Method and system for generating financial information graph of user through financing platform

Abstract

The invention relates to the technical field of financial information processing, in particular to a method and system for generating a financial information graph of a user through a financing platform, and the method comprises the following steps: extracting a financing limit, financing time, a financing interest rate and a repayment period in a user financing record based on user behavior data; and calling the timestamps to sort the financing behavior records according to a time sequence. According to the method, the quota, the time, the interest rate and the repayment period in the user financing data are dynamically extracted, time sequence analysis is integrated, continuous tracking and deep mining of the financing behavior are achieved, the fluctuation rule of the financing amount and the interest rate is effectively recognized, and the user repayment behavior and credit characteristics are combined. According to the method, a feature coupling matrix and a time weight distribution model are generated, the accuracy and comprehensiveness of user behavior prediction are remarkably improved, a visual financial information graph can be generated by combining dynamic parameters, time weights and credit features, and clearer data support and visual expression are provided for financing decision making.

Description

Technical Field

[0001] The present invention relates to the technical field of financial information processing, and in particular, to a method and system for generating a financial information graph of a user through a financing platform. Background Art

[0002] The technical field of financial information processing includes all methods and systems for collecting, storing, analyzing, and presenting financial data using information technology means. Its core content lies in providing accurate and efficient financial services and support for users through the integration and calculation of massive financial data. This technical field covers data collection technologies, structured and unstructured data processing, data analysis methods, and data visualization means, etc., and involves multiple financial scenarios, including investment management, credit assessment, risk control, and financing decision support, etc. The systematicness of financial information processing technology is reflected in the ability to perform multi-dimensional calculations and cross-domain integrations on various types of data, so as to realize the dynamic information generation and display based on user needs.

[0003] Among them, the method for generating a financial information graph of a user through a financing platform refers to a specific method that, in the context of a financing platform scenario, generates an intuitive graphical information expression for the user's financial data and financing needs by using data modeling analysis and visualization technology. The patent theme mainly covers the process of classifying and extracting features from the financing application data submitted by the user. By cleaning and correlating the user's financial information, combined with multi-dimensional feature modeling, a visualization graph including time series analysis and data distribution analysis is generated. At the same time, this method obtains dynamic transaction information on the financing platform through an embedded financial data interface and comprehensively compares it with the user's characteristic data to achieve the graphical generation process. The above method uses data correlation analysis and visualization expression means to systematically process and graphically display the user's financial data.

[0004] The existing technologies for processing user financing data mostly stay at the static analysis level, lack the full utilization of the time dynamics of the data, and are difficult to comprehensively capture the changing trends of user financing behaviors. The correlation analysis of time series and user characteristics is insufficient, resulting in limited accuracy of risk prediction. In terms of data visualization, the existing technologies have limited ability to interpret complex data and cannot form a clear display of user behaviors and characteristics, which may cause a decrease in the matching degree of financial services and limit the role of dynamic data in financing decisions. Summary of the Invention

[0005] The purpose of the present invention is to solve the deficiencies existing in the prior art, and to propose a method and system for generating a financial information graph of a user through a financing platform.

[0006] To achieve the above object, the present invention adopts the following technical solutions: A method for generating a user's financial information graph through a financing platform, comprising the following steps: S1: Based on user behavior data, extract the financing amount, financing time, financing interest rate, and repayment period in the user's financing record, call the timestamp to sort the financing behavior records in chronological order, calculate the difference in financing amounts between adjacent time points, integrate the interest rate change data to calculate the fluctuation range, and combine the change trend of the repayment ratio to generate a set of user financing dynamic parameters; S2: Based on the set of user financing dynamic parameters, analyze the change rate of the difference in financing amounts and the interest rate fluctuation range, use the time data in the user's repayment behavior for normalization processing, and construct a financing behavior feature vector for time nodes through cumulative calculation to generate a set of user financing cumulative feature vectors; S3: Based on the set of user financing cumulative feature vectors, extract time series data, combine the score and default probability in the user's credit characteristics, analyze the change of eigenvalue at multiple time nodes, calculate the feature coupling matrix, and perform classification processing by period to generate a user financing dynamic coupling feature matrix; S4: Based on the user financing dynamic coupling feature matrix, analyze the time dependence coefficient, combine the overdue data and repayment time distribution in the user's repayment behavior, calculate the time weight distribution, and classify the results by time period to generate a user financing time weight distribution model; S5: Based on the user financing time weight distribution model, combine it with the user financing dynamic coupling feature matrix, perform cumulative weight calculation for time periods, analyze the relationship between the financing amount and credit score data, construct the node distribution and edge weights, and generate a user financial information graph.

[0007] The set of user financing dynamic parameters specifically includes the difference in financing amounts, interest rate changes, and repayment ratio trends. The set of user financing cumulative feature vectors includes the change rate of financing amounts, interest rate fluctuation ranges, and time normalization data. The user financing dynamic coupling feature matrix specifically refers to time series data, credit scores, and default probabilities. The user financing time weight distribution model specifically includes time dependence coefficients, overdue data, and repayment time distributions. The user financial information graph includes node distributions and edge weights.

[0008] As a further solution of the present invention, the specific steps for obtaining the set of user financing dynamic parameters are as follows: S101: Read the financing amount and financing time of the user's financing record, arrange the data according to the timestamp, and calculate the difference sequence of the financing amounts based on consecutive time points. Analyze the change of the financing interest rate data, and classify the change of the financing amount and its associated parameters in combination with the repayment period data to generate a sequence of changes in the financing amount; S102: Combine the financing amount change sequence with the financing interest rate data, normalize the financing interest rate and the repayment ratio parameter within each fluctuation interval in the financing behavior, generate a trend feature quantity based on the numerical change between the two, and segment and extract the trend sequence of the numerical change of the trend feature quantity according to the repayment cycle to generate a repayment ratio trend sequence; S103: Based on the financing amount change sequence and the repayment ratio trend sequence, analyze the correlation characteristics of the change trend of the repayment ratio with respect to the fluctuation of the financing amount, and use the formula: ; Calculate the dynamic parameter of the change trend of the repayment ratio with respect to the fluctuation of the financing amount, and generate a set of user financing dynamic parameters; Wherein, represents the comprehensive index of the set of financing dynamic parameters, represents the financing amount at the i-th time point, represents the financing amount at the (i - 1)-th time point, represents the financing interest rate, represents the change rate of the repayment ratio, represents the total number of time points of the financing record.

