Automobile financial service method and system based on large language model

Abstract

The invention relates to the technical field of automobile finance, in particular to an automobile finance service method and system based on a big language model, and the method comprises the following steps: based on the transaction application information of a customer, analyzing the loan and lease application information of the customer through the big language model, and recognizing and recording the basic information and financial data of the customer; a customer information dataset is generated. According to the invention, batch data is automatically processed by using a large language model, the examination and approval time is reduced, the processing efficiency is improved, key financial data is extracted, the comprehensiveness and accuracy of customer credit evaluation are improved by combining historical repayment records of customers and the use frequency of credit cards, and through real-time monitoring of transaction behaviors and data-driven market analysis, the customer credit evaluation efficiency is improved. According to the method, abnormal transactions and fraudulent behaviors are quickly identified, financial risks are prevented in advance, and market adaptability and customer satisfaction of financial services are improved by predicting market trends and seasonal changes, adjusting loan, lease and insurance parameters and matching market and customer demands.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile finance, and in particular to an automobile financial service method and system based on a large language model. Background Art

[0002] The field of automotive financial technology supports consumers in purchasing and leasing cars by utilizing a variety of financial tools and services, combining traditional financial services with modern technologies to simplify transaction processes, improve capital utilization efficiency, and reduce related economic risks. It includes vehicle loans, leasing, insurance, and credit assessment services. It optimizes credit approval processes and risk control through data analysis and management technology, and combines artificial intelligence and big data to provide personalized and flexible financial products.

[0003] Among them, the automotive financial service method aims to provide and optimize the access to and management of financial products, including procedures and technologies for providing car buyers with loans, leasing plans, insurance and other financial services. It simplifies the financing process, speeds up approval time, reduces risks, and combines digital technology to improve the user's service experience, including through an automated credit assessment system to quickly process loan applications, and use online platforms to enable users to complete application, payment and management of multiple links by themselves. It combines data analysis and behavior prediction technology to accurately identify customer needs and improve services, improve business efficiency and provide customers with convenient and personalized financial services.

[0004] Traditional automotive financial service methods rely on traditional financial models and manual approval processes when processing customer data and evaluating credit, resulting in extended approval time and poor data utilization efficiency. When processing batch data, it is difficult to respond quickly to market changes and customer needs, affecting the timeliness of services and customer experience. In a rapidly changing financial environment, complex fraud cannot be identified and responded to in a timely manner, increasing operational risks. There is a lack of real-time data support in market analysis, and the response to market trends is not sensitive and accurate enough, making it difficult for financial service providers to accurately adjust product design and service strategies, resulting in a disadvantageous position in the highly competitive market. The delayed processing of market data leads to deviations between financial products and actual customer needs, affecting the business development and profit maximization of financial institutions. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an automobile financial service method and system based on a large language model.

[0006] In order to achieve the above object, the present invention adopts the following technical solution, a method for automobile financial service based on a large language model, comprising the following steps:

[0007] S1: Based on customer transaction application information, analyze the customer's loan and lease application information through a large language model, identify and record the customer's basic information and financial data, and generate a customer information dataset;

[0008] S2: Based on the customer information data set, by analyzing the customer's historical loan repayment record, credit card usage frequency, default record, and comparing the customer's financial stability with the market average, the customer's credit rating is evaluated and a customer credit rating score is generated;

[0009] S3: using the customer credit rating score, by analyzing the customer's deposit and income data, evaluating the customer's financial liquidity and asset status, calculating the customer's monthly repayment ability, and generating a repayment ability analysis result;

[0010] S4: Based on the repayment capacity analysis results, by real-time monitoring of customer transaction behavior information, identifying abnormal transaction records and fraudulent behavior, and generating abnormal transaction monitoring results;

[0011] S5: Based on the abnormal transaction monitoring results, collect automobile finance market data in real time, analyze and predict the trend and seasonal change pattern of the automobile finance market, and generate market demand analysis data;

[0012] S6: Based on the market demand analysis data and in combination with customer credit information, loan parameters, leasing parameters, and insurance parameters are adjusted to match market changes and customer needs, and service parameter adjustment results are generated.

[0013] As a further solution of the present invention, the customer information data set includes customer occupation information, customer age information, customer deposit and income data, the customer credit rating score includes credit score range, financial stability rating information, and market benchmark data, the repayment ability analysis results include monthly disposable income, estimated monthly repayment amount, and financial liquidity evaluation results, the abnormal transaction monitoring results include fraud type identification results, abnormal transaction patterns, and transaction amount ranges, the market demand analysis data includes loan demand growth rate data, leasing market activity, and insurance product preference information, and the service parameter adjustment results include adjusted loan interest rate range, leasing conditions adjustment records, and auto insurance product adjustment results.

