes: production purpose, infrastructure construction purpose, educational purpose, and medical purpose.

[0075] Optionally, the first loan amount information is determined based on the market price information of the means of production in the loan purpose information and the risk assessment index, including: obtaining multiple market price information of the means of production in the loan purpose information through different data sources, wherein the different data sources include: a market price database and an industry information platform; performing weighted summation on the multiple market price information to obtain the target market price information; calculating the total value of the means of production based on the quantity information of the means of production in the loan purpose information and the target market price information; when the risk assessment index is greater than a first preset threshold, multiplying the total value of the means of production by a first coefficient to obtain the loan amount information, wherein the first coefficient is less than 1; when the risk assessment index is less than the first preset threshold and greater than the second preset threshold, multiplying the total value of the means of production by a second coefficient to obtain the loan amount information, wherein the second preset threshold is less than the first preset threshold and the second coefficient is equal to 1; when the risk assessment index is less than the first preset threshold and greater than the second preset threshold, When the risk assessment index is greater than the third preset threshold, the target value is multiplied by the fourth coefficient to obtain the loan amount information, wherein the fourth coefficient is less than 1; when the risk assessment index is less than the third preset threshold and greater than the fourth preset threshold, the target value is multiplied by the fifth coefficient to obtain the loan amount information, wherein the fourth preset threshold is less than the third preset threshold and the fifth coefficient is equal to 1; when the risk assessment index is less than the fourth preset threshold, the target value is multiplied by the sixth coefficient to obtain the second loan amount information, wherein the sixth coefficient is greater than 1.

[0076] Optionally, the machine learning model is trained by the following method: obtaining basic information data, credit history data and transaction flow data of the target object, wherein the basic information data at least includes: name, gender, age, the credit history data at least includes: loan information, the loan information includes: loan type, loan amount, loan term, loan issuance time, repayment information, the repayment information at least includes: repaid amount, unpaid amount, number of overdue payments, overdue duration, transaction flow data includes: timestamp of each transaction, transaction amount, transaction direction, transaction category and counterparty information; integrating the basic information data, credit history data and transaction flow data, A first feature vector related to risk assessment is selected from the integrated data, wherein the first feature includes: age group, credit history indicator, transaction behavior characteristics, credit history indicators include: average overdue duration, total overdue times, loan default rate, and transaction behavior characteristics include: average monthly transaction times, ratio of expenditure to income; the first feature vector is normalized to obtain a second feature vector; the second feature vector is divided into a training set, a validation set, and a test set; the machine learning model is trained using the training set, the parameters of the model are adjusted through an optimization algorithm, the model is validated using the validation set during the training process, the performance of the trained model is evaluated using the test set, and a machine learning model that has completed the training is obtained when the preset evaluation indicators meet the preset conditions.

[0077] Optionally, after determining the loan contract information based on the loan amount information and sending the loan contract information to the target device, the method also includes: recording and storing the identification information of the borrower, and collecting the repayment information of the borrower corresponding to the identification information based on a preset time interval or a preset trigger event; converting the repayment information of the borrower corresponding to the identification information into a target code suitable for the machine learning model, and adjusting the hyperparameters of the machine learning model based on the target code.

[0078] Optionally, the loan contract information is determined based on the loan amount information, including: when the loan amount information meets preset conditions, the loan approval form including the loan amount information is sent to the first device, wherein the preset conditions include at least one of the following: the loan amount exceeds a preset amount threshold, the number of overdue repayments by the borrower exceeds a preset number or the overdue repayment period exceeds a preset period, the means of production in the loan purpose characteristics are changed from preset low-risk means of production to preset high-risk means of production, and the price fluctuation of agricultural products in the target area exceeds a preset range; receiving the verification result of the loan approval form sent by the first device, and generating the loan contract information based on the verification result.

[0079] It should be noted that the above Figure 2Each module in the can be a program module (for example, a set of program instructions that implement a specific function) or a hardware module. For the latter, it can be expressed in the following forms, but is not limited to this: the expression form of each of the above modules is a processor, or the functions of each of the above modules are implemented by a processor.

[0080] It should be noted that Figure 2 The preferred implementation of the illustrated embodiment can be found in Figure 1 The relevant description of the illustrated embodiment will not be repeated here.

[0081] Figure 3 The hardware structure block diagram of a computer terminal for implementing an intelligent management method based on deep learning technology is shown. Figure 3 As shown, the computer terminal 30 may include one or more (302a, 302b, ..., 302n are used to illustrate) processors 302 (the processor 302 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 304 for storing data, and a transmission module 306 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 3 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 3 More or fewer components as shown, or with Figure 3 Different configurations are shown.

[0082] It should be noted that the one or more processors 302 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 30. As involved in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0083] The memory 304 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the intelligent management method based on deep learning technology in the embodiment of the present application. The processor 302 executes various functional applications and data processing by running the software programs and modules stored in the memory 304, that is, realizing the above-mentioned intelligent management method based on deep learning technology. The memory 304 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 304 may further include a memory remotely arranged relative to the processor 302, and these remote memories may be connected to the computer terminal 30 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0084] The transmission module 306 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 30. In one example, the transmission module 306 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission module 306 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0085] The display may be, for example, a touch screen liquid crystal display (LCD), which may enable the target object to interact with the target object interface of the computer terminal 30 .