[0009] As a further solution of the present invention, the steps for obtaining the set of user financing cumulative feature vectors are specifically as follows: S201: Based on the set of user financing dynamic parameters, analyze the change rate of the difference in financing amount and the interest rate fluctuation interval, calculate the change amount of the financing amount between adjacent time points in the financing behavior record, combine the upper and lower bound differences of the interest rate fluctuation interval to judge the fluctuation characteristics of the change rate, screen the financing behavior records that meet the conditions and extract the feature values to generate the basic feature quantity of the financing behavior; S202: Use the basic feature quantity of the financing behavior to normalize the time dimension parameter in the feature quantity, adjust the distribution interval based on the average value and standard deviation of the basic feature quantity, and perform normalization operations on the difference in financing amount and the change rate of the upper and lower bounds of the interest rate fluctuation according to a unified range to generate a normalized financing behavior feature quantity; S203: Based on the normalized financing behavior feature quantity, construct a financing behavior feature vector for the time node, and use the formula: ; Through weighted cumulative calculation, obtain the set of user financing cumulative feature vectors; Wherein, represents the result of the set of user financing cumulative feature vectors, represents the normalized financing behavior feature quantity, is the average value calculated during the normalization process, is the standard deviation calculated during the normalization process, is the weight coefficient, which is used to adjust the influence degree of the normalized feature quantity in the cumulative feature vector. is the total number of normalized financing behavior feature quantities.

[0010] As a further solution of the present invention, the steps for obtaining the user financing dynamic coupling feature matrix are specifically as follows: S301: Extract time series data from the user financing cumulative feature vector set, combine the change parameters of the user credit feature score and the default probability, analyze the trend of the time node eigenvalue, and at the same time perform segmented processing on the difference of the eigenvalue based on the change trend and extract the feature change rate to generate a time series analysis result; S302: Call the time series analysis result data, calculate the coupling degree between the eigenvalue of each time point and the user credit score and the default probability, construct the feature distribution relationship between time points through the change rate of the eigenvalue, and at the same time measure the coupling strength between different features in the time series data based on the standardization method to generate a feature coupling analysis result; S303: Based on the feature coupling analysis result, calculate the feature coupling matrix, measure the mutual relationship between eigenvalues through normalization, and use the formula: ; Classify and process by period to generate a user financing dynamic coupling feature matrix; Among them, represents the coupling degree between feature i and feature j, represents the normalized value of feature i at time point t, represents the normalized value of feature j at time point t, is the total number of time points.

[0011] As a further solution of the present invention, the steps for obtaining the user financing time weight distribution model are specifically as follows: S401: According to the time dependence coefficient in the user financing dynamic coupling feature matrix, evaluate the behavior anomaly rate of multiple time nodes by matching the overdue days and repayment time distribution data in the user repayment behavior, extract the key credit behavior pattern features based on the change of the anomaly rate, and generate a time dependence analysis result; S402: Match the behavior pattern features of multiple time nodes in the time dependence analysis result with time periods, calculate the credit influence weight within each time period cumulatively according to the overdue degree, repayment time distribution law and feature weight, and at the same time combine the central tendency and change law of the behavior within the time period to generate a time weight calculation result; S403: According to the time weight calculation result, classify the user credit behavior by time period, and use the formula: ; Combined with Gaussian weighting and time-dependent features, through the weighted superposition analysis of multiple time periods and classified processing according to time periods, a user financing time weight distribution model is generated; Among them, represents the weight value at time t, which is used to measure the influence degree of time t on the user's financing behavior, represents the weight coefficient of time period classification, reflecting the key role of multiple time periods in the user's financing behavior, is the central time point of time period i, indicating the key monitoring point within the time period, is the standard deviation of the time period, indicating the width range of time distribution.

[0012] As a further solution of the present invention, the steps for obtaining the user financial information graph are specifically as follows: S501: Invoke the user financing time weight distribution model and the user dynamic coupling feature matrix, construct the weight cumulative feature within the time period through the coupling relationship between the time weight parameter and the feature matrix, and at the same time combine the interaction result of the weight cumulative feature to evaluate the dynamic association characteristics within the time period, and generate the time period cumulative weight analysis result; S502: For the time period cumulative weight analysis result, calculate the change range of the financing amount and the credit score within different time periods, extract the amount and score association features within the time period, evaluate the dynamic influence degree of the amount change on the score, and judge the key connection points through the relationship between the node features and the data change of adjacent nodes, and generate the relationship analysis result; S503: Combine the relationship analysis result, construct the node distribution and edge weight, and use the formula: ; Through the weighted processing of the node features and connection strength, construct the edge weight and node distribution, and generate the user financial information graph; Among them, represents the weight distribution of node v and edge e in the graph, is the basic weight of the node, indicating the initial key role of each node, is the adjustment coefficient, reflecting the sensitivity of the connection strength between nodes, is the eigenvalue of node i, indicating the financial attribute value of each node, is the center point of the adjustment coefficient, indicating the reference value for the weight balance between nodes.