[0014] As a further solution of the present invention, based on the customer transaction application information, the customer's loan and lease application information is analyzed by a large language model, and the customer's basic information and financial data are identified and recorded. The steps of generating a customer information data set are specifically as follows:

[0015] S101: Based on the customer transaction application information, identify and record the customer's personal information, including name, contact information, and address, and generate a customer basic information record;

[0016] S102: Based on the customer basic information record, analyzing the financial certification documents provided by the customer, including salary slips and bank deposit certificates, identifying the customer's monthly income and total deposit information, and generating customer financial information;

[0017] S103: Based on the customer's financial information, analyze the type of loan, loan amount and repayment period applied for by the customer, and generate a customer information data set in combination with the customer's basic information.

[0018] As a further solution of the present invention, based on the customer information data set, by analyzing the customer's historical loan repayment record, credit card usage frequency, default record, and comparing the customer's financial stability with the market average, the customer's credit rating is evaluated, and the steps of generating the customer's credit rating score are specifically as follows:

[0019] S201: Based on the customer information data set, collect and analyze the customer's loan and repayment information, including the loan amount and repayment time, record the number of overdue payments, and generate a loan repayment record;

[0020] S202: Based on the loan repayment record, calculate the usage frequency and overdue situation of the customer's credit card, analyze and compare the difference with the market average, and generate a credit behavior analysis result;

[0021] S203: Based on the credit behavior analysis result, taking into account the financial stability of the customer, a logistic regression algorithm is used to evaluate and calculate the customer's credit rating and generate a customer credit rating score.

[0022] As a further solution of the present invention, the logistic regression algorithm is according to the formula:

[0023]

[0024] Calculate the customer's credit default probability, where p is the prediction result, e is the base of the natural logarithm, β0 is the intercept term, β1 is the weight coefficient of annual income, β2 is the weight coefficient of credit card usage frequency, β3 is the weight coefficient of the financial responsibility index, β4 is the weight coefficient of the historical loan repayment rate, x1 is the annual income, x2 is the credit card usage frequency, x3 is the financial responsibility index, and x4 is the historical loan repayment rate.

[0025] As a further solution of the present invention, the steps of using the customer credit rating score to analyze the customer's deposit and income data, evaluate the customer's financial liquidity and asset status, and calculate the customer's monthly repayment ability to generate the repayment ability analysis result are specifically as follows:

[0026] S301: Using the customer credit rating score, by analyzing the customer's deposit certificate and income certificate documents, recording monthly income and deposit balance, and generating the customer's monthly financial record;

[0027] S302: Based on the monthly financial records of the customer, taking into account family status, age, and occupation, assess the customer's monthly fixed expenses and disposable income, calculate the asset-liability ratio, and generate financial liquidity assessment data;

[0028] S303: Based on the financial liquidity assessment data, the monthly repayment capabilities of multiple customers are calculated, and the credit rating scores of the customers are considered to generate a repayment capability analysis result.

[0029] As a further solution of the present invention, based on the repayment capacity analysis result, by real-time monitoring of customer transaction behavior information, identifying abnormal transaction records and fraudulent behavior, the steps of generating abnormal transaction monitoring results are specifically:

[0030] S401: Based on the repayment capacity analysis result, collect and analyze the customer's transaction behavior information in real time, including transaction date, amount and location information, and generate real-time transaction records;

[0031] S402: According to the real-time transaction record, by comparing and analyzing with the regular transaction pattern, identifying abnormal transaction behaviors, including abnormal large-amount transactions and transactions at abnormal locations, and generating a transaction abnormality identification record;

[0032] S403: Based on the transaction anomaly identification record, combined with the customer's device information and geographic location data, identify fraudulent behavior and generate abnormal transaction monitoring results.

[0033] As a further solution of the present invention, according to the abnormal transaction monitoring results, the steps of collecting automobile financial market data in real time, analyzing and predicting the trend and seasonal change pattern of the automobile financial market, and generating market demand analysis data are specifically as follows:

[0034] S501: According to the abnormal transaction monitoring results, the number of loan applications, leasing activity data and insurance purchase records in the automobile finance market are collected and analyzed in real time to generate market activity data;

[0035] S502: Based on the market activity data, calculate the monthly loan, lease and insurance demand growth rate, analyze and record seasonal changes, and generate demand change calculation data;

[0036] S503: According to the demand change calculation data, identify key factors and seasonal change patterns that affect the automobile financial market, and generate market demand analysis data.

[0037] As a further solution of the present invention, based on the market demand analysis data and in combination with customer credit information, the steps of adjusting loan parameters, leasing parameters, and insurance parameters to match market changes and customer needs and generating service parameter adjustment results are specifically as follows:

[0038] S601: Based on the market demand analysis data, evaluate the market competitiveness of the current market loan interest rate level and loan amount, identify and record adjustment needs, and generate an adjustment demand assessment record;

[0039] S602: Based on the adjustment demand assessment record, by analyzing the data of the leasing market, assessing the market adaptability of the current lease term and price, calculating the adjustment demand of the leasing parameters, and generating leasing parameter analysis information;

[0040] S603: Based on the lease parameter analysis information, adjust the coverage and compensation ratio of the insurance product, optimize the insurance price to match the customer's credit rating and market demand, and generate a service parameter adjustment result.