[0086] It should be noted that, in some optional embodiments, the above Figure 3 The computer terminal shown may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of hardware elements and software elements. It should be noted that Figure 3 This is merely one example of a particular embodiment and is intended to illustrate the types of components that may be present in the computer terminal described above.

[0087] It should be noted that Figure 3 The computer terminal shown is used to execute Figure 1 The intelligent management method based on deep learning technology shown in the figure, therefore the relevant explanations and instructions in the execution method of the above commands are also applicable to the electronic device and will not be repeated here.

[0088] An embodiment of the present application also provides a non-volatile storage medium, which includes a stored program, wherein when the program is running, the device where the storage medium is located is controlled to execute the above-mentioned intelligent management method based on deep learning technology.

[0089] A program for a non-volatile storage medium to perform the following functions: receiving information on the purpose of borrowing, wherein the information on the purpose of borrowing includes: a borrowing object; using a classification model to determine a borrowing purpose feature corresponding to the borrowing purpose information, and using a machine learning model to determine a risk assessment index corresponding to the borrowing object; when the borrowing purpose feature is production purpose, determining first loan amount information based on market price information of means of production in the borrowing purpose information and risk assessment indexes; when the borrowing purpose feature is infrastructure construction purpose, determining second loan amount information based on the construction cost, estimated service life, expected future income and risk assessment index of the infrastructure; determining loan contract information based on the loan amount information, and sending the loan contract information to a target device, wherein the loan amount information includes: first loan amount information and second loan amount information.

[0090] An embodiment of the present application also provides an electronic device, including: a memory and a processor, the processor being used to run a program stored in the memory, wherein the above-mentioned intelligent management method based on deep learning technology is executed when the program is running.

[0091] The processor is used to run a program that performs the following functions: receiving information on the purpose of borrowing, wherein the information on the purpose of borrowing includes: a borrowing object; using a classification model to determine a borrowing purpose feature corresponding to the borrowing purpose information, and using a machine learning model to determine a risk assessment index corresponding to the borrowing object; when the borrowing purpose feature is production purpose, determining the first loan amount information based on the market price information of the means of production and the risk assessment index in the borrowing purpose information; when the borrowing purpose feature is infrastructure construction purpose, determining the second loan amount information based on the construction cost, expected service life, expected future income and risk assessment index of the infrastructure; determining the loan contract information based on the loan amount information, and sending the loan contract information to the target device, wherein the loan amount information includes: the first loan amount information and the second loan amount information.

[0092] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0093] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0094] In the above-mentioned embodiments of the present application, the collected information is information and data authorized by the target object or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary protection measures, do not violate public order and good morals, and provide corresponding operation entrances for the target object to choose to authorize or refuse.

[0095] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be 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 interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0096] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0097] In addition, each functional unit in each embodiment of 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. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0098] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it 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 relevant technology or all or 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, including a number of instructions to enable a computer device (which can be a personal computer, server or 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: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.

[0099] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. An intelligent management method based on deep learning technology, characterized in that: include: receiving loan purpose information, wherein the loan purpose information includes: a borrower; Using a classification model to determine the loan purpose characteristics corresponding to the loan purpose information, and using a machine learning model to determine the risk assessment index corresponding to the loan object; In the case where the loan purpose characteristic is production purpose, determining the first loan amount information according to the market price information of the means of production in the loan purpose information and the risk assessment index; In the case where the borrowing purpose is characterized as infrastructure construction, the second loan amount information is determined based on the construction cost, expected service life, expected future income and risk assessment index of the infrastructure; Determine loan contract information based on the loan amount information, and send the loan contract information to a target device, wherein the loan amount information includes: the first loan amount information and the second loan amount information.

2. The method according to claim 1, characterized in that The classification model is obtained by training through the following method: Collecting text data containing loan purpose information from multiple data sources, wherein the data sources include: loan application documents, credit business records, customer questionnaire results, and communication records between customers and financial institution staff; the text data includes natural language text for describing loan purpose characteristics; Performing data cleaning on the text data, and performing word segmentation processing on the cleaned data to obtain a word segmentation result, wherein the data cleaning includes: removing preset special characters, correcting spelling errors, and unifying uppercase and lowercase formats; Performing lexical analysis on the word segmentation results to extract basic vocabulary units in the text data to obtain target text data, wherein the lexical analysis includes: part-of-speech tagging, stem extraction and lemma restoration operations; The target text data is converted into a semantic feature vector by using natural language processing technology; the classification model is trained using the semantic feature vector and the target label corresponding to the semantic feature vector, and the training of the classification model is completed when the loss function of the classification model meets the preset convergence condition, wherein the target label includes: production use, infrastructure construction use, educational use, and medical use.