[0013] A system for generating a user financial information graph through a financing platform, the system for generating a user financial information graph through a financing platform is used to execute the method for generating a user financial information graph through a financing platform, and the system includes: The financing behavior extraction module extracts the financing amount, financing time, financing interest rate, and repayment cycle from the financing records based on the user behavior data, sorts the records in timestamp order, calculates the difference in the financing amount between adjacent time points, integrates the interest rate change data to calculate the fluctuation range, calculates the change trend of the repayment ratio, and generates a set of financing dynamic parameters; The financing feature vector generation module analyzes the change rate of the difference in the financing amount and the interest rate fluctuation range based on the set of financing dynamic parameters, normalizes the repayment time data, constructs the time node eigenvalue of the financing behavior through cumulative calculation, and generates a set of financing cumulative feature vectors; The feature matrix analysis module extracts the time series data from the set of financing cumulative feature vectors, combines the user credit score and the default probability, analyzes the change of the multi-time node eigenvalue, calculates the feature coupling relationship, classifies and processes the feature coupling data, and generates a financing dynamic coupling feature matrix; The time weight modeling module analyzes the time dependence coefficient based on the financing dynamic coupling feature matrix, combines the overdue data and the repayment time distribution in the repayment behavior, calculates the time weight distribution, classifies and calculates the time weight value, and generates a financing time weight distribution model; The financial information graph generation module calculates the cumulative weight of the time period based on the financing time weight distribution model, combines the financing dynamic coupling feature matrix, analyzes the relationship between the financing amount and the credit score, calculates the node distribution and the edge weight value, and generates a user financial information graph.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, by dynamically extracting the amount, time, interest rate, and repayment cycle in the user financing data and integrating the time series analysis, the continuous tracking and in-depth mining of the financing behavior are realized, and the fluctuation rules of the financing amount and the interest rate are effectively identified. Combining the user repayment behavior and credit characteristics, a feature coupling matrix and a time weight distribution model are generated, significantly improving the accuracy and comprehensiveness of the prediction of the user behavior. By combining the dynamic parameters, time weights, and credit characteristics, an intuitive financial information graph can be generated, providing clearer data support and visual expression for the financing decision-making, while enhancing the correlation analysis ability of the data, optimizing the risk identification and personalized service effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a schematic diagram of the working process of the present invention; Figure 2 is a flowchart of the acquisition steps of the user financing dynamic parameter set of the present invention; Figure 3 is a flowchart of the acquisition steps of the user financing cumulative feature vector set of the present invention; Figure 4 is a flowchart of the acquisition steps of the user financing dynamic coupling feature matrix of the present invention; Figure 5 This is a flowchart of the steps for obtaining the user financing time weight distribution model of the present invention; Figure 6 This is a flowchart of the steps for obtaining the user financial information graph of the present invention. Detailed implementation manners

[0016] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to 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 used to limit the present invention.

[0017] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is 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 should not be construed as a limitation of the present invention. In addition, in the description of the present invention, the meaning of "a plurality of" is two or more unless otherwise specifically defined. Embodiment

[0018] Please refer to Figure 1 , the present invention provides a technical solution: a method for generating a user financial information graph through a financing platform, including the following steps: S1: Based on user behavior data, extract the financing amount, financing time, financing interest rate and repayment period in the user financing record, call the time stamp to sort the financing behavior records in chronological order, calculate the difference in financing amounts between adjacent time points, integrate the interest rate change data to calculate the fluctuation range, and combine the repayment ratio change trend to generate a set of user financing dynamic parameters; S2: Based on the set of user financing dynamic parameters, analyze the change rate of the difference in financing amounts and the interest rate fluctuation range, use the time data in the user repayment behavior for normalization processing, and construct a financing behavior feature vector for time nodes through cumulative calculation to generate a set of user financing cumulative feature vectors; S3: Based on the set of user financing cumulative feature vectors, extract time series data, combine the score and default probability in the user credit characteristics, analyze the change of eigenvalue at multiple time nodes, calculate the feature coupling matrix, and perform classification processing according to the period to generate a user financing dynamic coupling feature matrix; S4: Based on the user financing dynamic coupling feature matrix, analyze the time dependence coefficient, combine the overdue data and repayment time distribution in the user repayment behavior, calculate the time weight distribution, classify the results by time period, and generate a user financing time weight distribution model; S5: Based on the user financing time weight distribution model, combined with the user financing dynamic coupling feature matrix, the cumulative weight calculation of the time period is performed, the relationship between the financing amount and the credit score data is analyzed, the node distribution and edge weight are constructed, and the user financial information graph is generated.

[0019] The user financing dynamic parameter set specifically includes the financing amount difference, interest rate change, and repayment ratio trend. The user financing cumulative feature vector set includes the financing amount change rate, interest rate fluctuation range, and time normalized data. The user financing dynamic coupling feature matrix specifically refers to time series data, credit score, and default probability. The user financing time weight distribution model specifically includes time dependence coefficient, overdue data, and repayment time distribution. The user financial information graph includes node distribution and edge weight.

[0020] See also Figure 2 , the specific steps for obtaining the user financing dynamic parameter set are: S101: Read the financing amount and financing time of the user's financing record, arrange the data by timestamp, calculate the difference sequence of the financing amount based on consecutive time points, analyze the change of financing interest rate data, classify the change of financing amount and its associated parameters in combination with the repayment cycle data, and generate a financing amount change sequence; Call the timestamp to sort the financing behavior records in chronological order, extract the difference in financing amount between adjacent time points in the financing behavior, perform structured integration of various parameters based on data characteristics, generate a time series through the difference in financing amount at consecutive time points, analyze the data range of the financing amount field in the financing record and clean up the outliers, remove the abnormal data points through the upper and lower limit thresholds, and then recalculate the difference in financing amount between adjacent time points. Based on the cleaned data, establish a continuous financing amount change sequence, call the timestamp field to standardize the financing time, and perform range normalization on the financing interest rate field in the financing behavior according to the actual distribution characteristics, and generate a financing amount change sequence by combining the normalized interest rate data with the difference in financing amount changes. After the financing amount change sequence is generated, call the subsequent steps as parameters.

[0021] S102: combining the financing amount change sequence with the financing interest rate data, normalizing the financing interest rate and the repayment ratio parameter in each fluctuation range of the financing behavior, generating a trend feature quantity based on the numerical change between the two, and dividing the numerical change of the trend feature quantity according to the repayment period and extracting the trend sequence to generate a repayment ratio trend sequence; Based on the sequence of changes in the financing amount and the interest rate change data, the fluctuation range of the financing behavior is integrated and calculated. The time series data of the financing amount change difference field is segmented according to time periods, and the relative fluctuation characteristics are analyzed based on the financing interest rate data of each segment. The mean and standard deviation are calculated by matching the financing amount change difference and the interest rate value in each time slice. The standard deviation field of the financing fluctuation within the segmented time is calculated as the basis for evaluating the fluctuation characteristics. A fluctuation characteristic score is generated for the fluctuation characteristics of each time slice. According to the score, a threshold is set to eliminate the time slices that do not meet the set criteria. The mean value of the fluctuation characteristics of the remaining time slices is extracted as the core parameter of the financing behavior fluctuation range. The change trend of the repayment ratio field in the data that meets the fluctuation range is tracked and recorded. The trend difference is calculated through adjacent repayment ratio data points to generate a repayment ratio change trend sequence. Combining the financing behavior fluctuation range and the repayment ratio change trend sequence provides basic parameters for the next analysis.