[0041] An automobile financial service system based on a large language model, the automobile financial service system based on a large language model is used to execute the above-mentioned automobile financial service method based on a large language model, and the system includes:

[0042] The customer information analysis module analyzes the data based on the customer transaction application information, identifies and records occupation, age, deposit and income data, and generates a basic customer profile;

[0043] The customer credit assessment module analyzes the customer's historical loan repayment record and credit card usage frequency based on the customer basic file, conducts a comparative analysis of financial stability, and generates a credit score assessment result;

[0044] The financial status analysis module analyzes the deposit and income data based on the credit score evaluation result, calculates the customer's monthly repayment ability, and generates repayment ability evaluation information;

[0045] The abnormal transaction monitoring module monitors customer transaction behaviors in real time based on the repayment ability assessment information, including login information and device fingerprints, identifies abnormal transactions and fraudulent behaviors, and generates abnormal activity identification records;

[0046] The market matching adjustment module collects automobile financial market data based on the abnormal activity identification records, analyzes market trends and demand changes, adjusts loan, lease and insurance parameters in combination with customer credit information, and generates service parameter adjustment results.

[0047] Compared with the prior art, the advantages and positive effects of the present invention are:

[0048] In the present invention, a large language model is used to automatically process a large amount of data, reduce approval time and improve processing efficiency, extract key financial data, and combine the customer's historical repayment record and credit card usage frequency to improve the comprehensiveness and accuracy of customer credit assessment. Through real-time monitoring of transaction behavior and data-driven market analysis, abnormal transactions and fraudulent behavior can be quickly identified to prevent financial risks in advance. By predicting market trends and seasonal changes, loan, leasing and insurance parameters are adjusted to match market and customer needs, thereby improving the market adaptability and customer satisfaction of financial services. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0050] Figure 2 This is a detailed flow chart of S1 of the present invention;

[0051] Figure 3 This is a detailed flow chart of S2 of the present invention;

[0052] Figure 4 This is a detailed flow chart of S3 of the present invention;

[0053] Figure 5 This is a detailed flow chart of S4 of the present invention;

[0054] Figure 6 This is a detailed flow chart of S5 of the present invention;

[0055] Figure 7 This is a detailed flow chart of S6 of the present invention;

[0056] Figure 8 It is a system flow chart of the present invention. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0058] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0059] See also Figure 1 The present invention provides a technical solution, a method for automobile financial service based on a large language model, comprising the following steps:

[0060] S1: Based on customer transaction application information, analyze the customer's loan and lease application information through a large language model, identify and record the customer's basic information and financial data, and generate a customer information dataset;

[0061] S2: Based on the customer information data set, by analyzing the customer's historical loan repayment record, credit card usage frequency, default record, and comparing the customer's financial stability with the market average, the customer's credit rating is evaluated and a customer credit rating score is generated;

[0062] S3: Use the customer credit rating score to analyze the customer's deposit and income data, assess the customer's financial liquidity and asset status, calculate the customer's monthly repayment ability, and generate repayment ability analysis results;

[0063] S4: Based on the repayment capacity analysis results, by real-time monitoring of customer transaction behavior information, abnormal transaction records and fraudulent behavior are identified, and abnormal transaction monitoring results are generated;

[0064] S5: Based on the abnormal transaction monitoring results, collect auto finance market data in real time, analyze and predict the trend and seasonal change pattern of the auto finance market, and generate market demand analysis data;

[0065] S6: Based on market demand analysis data and combined with customer credit information, adjust loan parameters, leasing parameters, and insurance parameters to match market changes and customer needs, and generate service parameter adjustment results.

[0066] The customer information data set includes customer occupation information, customer age information, customer deposit and income data. The customer credit rating score includes credit score range, financial stability rating information, and market benchmark data. The repayment ability analysis results include monthly disposable income, estimated monthly repayment amount, and financial liquidity evaluation results. The abnormal transaction monitoring results include fraud type identification results, abnormal transaction patterns, and transaction amount ranges. The market demand analysis data includes loan demand growth rate data, leasing market activity, and insurance product preference information. The service parameter adjustment results include the adjusted loan interest rate range, leasing conditions adjustment records, and auto insurance product adjustment results.

[0067] See also Figure 2 Based on the customer transaction application information, the customer's loan and lease application information is analyzed through a large language model, and the customer's basic information and financial data are identified and recorded. The specific steps for generating a customer information data set are:

[0068] S101: Based on the customer transaction application information, identify and record the customer's personal information, including name, contact information, and address, and generate the customer basic information record.

[0069] In sub-step S101, based on the customer transaction application information, the submitted form data is structured and parsed, the name, contact information, and address fields are explicitly identified and verified, the data format is standardized, and the input information meets the database entry format requirements. The information verification formula is used to check the legitimacy of each data item. The formula is: vz =α z x z +β z y z +γ z z z , where I vz represents the score of information verification, which is used to measure the accuracy and availability of extracted data, α z , β z , γ z are the weight coefficients of name, contact information, and address information, which are used to adjust the sensitivity of the recognition result. z ,y z 、z z Indicates the validity score of the name, contact information, and address information parsed from the table to generate a basic customer information record.