3. The method according to claim 1, characterized in that Determining first loan amount information according to the market price information of the means of production in the loan purpose information and the risk assessment index includes: Acquire multiple market price information of the production materials in the loan purpose information through different data sources, wherein the different data sources include: a market price database and an industry information platform; Performing weighted summation on the plurality of market price information to obtain target market price information; calculating the total value of the production materials according to the quantity information of the production materials in the loan purpose information and the target market price information; When the risk assessment index is greater than a first preset threshold, the total value of the means of production is multiplied by a first coefficient to obtain the loan amount information, wherein the first coefficient is less than 1; When the risk assessment index is less than the first preset threshold and greater than a second preset threshold, the total value of the means of production is multiplied by a second coefficient to obtain the loan amount information, wherein the second preset threshold is less than the first preset threshold and the second coefficient is equal to 1; When the risk assessment index is less than the second preset threshold, the total value of the means of production is multiplied by a third coefficient to obtain the first loan amount information, wherein the third coefficient is greater than 1; The second loan amount information is determined based on the construction cost, expected service life, expected future income and risk assessment indicators of the infrastructure, including: Determine the target value of the infrastructure based on its construction cost, estimated useful life, and expected future benefits; When the risk assessment index is greater than a third preset threshold, multiplying the target value by a fourth coefficient to obtain the loan amount information, wherein the fourth coefficient is less than 1; When the risk assessment index is less than the third preset threshold and greater than the fourth preset threshold, multiplying the target value by a fifth coefficient to obtain the loan amount information, wherein the fourth preset threshold is less than the third preset threshold and the fifth coefficient is equal to 1; When the risk assessment index is less than the fourth preset threshold, the target value is multiplied by a sixth coefficient to obtain the second loan amount information, wherein the sixth coefficient is greater than 1.

4. The method according to claim 1, characterized in that: The machine learning model is obtained by training using the following method: Obtaining basic information data, credit history data and transaction flow data of the target object, wherein the basic information data at least includes: name, gender, age, the credit history data at least includes: loan information, the loan information includes: loan type, loan amount, loan term, loan issuance time, repayment information, the repayment information at least includes: repaid amount, unpaid amount, overdue times, overdue duration, the transaction flow data includes: timestamp of each transaction, transaction amount, transaction direction, transaction category and counterparty information; Integrate the basic information data, the credit history data and the transaction flow data, and select a first feature vector related to risk assessment from the integrated data, wherein the first feature includes: age group, credit history index, and transaction behavior characteristics, the credit history index includes: average overdue duration, total overdue times, and loan default rate, and the transaction behavior characteristics include: average monthly transaction times and ratio of expenditure to income; Normalizing the first eigenvector to obtain a second eigenvector; The second feature vector is divided into a training set, a validation set and a test set; the training set is used to train the machine learning model, the parameters of the model are adjusted by an optimization algorithm, the validation set is used to verify the model during the training process, and the test set is used to evaluate the performance of the trained model. When the preset evaluation indicators meet the preset conditions, a machine learning model that has completed the training is obtained.

5. The method according to claim 1, characterized in that After determining the loan contract information according to the loan amount information and sending the loan contract information to the target device, the method further includes: Record and store the identification information of the borrower, and collect the repayment information of the borrower corresponding to the identification information based on a preset time interval or a preset trigger event; The repayment information of the borrower corresponding to the identification information is converted into a target code applicable to the machine learning model, and the hyperparameters of the machine learning model are adjusted based on the target code.

6. The method according to claim 1, characterized in that Determine the loan contract information based on the loan amount information, including: In the case where the loan amount information satisfies a preset condition, a loan approval form including the loan amount information is sent to the first device, wherein the preset condition includes at least one of the following: the loan amount exceeds a preset amount threshold, the number of overdue repayments of the borrower exceeds a preset number or the overdue repayment duration exceeds a preset duration, the means of production in the loan use characteristics are changed from preset low-risk means of production to preset high-risk means of production, and the price fluctuation of agricultural products in the target area exceeds a preset range; Receive the verification result of the loan approval form sent by the first device, and generate the loan contract information according to the verification result.

7. An intelligent management system based on deep learning technology, characterized in that: include: The receiving module is used to receive the loan purpose information, wherein the loan purpose information includes: the borrower; A first determination module is used to determine the loan purpose characteristics corresponding to the loan purpose information using a classification model, and to determine the risk assessment index corresponding to the loan object using a machine learning model; A second determination module is used to determine the first loan amount information according to the market price information of the means of production in the loan purpose information and the risk assessment index when the loan purpose characteristic is production purpose; A third determination module is used to determine the second loan amount information according to the construction cost, expected service life, expected future income and the risk assessment index of the infrastructure when the loan purpose characteristic is infrastructure construction; The fourth determination module is used to determine the loan contract information according to the loan amount information, and send the loan contract information to the target device, wherein the loan amount information includes: the first loan amount information and the second loan amount information.

8. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the intelligent management method based on deep learning technology as described in any one of claims 1 to 6.

9. An electronic device, characterized in that: include: A memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the program, when running, executes the intelligent management method based on deep learning technology as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the intelligent management method based on deep learning technology described in any one of claims 1 to 6 is implemented.

Images