[0022] S103: Based on the sequence of changes in the financing amount and the repayment ratio trend sequence, analyze the correlation characteristics of the repayment ratio change trend with the financing amount fluctuation, and use the formula: ; Calculate the dynamic parameters of the repayment ratio change trend with respect to the financing amount fluctuation, and generate a set of user financing dynamic parameters; Among them, represents the comprehensive index of the set of financing dynamic parameters, represents the financing amount at the i-th time point, represents the financing amount at the (i - 1)-th time point, represents the financing interest rate, represents the repayment ratio change rate, represents the total number of time points of the financing record.

[0023] Formula: ; The advantage of the formula is that by comprehensively considering the synergistic relationship among the financing amount change difference, the financing interest rate, and the repayment ratio change rate, it can accurately capture the dynamic characteristics of the financing behavior and enhance the model's comprehensive description ability of the fluctuation range and repayment behavior; Detailed explanation of the formula and the formula calculation derivation process: According to the cleaned financing records, using the sequence of changes in the financing amount as the basic data, set the -th financing amount in the sequence of changes in the financing amount as , the financing amount at the adjacent time point as , the -th repayment ratio change value in the repayment ratio change rate sequence as , and the financing interest rate as , first calculate the square of the difference in the financing amount between adjacent time points to obtain , then divide the square of each difference by , perform normalization calculation and execute the summation operation, with the summation range from to , finally calculate the square root to obtain the result , assuming that the financing amount data is , the financing interest rate is , the change rate of the repayment ratio is , the calculation is as follows: 1. Calculate the square of the change difference in the financing amount: ; 2. Calculate the normalization of each component: ; 3. Execute the summation: ; 4. Calculate the square root to obtain : ; This result shows that the comprehensive index value of the financing dynamic parameter set is 409.22, indicating the dynamic fluctuation characteristics of the current financing behavior and providing a key reference for further optimizing the financing behavior.

[0024] Please refer to Figure 3 , the specific steps for obtaining the set of user financing cumulative feature vectors are as follows: S201: Based on the set of user financing dynamic parameters, analyze the change rate of the difference in the financing amount and the interest rate fluctuation range, calculate the change in the financing amount between adjacent time points in the financing behavior record, combine the upper and lower bound differences of the interest rate fluctuation range to judge the fluctuation characteristics of the change rate, screen the financing behavior records that meet the conditions and extract the feature values, and generate the basic feature quantities of the financing behavior; By calculating the change in the financing amount between adjacent time points in the financing behavior record, extract the financing amount data of adjacent time points and calculate the difference, and use the formula to obtain the change in the financing amount for each time period. Combine the financing interest rate fluctuation range, screen the difference between the upper and lower limits of the financing interest rate, and use the formula to calculate the amplitude change within the interest rate fluctuation range, judge whether the change in the financing amount fluctuates significantly, and determine the screening criteria by comparing the change amount with the fluctuation thresholds and 's correlation relationship. Extract the records with a change amount greater than the threshold and calculate the feature quantities. Extract the financing behavior records that meet the conditions through the index timestamp, and combine the results of the change amount calculation to extract the change feature values in the financing behavior; generate the basic feature quantities of the financing behavior.

[0025] S202: Use the basic feature quantities of the financing behavior to normalize the time dimension parameters in the feature quantities, adjust the distribution interval based on the average value and standard deviation of the basic feature quantities, and perform normalization operations on the difference in financing amount and the change rate of the upper and lower bounds of interest rate fluctuations according to a unified range to generate normalized financing behavior feature quantities; Extract the data of the difference in financing amount and the amplitude of interest rate fluctuation, and limit the range of the time dimension data to , and map the values of the time dimension to the interval through the normalization formula . For the difference in financing amount and the amplitude of interest rate fluctuation, use the standardization formulas and for processing respectively, where , are the mean values of the difference in financing amount and the amplitude of interest rate fluctuation, , are the standard deviations. After standardization, all eigenvalue uniformly distributed data is obtained, and the normalized financing behavior feature quantities are generated by combining the normalization results of the time dimension and the standardization results of the difference in financing amount and the amplitude of interest rate fluctuation.

[0026] S203: Based on the normalized financing behavior feature quantities, construct the financing behavior feature vectors of time nodes, using the formula: ; Through weighted cumulative calculation, obtain the set of user financing cumulative feature vectors; Among them, represents the result of the set of user financing cumulative feature vectors, represents the normalized financing behavior feature quantities, is the average value calculated in the normalization process, is the standard deviation calculated in the normalization process, is the weight coefficient, which is used to adjust the influence degree of the normalized feature quantities in the cumulative feature vectors, is the total number of the normalized financing behavior feature quantities.

[0027] Formula: ; The advantage of the formula is that by adding the weight coefficient to adjust the influence degree of different feature quantities, and combining the normalized financing behavior feature quantities , calculate the cumulative feature vectors, which can more flexibly adapt to the differences between multi-feature data and eliminate the dimensional differences between features at the same time.

[0028] Detailed explanation of the formula and the derivation process of formula calculation: Represents the set of cumulative feature vectors of user financing, Represents the normalized financing behavior feature quantity, Is the mean value of the feature quantity, through the formula Calculated, Is the standard deviation of the feature quantity, through the formula Calculated, Is the weight coefficient, determined by analyzing the contribution degree of each feature quantity to the overall behavior, Is the total number of feature quantities, and the calculation steps of the cumulative feature vector are as follows: For the normalized financing behavior feature quantity , first calculate its mean value through the formula And the standard deviation ; Combined with the weight coefficient , after standardizing the feature quantity, accumulate and calculate according to the weight; Calculate the final cumulative feature vector of financing through the cumulative formula .