[0070] S102: Based on the basic customer information record, analyze the financial certification documents provided by the customer, including salary slips and bank deposit certificates, identify the customer's monthly income and total deposit information, and generate the customer's financial information in the following specific process;

[0071] In sub-step S102, based on the basic customer information records, the financial certification documents are parsed using text analysis technology to identify key information in the pay slip and bank deposit certificate, including monthly income and total deposit amount. Non-standard items are removed using data cleaning technology, and the financial data is converted into a unified format. The customer's credit score is calculated using the financial data analysis formula, which is: Among them, C sz represents the customer's credit score, μ z and λ z are the weight coefficients of monthly income and deposit, M iz is the monthly income, D kz Generate customer financial information for deposit totals.

[0072] S103: Based on the customer's financial information, the loan type, loan amount and repayment period applied by the customer are analyzed, and combined with the customer's basic information, the process of generating a customer information data set is as follows;

[0073] In sub-step S103, based on the customer's financial information, a comprehensive analysis of loan conditions is conducted, including loan type, loan amount and repayment period. Combined with the customer's credit score and basic information, a composite scoring model is used to calculate the probability of loan approval. The formula is: L pz =δ z C sz +∈ z T kz +φ z A mz +ψ z R tz , where L pz represents the probability of loan approval, δ z ,∈ z ,φ z , z are the weight coefficients of credit score, loan type, loan amount, and repayment period, respectively. sz represents the customer's credit score, T kz is the loan type identifier, A mz represents the loan amount, R tz Indicates the repayment period and generates a customer information dataset.

[0074] See also Figure 3 Based on the customer information data set, by analyzing the customer's historical loan repayment record, credit card usage frequency, default record, and comparing the customer's financial stability with the market average, the customer's credit rating is evaluated. The specific steps for generating a customer credit rating score are:

[0075] S201: Based on the customer information data set, collect and analyze the customer's loan and repayment information, including the loan amount and repayment time, record the number of overdue payments, and generate the loan repayment record. The specific process is as follows;

[0076] In sub-step S201, based on the customer information data set, data extraction technology is used to obtain information about each loan from the database, including the loan amount, loan date, and repayment date. Each piece of data is verified through a data verification process to ensure completeness and accuracy. The repayment delay days are calculated for each loan, and the delay exceeds the standard overdue threshold T. px The situation is counted and the number of overdue times is identified. The formula is:

[0077] Among them, R ox Indicates the number of overdue records. represents the number of days overdue from the i0th loan to the repayment date, T pxRepresents the set overdue judgment threshold, n0 represents the number of loan records. This statistical method tracks and records the repayment behavior of each loan and generates a loan repayment record.

[0078] S202: Based on the loan repayment record, the frequency of use and overdue situation of the customer's credit card are calculated, and the difference with the market average is analyzed and compared to generate the credit behavior analysis result. The specific process is as follows;

[0079] In sub-step S202, based on the loan repayment record, the frequency of each credit card usage is calculated through a data analysis tool, the overdue status of the credit card is determined based on the overdue record, and compared with the market average usage frequency in the industry database. The deviation between the customer's credit behavior and the market average is identified through a difference analysis module. The formula is: Among them, C dx A measure of how different a customer is from the market average, represents the frequency of using the credit card for the j0th time, m0 represents the total number of measurements, It indicates the average usage frequency in the market, identifies the credit behavior of individual customers and compares it with the market behavior to generate credit behavior analysis results.

[0080] S203: Based on the credit behavior analysis results, taking into account the financial stability of the customer, a logistic regression algorithm is used to evaluate and calculate the customer's credit rating, and the specific process of generating the customer's credit rating score is as follows;

[0081] In sub-step S203, based on the results of the credit behavior analysis, a logistic regression model is used to analyze the customer's financial stability and credit behavior data. By collecting multiple financial data such as the customer's annual income, credit card usage frequency, financial responsibility index and historical loan repayment rate, the credit rating score is calculated based on the historical data set and combined with the gradient descent method to evaluate the customer's credit risk and generate a customer credit rating score.

[0082] Logistic regression algorithm, according to the formula:

[0083]

[0084] Calculate the customer's credit default probability, where p is the prediction result, indicating the probability of customer default, e is the base of the natural logarithm, used to calculate the exponential function, β0 is the intercept term, used to adjust the bias in the prediction model and affect the baseline probability of the model, β1 is the weight coefficient of annual income, used to measure the importance and direction of the impact of annual income on the prediction result, β2 is the weight coefficient of credit card usage frequency, used to measure the importance and direction of the impact of credit card usage frequency on the prediction result, β3 is the weight coefficient of the financial responsibility index, used to measure the importance and direction of the impact of the financial responsibility index on the prediction result, β4 is the weight coefficient of the historical loan repayment rate, used to measure the importance and direction of the impact of the historical loan repayment rate on the prediction result, x1 is the annual income, used to measure the customer's economic situation, x2 is the credit card usage frequency, used to measure the customer's consumption behavior, x3 is the financial responsibility index, used to measure the customer's attitude towards financial responsibility, including debt-to-income ratio, savings and investment behavior, and repayment timeliness, and x4 is the historical loan repayment rate, used to measure the customer's repayment reliability.

[0085] The specific execution process of the formula is as follows:

[0086] Collect multiple financial data such as customers’ annual income, credit card usage frequency, financial responsibility index and historical loan repayment rate. Based on the historical data set, use the gradient descent method to train the logistic regression model to obtain multiple weight coefficients β1, β2, β3, β4. Substitute the parameters into the formula to calculate the default probability p of each customer, which is used to evaluate and classify the customer’s credit rating.

[0087] See also Figure 4 , using the customer's credit rating score, by analyzing the customer's deposit and income data, assessing the customer's financial liquidity and asset status, and calculating the customer's monthly repayment ability. The steps to generate the repayment ability analysis results are as follows:

[0088] S301: Using the customer credit rating, by analyzing the customer's deposit certificate and income certificate documents, recording the monthly income and deposit balance, the specific process of generating the customer's monthly financial record is as follows;

[0089] In sub-step S301, based on the customer credit rating, document analysis technology is used to extract key data from the deposit certificate and income certificate documents submitted by the customer, including identifying and recording the monthly income and deposit balance, creating monthly financial records, and recording the monthly income and deposit balance. The formula is: F mc =I c +D c , where F mc Represents the monthly financial records of the customer, I c represents the monthly income data extracted from the income certificate document, D cRepresents the monthly deposit balance data extracted from the deposit certificate file, providing basic data for financial analysis and generating monthly financial records for customers.

[0090] S302: Based on the monthly financial records of the customer, considering the family status, age, and occupation, the customer's monthly fixed expenses and disposable income are evaluated, the asset-liability ratio is calculated, and the specific process of generating financial liquidity assessment data is as follows;

[0091] In sub-step S302, based on the customer's monthly financial records, combined with the customer's family status, age and occupation and other personal information, the customer's monthly fixed expenses and disposable income are evaluated, and the customer's asset-liability ratio is calculated using the formula: Among them, L fc Represents financial liquidity assessment data, A c Represents total assets, including fixed assets and current assets, L c Represents total liabilities, including long-term and short-term liabilities, measures the financial health of customers and conducts loan approvals, and generates financial liquidity assessment data.

[0092] S303: Based on the financial liquidity assessment data, the monthly repayment capacity of multiple customers is calculated, and the credit rating scores of the customers are considered to generate the repayment capacity analysis results. Specifically, the process is as follows;

[0093] In sub-step S303, based on the financial liquidity assessment data, a composite calculation model is used to analyze the monthly repayment capacity of multiple customers, taking into account the credit rating score and monthly net income of each customer, and calculating the repayment capacity index of each customer, the formula is: Among them, P ac Represents the overall solvency analysis results, represents the net monthly income of the i1th customer, It represents the credit rating score of the i1th customer, n1 represents the number of customers included in the analysis, quantifies and compares the repayment ability of differentiated customers, provides a data basis for credit decision-making, and generates repayment ability analysis results.

[0094] See also Figure 5 Based on the repayment capacity analysis results, by real-time monitoring of customer transaction behavior information, identifying abnormal transaction records and fraudulent behavior, the steps for generating abnormal transaction monitoring results are as follows:

[0095] S401: Based on the repayment capacity analysis results, the customer's transaction behavior information is collected and analyzed in real time, including transaction date, amount and location information, and the specific process of generating real-time transaction records is as follows;

[0096] In sub-step S401, based on the repayment capacity analysis results, the customer's transaction behavior information is monitored and collected in real time, the date, amount and location of each transaction are analyzed, and the record score of each transaction is calculated. The formula is: Among them, T rv Represents the comprehensive score of all transactions, used to generate real-time transaction records. Represent the scores of the date, amount and location of the i2th transaction, w d 、w a 、w l is the weight coefficient of date, amount and location, reflecting the importance of multiple factors in record evaluation, and n2 is the total number of transactions during the monitoring period, generating real-time transaction records.

[0097] S402: Based on the real-time transaction records, by comparing and analyzing with the regular transaction patterns, abnormal transaction behaviors are identified, including abnormal large-amount transactions and transactions at abnormal locations. The specific process of generating transaction abnormality identification records is as follows;

[0098] In sub-step S402, real-time transaction records are used to identify abnormal transaction behaviors by calculating standard deviations and means, and standard transaction patterns are compared and analyzed to identify transactions with large amounts and abnormal locations, and abnormal scores are calculated. The formula is: Among them, E tv Indicates the transaction anomaly score, denote the amount and location of the j2th transaction, σ l is the standard deviation of amount and position, is the average of amount and position, m2 is the total number of transactions analyzed, and transaction anomaly identification records are generated.