[0029] Example calculation: Suppose the financing behavior feature quantity is , the mean value , the standard deviation , the weight coefficient is , the calculation process is: , , ; 2. Cumulative calculation is: ; This result shows that the cumulative feature vector value of the financing behavior feature quantity is , and this value reflects the comprehensive performance of all normalized feature quantities after weighted cumulative calculation, and is used to further analyze the dynamic trend of financing behavior.

[0030] Please refer to Figure 4 , and the specific steps for obtaining the dynamic coupling feature matrix of user financing are as follows: S301: Extract time series data through the set of cumulative feature vectors of user financing, combine the change parameters of user credit feature scores and default probabilities, analyze the trend of time node feature values, and at the same time perform segmented processing on the differentiation of feature values based on the change trend and extract the feature change rate to generate the time series analysis result; Extract the timestamp information from each financing behavior record in the user financing cumulative feature vector set, sort each record in chronological order, extract the amount change, interest rate fluctuation, and repayment cycle change of the financing behavior in the time series as time series features, combine the user credit score and default probability, calculate the credit risk index in each record, and map each eigenvalue in the time series to the normalized interval of the credit risk dimension. The normalized interval is bounded by calculating the minimum and maximum values of each eigenvalue. The mapping formula is , where represents the original eigenvalue, represents the value after normalization. By calculating the change rate of each node feature in the time series and the normalized credit eigenvalue, construct a time series analysis matrix with these normalized eigenvalues. Each row in the matrix corresponds to a time node, and each column is the change rate of the eigenvalue or the normalized credit eigenvalue. Calculate the trend characteristics of the time series based on the matrix data and extract the feature change rate. Combine the change rate with the credit eigenvalue to generate the time series analysis result.

[0031] S302: Call the time series analysis result data, calculate the coupling degree for the eigenvalue at each time point with the user credit score and default probability, construct the feature distribution relationship between time points through the change rate of the eigenvalue, and at the same time measure the coupling strength between different features in the time series data based on the standardization method to generate the feature coupling analysis result; Apply standardization processing to measure the correlation between each feature and the user credit score and default probability. Through the data in each column of the time series analysis result matrix, normalize the change rate of each eigenvalue one by one. The normalization method is to calculate the mean and standard deviation of each column of data, and standardize each eigenvalue through the formula , where is the current value, is the mean of this column, is the standard deviation of this column. The standardized eigenvalues are evenly distributed in the interval of zero mean and unit standard deviation. Calculate the coupling relationship between each eigenvalue through the normalized eigenvalue data matrix. The coupling degree is calculated through the Pearson correlation coefficient of the eigenvalue change. The correlation coefficient formula is: , where and are respectively the standardized values of the feature at time and , and are the feature and The mean value is used to calculate the coupling matrix between eigenvalues through the above formula. Combining the normalized correlation data of eigenvalues with credit scores and default probabilities, the feature coupling analysis results are further generated.

[0032] S303: Based on the feature coupling analysis results, calculate the feature coupling matrix, measure the mutual relationship between eigenvalues through normalization, and use the formula: ; Classify and process by period to generate the user financing dynamic coupling feature matrix; Among them, represents the coupling degree between feature i and feature j, represents the normalized value of feature i at time point t, represents the normalized value of feature j at time point t, is the total number of time points.

[0033] Formula: ; The advantage of the formula is that through the weighted inner product calculation between the normalized time series eigenvalues, it can accurately reflect the mutual coupling relationship of each eigenvalue in the entire time series and improve the accuracy of dynamic feature matrix analysis.

[0034] Detailed explanation of the formula and the derivation process of formula calculation: Let the eigenvalue be the normalized value of feature at time , and the eigenvalue be the normalized value of feature at time . The calculation formula is as follows: 1. Calculate the product of feature and feature at time . The formula is . Sum this product over all time points to get ; 2. Calculate the sum of squares of feature and feature over all time points respectively. The formulas are and respectively; 3. Take the square root of the sum of squares and calculate and respectively; 4. Substitute the calculation results of the above two steps into the formula to get the coupling degree ; 5. By substituting the data and , calculate respectively: ; ; ; , ; ; 6. Calculate and obtain .

[0035] This result indicates that the coupling degree between feature and feature is relatively high in the time series, indicating that the changes of the two in the time series have a strong correlation. This coupling degree can be further used to generate a user financing dynamic coupling feature matrix, and the matrix result is obtained through periodic classification processing.

[0036] Please refer to Figure 5 , and the specific steps for obtaining the user financing time weight distribution model are as follows: S401: According to the time-dependent coefficients in the user financing dynamic coupling feature matrix, by matching the overdue days and repayment time distribution data in the user's repayment behavior, evaluate the behavior abnormality rate at multiple time nodes, extract the key credit behavior pattern features based on the change of the abnormality rate, and generate the time-dependent analysis result; First, extract the time-dependent coefficients. Match the features of each time node in the matrix with the corresponding user repayment overdue records, and extract the delayed time days and payment amount data in the overdue records. For the delayed days data, map it to a proportional value through normalization. The normalization formula is , where represents the delayed days data, and represent the minimum value and maximum value of the data respectively, and the proportionalized delayed days are calculated. Secondly, conduct segmented statistics on the repayment time distribution data. Divide all time nodes into fixed time intervals, accumulate the total repayment amount within the time interval, calculate the repayment proportion of each time period, and judge the behavior abnormality of the time node through the proportion. Then, compare the normalized delayed days data with the repayment time distribution data, and calculate the abnormality score of each time period in combination with the fluctuation value of the abnormality ratio. The score formula is , where represents the proportion of delayed days at time point t, represents the proportion of the repayment amount at time point t, is the total number of time nodes; through the above steps, the abnormality rate distribution of each time node is obtained, and the time-dependent analysis result is generated.