[0099] S403: Based on the transaction abnormality identification record, combined with the customer's device information and geographic location data, the process of identifying fraudulent behavior and generating abnormal transaction monitoring results is as follows;

[0100] In sub-step S403, based on the transaction anomaly identification record, combined with the customer's device information and geographic location data, potential fraud is identified, and the possibility of fraud is evaluated using a composite analysis formula, which is: Among them, F av Indicates the total score of abnormal transaction monitoring results. represents the device information score of the k2th transaction, represents the geographic location score of the k2th transaction, f d 、f g is the weight coefficient of device information and geographic location, which is used to evaluate the impact of fraudulent behavior, and p2 is the total number of transactions analyzed to generate abnormal transaction monitoring results.

[0101] See also Figure 6 ,According to the abnormal transaction monitoring results, the auto finance market data is collected in real time, the trend and seasonal change pattern of the auto finance market are analyzed and predicted, and the steps for generating market demand analysis data are as follows:

[0102] S501: Based on the abnormal transaction monitoring results, the number of loan applications, leasing activity data and insurance purchase records in the automobile finance market are collected and analyzed in real time, and the specific process of generating market activity data is as follows;

[0103] In sub-step S501, based on the abnormal transaction monitoring results, the real-time data collection system is used to monitor the dynamics of the auto finance market, including the number of loan applications, leasing activity data, and insurance purchase records, which are obtained through API interfaces connected to financial institutions and insurance companies, and data cleaning and standardization are used to ensure the accuracy and consistency of the data, and market activities are evaluated. The formula is: Among them, M bv Represents the summary results of market activity data, They represent the number of loan applications, leasing activity data, and insurance purchase records for the i3th time, respectively, and n3 is the total number of data points considered during the analysis to generate market activity data.

[0104] S502: Based on the market activity data, calculate the monthly loan, lease and insurance demand growth rate, analyze and record seasonal changes, and generate demand change calculation data. The specific process is as follows;

[0105] In sub-step S502, based on the market activity data, a statistical method is used to calculate the monthly growth rate of loan, lease and insurance demand. By analyzing the time series of the data, the seasonal change trend is identified. The calculation formula for the demand growth rate and seasonal change is: Among them, G bv Indicates the summary results of demand change calculation data, It represents the demand data of the j3rd month, q3 represents the number of months analyzed, records the key seasonal changes, and generates demand change calculation data.

[0106] S503: The specific process of calculating data based on demand changes, identifying key factors and seasonal change patterns that affect the automobile financial market, and generating market demand analysis data is as follows;

[0107] In sub-step S503, data is calculated based on demand changes, and data analysis techniques are used to identify key factors and seasonal change patterns that significantly affect the auto finance market. Multivariate regression analysis is used to associate multiple factors with market demand changes, including economic indicators, consumer confidence, and historical sales data. The formula for market demand analysis is: bv =β0+β1F1b +β2F 2b +…+β k F kb , where A bv represents the summary result of market demand analysis data, β0 is the intercept of the regression equation, …, β k is the regression coefficient of each variable in the model, …, F kb Represent different market influencing factors respectively, identify the market's driving forces and seasonal patterns, and generate market demand analysis data.

[0108] See also Figure 7 Based on the market demand analysis data and combined with customer credit information, loan parameters, leasing parameters, and insurance parameters are adjusted to match market changes and customer needs. The specific steps for generating service parameter adjustment results are as follows:

[0109] S601: Based on the market demand analysis data, evaluate the market competitiveness of the current market loan interest rate level and loan amount, identify and record the adjustment demand, and generate the adjustment demand assessment record. The specific process is as follows;

[0110] In sub-step S601, based on the market demand analysis data, the market competitiveness of the current market loan interest rate level and loan amount is analyzed, including comparing the current market data with the historical average, and calculating the required adjustment amount to improve market response. The evaluation formula for adjustment demand is: Among them, C qv Indicates the need for adjustment assessment record, represents the loan interest rate of the i4th data point, represents the loan amount of the i4th data point, Represent the market averages of loan interest rate and loan amount, α q , β q is the adjustment factor for the sensitivity of loan interest rate and loan amount, n4 represents the number of data points in the analysis, and the adjustment need assessment record is generated.

[0111] S602: Based on the adjustment demand assessment record, by analyzing the data of the leasing market, assessing the market adaptability of the current lease term and price, calculating the adjustment demand of the leasing parameters, and generating the leasing parameter analysis information, the specific process is as follows;

[0112] In sub-step S602, based on the adjustment demand assessment record, the term and price adaptability of the current leasing market is assessed through data analysis methods, including calculating the deviation from the market average conditions, and determining the necessary adjustments. The calculation formula for the leasing parameter adjustment demand is: Among them, L qv Indicates the lease parameter analysis information. represents the lease term of the j4th data point, represents the rental price of the j4th data point, are the market averages of lease term and price, respectively, q , δ q is the adjustment factor for term and price sensitivity, m4 is the number of data points in the analysis, and generates leasing parameter analysis information.