[0037] S402: Match the behavioral pattern features of multiple time nodes in the time-dependency analysis results with time periods, and cumulatively calculate the credit impact weights for each time period according to the overdue degree, repayment time distribution pattern, and feature weights. Meanwhile, combine the central tendency and variation pattern of the behavior within the time period to generate the time weight calculation result; First, perform segmented statistics on the abnormal rate data of each time point in the time-dependency analysis results, sum up the abnormal rates of all time points within the time period to obtain the initial time period weight value; second, standardize the abnormal rate within the time period, calculate the normalized abnormal rate distribution, and the normalization formula is where is the abnormal rate weight value of time period i, is the abnormal rate weight value of time period and is the total number of time periods; then, combine the central tendency of the behavior within the time period, calculate the mean and standard deviation of the behavior distribution, and judge the discreteness of the distribution. For time periods with significant central tendency, increase their corresponding weight parameters, and the adjusted weight formula is where is the central tendency adjustment coefficient, is the central tendency score of time period i, specifically the ratio of the distribution mean to the standard deviation; re-normalize the adjusted weight values and map the normalized weight distribution to each time period to generate the time weight calculation result.

[0038] S403: According to the time weight calculation result, classify the user's credit behavior by time period, using the formula: ; Combine Gaussian weighting and time-dependent features, perform weighted superposition analysis through multiple time periods, and classify by time period to generate the user's financing time weight distribution model; where represents the weight value at time t, which is used to measure the influence degree of time t on the user's financing behavior, represents the weight coefficient of time period classification, which reflects the key importance of multiple time periods in the user's financing behavior, is the central time point of time period i, indicating the key monitoring point within the time period, is the standard deviation of the time period, indicating the width range of the time distribution.

[0039] Formula: ; The advantage of the formula is that by combining the classification weight coefficient of time periods and the time-dependent characteristics, it accurately quantifies the degree of influence on time t, and introduces Gaussian distribution parameters to adjust the relationship between time points and time periods, thereby enhancing the flexibility and accuracy of the time weight distribution.

[0040] Detailed explanation of the formula and the derivation process of formula calculation: represents the weight value of time t, which represents the dynamic influence weight of time t on the user's financing behavior. represents the classification weight coefficient of the i-th time period, and its value is calculated through the distribution of time period anomaly rates and adjusted according to the distribution concentration. is the center point of the i-th time period, and its value is the average of the start and end times of the time period, representing the central time point of the time period. is the standard deviation of the time period, and its value is the standard deviation of the behavior distribution within the time period, representing the distribution width of the time period. is a specific time node, obtained by extracting from the actual time-dependent data; 1. Data acquisition: Assume the time period is , , corresponding to the center point , ; 2. Substitute into the calculation: ; = ; 3. Step-by-step calculation: = ; = ; = ; = ; ≈0.1100; This result shows that at time point 3, the weight value is 0.1100, indicating that time point 3 has a medium influence on the financing behavior. After combining the result with the time period anomaly rate and time-dependent characteristics, a user financing time weight distribution model can be further generated.

[0041] Please refer to Figure 6 , the specific steps for obtaining the user's financial information graph are as follows: S501: Call the user financing time weight distribution model and the user dynamic coupling feature matrix, construct the weight accumulation characteristics within the time period through the coupling relationship between the time weight parameter and the feature matrix, and at the same time combine the interaction results of the weight accumulation characteristics to evaluate the dynamic association characteristics within the time period and generate the time period cumulative weight analysis result; First, extract the time weight values in the user financing time weight distribution model, group them by time period, and call the dynamic coupling eigenvalue of each node in the financing dynamic coupling feature matrix to perform interactive calculations on the weights and eigenvalues within the same time period. During the calculation process, use the time weight to weight the dynamic features. The specific calculation is as follows: Represent the dynamic coupling eigenvalue in matrix form as , and the time weight is represented as , where represents the coupling value of the j-th node in the i-th time period, represents the weight value of the i-th time period. Combine the two into the cumulative feature weight value matrix , and the formula is . Through matrix multiplication, obtain the cumulative feature weight value matrix, where each element in the matrix represents the cumulative eigenvalue of node j in time period i. Then, perform node-by-node normalization on each node value in the cumulative feature weight value matrix. The normalization formula is . Through the normalization operation, obtain the time period cumulative weight matrix. The elements in the matrix represent the standardized cumulative feature weight values between the time period and the nodes, and generate the time period cumulative weight analysis result.

[0042] S502: For the time period cumulative weight analysis result, calculate the change range of the financing amount and the credit score within different time periods, extract the associated features of the amount and the score within the time period, evaluate the dynamic influence degree of the amount change on the score, and judge the key connection points through the relationship between the node features and the data changes of adjacent nodes to generate the relationship analysis result; First, extract the time period normalized weight value in the time period cumulative weight analysis result, combine it with the amount change in the financing amount record and the score change in the user credit score data within the time period. By calculating the change range of the amount and the score, evaluate their associated characteristics, and calculate the ratio of the amount change and the score change within the time period. The formula is , where represents the relationship eigenvalue of the financing amount change and the score change in the i-th time period. Further calculate the mean feature within all time periods, calculate the global mean eigenvalue, combine with to screen out the time periods with deviation values higher than the set threshold as the key time periods through the absolute difference

[0043] S503: Based on the combined relationship analysis results, construct the node distribution and edge weights using the formula: ; Through weighted processing of node features and connection strength, construct edge weights and node distribution to generate a user financial information graph; Among them, represents the weight distribution of node v and edge e in the graph, is the basic weight of the node, indicating the initial criticality of each node, is the adjustment coefficient, reflecting the sensitivity of the connection strength between nodes, is the eigenvalue of node i, representing the financial attribute value of each node, is the center point of the adjustment coefficient, indicating the reference value for the weight balance between nodes.

[0044] Formula: ; The benefit of the formula is that by combining the basic weight of the node, the adjustment coefficient, and the eigenvalue of the node, it quantifies the connection strength between nodes, comprehensively considers the dynamic characteristics and adjustment effects of the nodes, and improves the ability to refine the description of user financial information.

[0045] Detailed explanation of the formula and the derivation process of the formula calculation: Assume the number of nodes is 5, and the eigenvalues of the nodes are respectively , , , , , the adjustment coefficients are respectively , , , , , the basic weights are respectively , , , , , and the center point adjustment values are , , , , .

[0046] Gradually substitute into the formula for calculation: For node 1: ; For node 2: ; For node 3: ; For node 4: ; For node 5: ; Final calculation result: ; This result indicates that the comprehensive weight distribution of nodes and edges is 1.028, showing that the overall connection strength between the current nodes and edges is relatively balanced, and the distribution among nodes is consistent with the changes in the basic weight, adjustment coefficient, and eigenvalue. This value can be further used to draw the specific weight distribution of nodes and edges in the user's financial information graph.

[0047] A system for generating a user's financial information graph through a financing platform. The system for generating a user's financial information graph through a financing platform is used to execute the above method for generating a user's financial information graph through a financing platform. The system includes: The financing behavior extraction module extracts the financing amount, financing time, financing interest rate, and repayment period from the financing records based on the user behavior data, sorts the records in chronological order, calculates the difference in financing amounts between adjacent time points, integrates the interest rate change data to calculate the fluctuation range, calculates the change trend of the repayment ratio, and generates a set of financing dynamic parameters; The financing feature vector generation module analyzes the change rate of the difference in financing amounts and the interest rate fluctuation range based on the set of financing dynamic parameters, normalizes the repayment time data, constructs the time node eigenvalue of the financing behavior through cumulative calculation, and generates a set of financing cumulative feature vectors; The feature matrix analysis module extracts the time series data from the set of financing cumulative feature vectors, combines the user credit score and default probability, analyzes the change of the multi-time node eigenvalue, calculates the feature coupling relationship, classifies and processes the feature coupling data, and generates a financing dynamic coupling feature matrix; The time weight modeling module analyzes the time dependence coefficient based on the financing dynamic coupling feature matrix, combines the overdue data and repayment time distribution in the repayment behavior, calculates the time weight distribution, classifies and calculates the time weight value, and generates a financing time weight distribution model; The financial information graph generation module calculates the cumulative weight of the time period based on the financing time weight distribution model, combines the financing dynamic coupling feature matrix, analyzes the relationship between the financing amount and the credit score, calculates the node distribution and edge weight value, and generates a user's financial information graph.

[0048] The above are only the preferred embodiments of the present invention, and do not limit the present invention in other forms. Any person skilled in the art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for generating a user's financial information graph through a financing platform, characterized in that: The following steps are involved: S1: Based on user behavior data, extract the financing amount, financing time, financing interest rate and repayment period from the user's financing record, call the timestamp to sort the financing behavior records in chronological order, calculate the difference in financing amount at adjacent time points, integrate the interest rate change data to calculate the fluctuation range, and combine the repayment ratio change trend to generate the user financing dynamic parameter set; S2: Based on the user financing dynamic parameter set, analyze the change rate of the financing amount difference and the interest rate fluctuation range, use the time data in the user repayment behavior for normalization processing, construct the financing behavior feature vector of the time node through cumulative calculation, and generate the user financing cumulative feature vector set; S3: Based on the accumulated feature vector set of user financing, extract time series data, combine the score and default probability in the user credit feature, analyze the changes in the feature values ​​of multiple time nodes, calculate the feature coupling matrix, classify and process by period, and generate the user financing dynamic coupling feature matrix; S4: Based on the user financing dynamic coupling characteristic matrix, analyze the time dependency coefficient, combine the overdue data and repayment time distribution in the user repayment behavior, calculate the time weight distribution, classify the results by time period, and generate a user financing time weight distribution model; S5: Based on the user financing time weight distribution model, combined with the user financing dynamic coupling feature matrix, perform time period cumulative weight calculation, analyze the relationship between financing amount and credit score data, construct node distribution and edge weight, and generate a user financial information graph.

2. The method for generating a user's financial information graph through a financing platform according to claim 1, characterized in that: The user financing dynamic parameter set specifically includes the financing amount difference, interest rate change, and repayment ratio trend. The user financing cumulative feature vector set includes the financing amount change rate, interest rate fluctuation range, and time normalized data. The user financing dynamic coupling feature matrix specifically refers to time series data, credit score, and default probability. The user financing time weight distribution model specifically includes time dependence coefficient, overdue data, and repayment time distribution. The user financial information graph includes node distribution and edge weight.

3. The method for generating a user's financial information graph through a financing platform according to claim 2, characterized in that: The steps for obtaining the user financing dynamic parameter set are specifically as follows: S101: Read the financing amount and financing time of the user's financing record, arrange the data by timestamp, calculate the difference sequence of the financing amount based on consecutive time points, analyze the change of financing interest rate data, classify the change of financing amount and its associated parameters in combination with the repayment cycle data, and generate a financing amount change sequence; S102: combining the financing amount change sequence with the financing interest rate data, normalizing the financing interest rate and the repayment ratio parameter in each fluctuation range in the financing behavior, generating a trend feature quantity based on the numerical change between the two, and dividing the numerical change of the trend feature quantity according to the repayment cycle and extracting the trend sequence to generate a repayment ratio trend sequence; S103: Based on the financing amount change sequence and the repayment ratio trend sequence, the correlation characteristics of the repayment ratio change trend to the financing amount fluctuation are analyzed, using the formula: ； Calculate the dynamic parameters of the repayment ratio change trend to the financing amount fluctuation, and generate a user financing dynamic parameter set; in, A composite indicator representing a set of financing dynamics parameters, represents the financing amount at the i-th time point, Represents the financing amount at the i-1th time point, represents the financing rate, represents the repayment ratio change rate, Represents the total number of financing records.