[0113] S603: Based on the leasing parameter analysis information, the coverage and compensation ratio of the insurance product are adjusted to optimize the insurance price to match the customer's credit rating and market demand. The specific process of generating the service parameter adjustment result is as follows;

[0114] In sub-step S603, based on the leasing parameter analysis information, the insurance product is adjusted, including the optimization of coverage and compensation ratio, and the insurance pricing strategy is adjusted according to the customer's credit rating and market demand. The formula for adjusting the service parameters is: Among them, B qv Indicates the result of service parameter adjustment. represents the coverage of the k4th data point, represents the payout ratio of the k4th data point, are the market averages of coverage and loss ratio, ∈ q , q is the adjustment factor for coverage and loss ratio sensitivity, and p4 is the number of data points in the analysis, generating the service parameter adjustment results.

[0115] See also Figure 8 , an automobile financial service system based on a large language model, the automobile financial service system based on a large language model is used to execute the above-mentioned automobile financial service method based on a large language model, and the system includes:

[0116] The customer information analysis module analyzes the data based on the customer transaction application information, identifies and records occupation, age, deposit and income data, and generates a basic customer profile;

[0117] The customer credit assessment module analyzes the customer's historical loan repayment record and credit card usage frequency based on the customer's basic files, conducts a comparative analysis of financial stability, and generates a credit score assessment result;

[0118] The financial status analysis module analyzes deposit and income data based on the credit score assessment results, calculates the customer's monthly repayment ability, and generates repayment ability assessment information;

[0119] The abnormal transaction monitoring module monitors customer transaction behaviors in real time based on repayment capacity assessment information, including login information and device fingerprints, identifies abnormal transactions and fraudulent behaviors, and generates abnormal activity identification records;

[0120] The market matching adjustment module collects automobile financial market data based on abnormal activity identification records, analyzes market trends and demand changes, adjusts loan, lease and insurance parameters based on customer credit information, and generates service parameter adjustment results.

[0121] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for automobile financial service based on a large language model, characterized in that: The following steps are involved: Based on customer transaction application information, analyze customer loan and lease application information through a large language model, identify and record customer basic information and financial data, and generate a customer information data set; Based on the customer information data set, by analyzing the customer's historical loan repayment record, credit card usage frequency, default record, and comparing the customer's financial stability with the market average, the customer's credit rating is evaluated to generate a customer credit rating score; Using the customer credit rating score, by analyzing the customer's deposit and income data, assessing the customer's financial liquidity and asset status, calculating the customer's monthly repayment ability, and generating a repayment ability analysis result; Based on the repayment capacity analysis results, by real-time monitoring of customer transaction behavior information, identifying abnormal transaction records and fraudulent behavior, and generating abnormal transaction monitoring results; Based on the abnormal transaction monitoring results, real-time collection of automobile finance market data, analysis and prediction of automobile finance market trends and seasonal change patterns, and generation of market demand analysis data; Based on the market demand analysis data and combined with customer credit information, loan parameters, leasing parameters, and insurance parameters are adjusted to match market changes and customer needs, generating service parameter adjustment results.

2. The automobile financial service method based on a large language model according to claim 1, characterized in that: The customer information data set includes customer occupation information, customer age information, customer deposit and income data; the customer credit rating score includes credit score range, financial stability rating information, and market benchmark data; the repayment capacity analysis results include monthly disposable income, estimated monthly repayment amount, and financial liquidity evaluation results; the abnormal transaction monitoring results include fraud type identification results, abnormal transaction patterns, and transaction amount ranges; the market demand analysis data includes loan demand growth rate data, leasing market activity, and insurance product preference information; the service parameter adjustment results include adjusted loan interest rate range, leasing conditions adjustment records, and auto insurance product adjustment results.

3. The automobile financial service method based on a large language model according to claim 1, characterized in that: Based on the customer transaction application information, the customer's loan and lease application information is analyzed through a large language model, and the customer's basic information and financial data are identified and recorded. The specific steps for generating a customer information data set are as follows: Based on the customer transaction application information, identify and record the customer's personal information, including name, contact information, address, and generate a customer basic information record; Based on the basic customer information records, analyzing the financial certification documents provided by the customer, including salary slips and bank deposit certificates, identifying the customer's monthly income and total deposit information, and generating customer financial information; Based on the customer's financial information, the type of loan applied for by the customer, the loan amount and the repayment period are analyzed, and combined with the customer's basic information to generate a customer information data set.

4. The automobile financial service method based on a large language model according to claim 1, characterized in that: Based on the customer information data set, by analyzing the customer's historical loan repayment record, credit card usage frequency, default record, and comparing the customer's financial stability with the market average, the customer's credit rating is evaluated, and the steps for generating the customer's credit rating score are as follows: Based on the customer information data set, collect and analyze the customer's loan and repayment information, including loan amount and repayment time, record the number of overdue payments, and generate loan repayment records; Based on the loan repayment record, calculate the customer's credit card usage frequency and overdue situation, analyze and compare the difference with the market average, and generate credit behavior analysis results; Based on the credit behavior analysis results, taking into account the financial stability of the customer, a logistic regression algorithm is used to evaluate and calculate the customer's credit rating and generate a customer credit rating score.