4. The method for generating a user's financial information graph through a financing platform according to claim 3, characterized in that: The steps for obtaining the user financing cumulative feature vector set are specifically as follows: S201: Based on the user financing dynamic parameter set, analyze the change rate of the financing amount difference and the interest rate fluctuation range, calculate the change amount of the financing amount at adjacent time points in the financing behavior record, and judge the fluctuation characteristics of the change rate in combination with the difference between the upper and lower bounds of the interest rate fluctuation range, screen the financing behavior records that meet the conditions and extract the characteristic values ​​to generate the basic characteristic amount of the financing behavior; S202: using the basic characteristic quantity of the financing behavior, normalizing the time dimension parameters in the characteristic quantity, adjusting the distribution interval based on the average value and standard deviation of the basic characteristic quantity, and normalizing the financing amount difference and the change rate of the upper and lower limits of interest rate fluctuation within a unified range to generate a normalized financing behavior characteristic quantity; S203: Based on the normalized financing behavior feature quantity, construct a financing behavior feature vector of the time node, using the formula: ； Through weighted cumulative calculation, the user financing cumulative feature vector set is obtained; in, Represents the result of the cumulative feature vector set of user financing, represents the normalized financing behavior characteristic quantity, is the average value calculated during the normalization process, is the standard deviation calculated during normalization, is the weight coefficient, which is used to adjust the influence of the normalized feature quantity in the cumulative feature vector. It is the total number of normalized financing behavior characteristics.

5. The method for generating a user's financial information graph through a financing platform according to claim 4, characterized in that: The steps for obtaining the user financing dynamic coupling feature matrix are specifically as follows: S301: extracting time series data through the user financing cumulative feature vector set, combining the user credit feature score and the change parameter of the default probability, analyzing the trend of the feature value of the time node, and performing segmentation processing on the differentiation of the feature value based on the change trend and extracting the feature change rate to generate the time series analysis result; S302: calling the time series analysis result data, calculating the coupling degree between the characteristic value at each time point and the user credit score and default probability, constructing the characteristic distribution relationship between the time points through the change rate of the characteristic value, and measuring the coupling strength between the differentiated features in the time series data based on the standardized method, and generating the characteristic coupling analysis result; S303: Based on the characteristic coupling analysis results, a characteristic coupling matrix is ​​calculated, and the relationship between the characteristic values ​​is measured by normalization, using the formula: ； Classify and process by period to generate the user financing dynamic coupling feature matrix; in, represents the coupling degree between feature i and feature j, represents the normalized value of feature i at time point t, represents the normalized value of feature j at time point t, is the total number of time points.

6. The method for generating a user's financial information graph through a financing platform according to claim 5, characterized in that: The steps for obtaining the user financing time weight distribution model are specifically as follows: S401: According to the time dependency coefficient in the user financing dynamic coupling feature matrix, by matching the overdue days and repayment time distribution data in the user repayment behavior, evaluating the behavior abnormality rate of multiple time nodes, extracting key credit behavior pattern features based on the abnormality rate changes, and generating time dependency analysis results; S402: Matching the behavior pattern characteristics of multiple time nodes in the time dependency analysis result with the time period, cumulatively calculating the credit impact weight in each time period according to the degree of overdue, the distribution law of repayment time and the characteristic weight, and generating the time weight calculation result by combining the behavior concentration trend and change law in the time period; S403: According to the time weight calculation result, the user's credit behavior is classified by time period, using the formula: ； Combining Gaussian weighting and time-dependent features, we generate a user financing time weight distribution model by performing weight superposition analysis of multiple time periods and classifying them by time period. in, Represents the weight value at time t, which is used to measure the impact of time t on user financing behavior. Represents the weight coefficient of time period classification, reflecting the importance of multiple time periods in user financing behavior. is the central time point of time period i, indicating the key monitoring point within the time period, is the standard deviation of the time period, indicating the width of the time distribution.

7. The method for generating a user's financial information graph through a financing platform according to claim 6, characterized in that: The steps for obtaining the user financial information graph are specifically as follows: S501: calling the user financing time weight distribution model and the user dynamic coupling feature matrix, constructing the weight accumulation feature within the time period through the coupling relationship between the time weight parameter and the feature matrix, and combining the interaction result of the weight accumulation feature to evaluate the dynamic correlation characteristics within the time period, and generating the time period cumulative weight analysis result; S502: Calculate the change range of the financing amount and the credit score in the differentiated time period according to the cumulative weight analysis results of the time period, extract the correlation characteristics between the amount and the score in the time period, evaluate the dynamic impact of the amount change on the score, determine the key connection points through the relationship between the node characteristics and the change of the adjacent node data, and generate the relationship analysis results; S503: Based on the relationship analysis results, construct node distribution and edge weights using the formula: ； Through weighted processing of node features and connection strength, edge weights and node distribution are constructed to generate user financial information graphs; in, represents the weight distribution of nodes v and edges e in the graph, is the basic weight of the node, indicating the initial criticality of each node, is the adjustment coefficient, reflecting the sensitivity of the connection strength between nodes, is the characteristic value of node i, indicating the financial attribute value of each node, It is the center point of the adjustment coefficient, which represents the benchmark value of the weight balance between nodes.

8. A system for generating a user's financial information graph through a financing platform, characterized in that: According to the method for generating a user's financial information graph through a financing platform according to any one of claims 1 to 7, the system comprises: The financing behavior extraction module extracts the financing amount, financing time, financing interest rate and repayment period from the financing records based on user behavior data, sorts the records in timestamp order, calculates the difference in financing amount at adjacent time points, integrates interest rate change data to calculate the fluctuation range, calculates the repayment ratio change trend, and generates a set of financing dynamic parameters; The financing feature vector generation module analyzes the change rate of the financing amount difference and the interest rate fluctuation range based on the financing dynamic parameter set, normalizes the repayment time data, constructs the time node feature value of the financing behavior through cumulative calculation, and generates a financing cumulative feature vector set; The feature matrix analysis module extracts time series data from the accumulated feature vector set of financing, combines the user credit score with the probability of default, analyzes the changes in feature values ​​at multiple time nodes, calculates feature coupling relationships, classifies and processes feature coupling data, and generates a financing dynamic coupling feature matrix; The time weight modeling module analyzes the time dependency coefficient based on the financing dynamic coupling characteristic matrix, combines the overdue data in the repayment behavior with the repayment time distribution, calculates the time weight distribution, classifies and calculates the time weight value, and generates a financing time weight distribution model; The financial information graph generation module is based on the financing time weight distribution model, combined with the financing dynamic coupling feature matrix, calculates the cumulative weight of the time period, analyzes the relationship between the financing amount and the credit score, calculates the node distribution and edge weight value, and generates a user financial information graph.

Images