5. The automobile financial service method based on a large language model according to claim 4 is characterized in that: The logistic regression algorithm is based on the formula: Calculate the customer's credit default probability, where p is the prediction result, e is the base of the natural logarithm, β0 is the intercept term, β1 is the weight coefficient of annual income, β2 is the weight coefficient of credit card usage frequency, β3 is the weight coefficient of the financial responsibility index, β4 is the weight coefficient of the historical loan repayment rate, x1 is the annual income, x2 is the credit card usage frequency, x3 is the financial responsibility index, and x4 is the historical loan repayment rate.

6. The automobile financial service method based on a large language model according to claim 1, characterized in that: Using the customer credit rating score, by analyzing the customer's deposit and income data, assessing the customer's financial liquidity and asset status, and calculating the customer's monthly repayment ability, the steps for generating the repayment ability analysis result are specifically as follows: Using the customer credit rating score, generate a monthly financial record for the customer by analyzing the customer's deposit certificate and income certificate documents, recording monthly income and deposit balances; Based on the monthly financial records of the client, taking into account family status, age, and occupation, assess the client's monthly fixed expenses and disposable income, calculate the asset-liability ratio, and generate financial liquidity assessment data; Based on the financial liquidity assessment data, the monthly repayment capacity of multiple customers is calculated, and the credit rating scores of the customers are taken into consideration to generate a repayment capacity analysis result.

7. The automobile financial service method based on a large language model according to claim 1, characterized in that: Based on the repayment capacity analysis results, by real-time monitoring of customer transaction behavior information, identifying abnormal transaction records and fraudulent behavior, the steps of generating abnormal transaction monitoring results are specifically as follows: Based on the repayment capacity analysis results, collect and analyze the customer's transaction behavior information in real time, including transaction date, amount and location information, and generate real-time transaction records; Based on the real-time transaction records, by comparing and analyzing with regular transaction patterns, abnormal transaction behaviors are identified, including abnormal large-amount transactions and transactions at abnormal locations, and transaction abnormality identification records are generated; Based on the transaction anomaly identification records, combined with the customer's device information and geographic location data, fraudulent behavior is identified and abnormal transaction monitoring results are generated.

8. The automobile financial service method based on a large language model according to claim 1, characterized in that: According to the abnormal transaction monitoring results, the steps of collecting automobile financial market data in real time, analyzing and predicting the trend and seasonal change pattern of the automobile financial market, and generating market demand analysis data are as follows: Based on the abnormal transaction monitoring results, the number of loan applications, leasing activity data and insurance purchase records in the auto finance market are collected and analyzed in real time to generate market activity data; Based on the market activity data, calculate the monthly loan, lease and insurance demand growth rates, analyze and record seasonal changes, and generate demand change calculation data; Based on the demand change calculation data, key factors and seasonal change patterns affecting the automobile finance market are identified to generate market demand analysis data.

9. The automobile financial service method based on a large language model according to claim 1, characterized in that: Based on the market demand analysis data, combined with customer credit information, loan parameters, leasing parameters, and insurance parameters are adjusted to match market changes and customer needs, and the steps for generating service parameter adjustment results are as follows: Based on the market demand analysis data, evaluate the market competitiveness of the current market loan interest rate level and loan amount, identify and record adjustment needs, and generate adjustment needs assessment records; Based on the adjustment demand assessment record, by analyzing the data of the leasing market, assessing the market adaptability of the current lease term and price, calculating the adjustment demand of the leasing parameters, and generating leasing parameter analysis information; Based on the lease parameter analysis information, the coverage and compensation ratio of the insurance product are adjusted, the insurance price is optimized to match the customer's credit rating and market demand, and the service parameter adjustment result is generated.

10. An automobile financial service system based on a large language model, characterized in that: According to the automobile financial service method based on a large language model according to any one of claims 1 to 9, the system comprises: The customer information analysis module analyzes the data based on the customer transaction application information, identifies and records occupation, age, deposit and income data, and generates a basic customer profile; The customer credit assessment module analyzes the customer's historical loan repayment record and credit card usage frequency based on the customer basic file, conducts a comparative analysis of financial stability, and generates a credit score assessment result; The financial status analysis module analyzes the deposit and income data based on the credit score evaluation result, calculates the customer's monthly repayment ability, and generates repayment ability evaluation information; The abnormal transaction monitoring module monitors customer transaction behaviors in real time based on the repayment ability assessment information, including login information and device fingerprints, identifies abnormal transactions and fraudulent behaviors, and generates abnormal activity identification records; The market matching adjustment module collects automobile financial market data based on the abnormal activity identification records, analyzes market trends and demand changes, adjusts loan, lease and insurance parameters in combination with customer credit information, and generates service parameter adjustment results.

Images