Data management and control system for post-loan risk monitoring

Abstract

The invention relates to the technical field of risk management and control and data processing, and discloses a data management and control system for post-loan risk monitoring, and the system comprises a multi-source data fusion module which is used for obtaining post-loan financial information of a borrower and external association information, and generating a multi-dimensional feature vector; the post-loan risk assessment module is used for constructing and training an integrated learning model comprising a feature decoupling layer, a base model layer and a meta-model layer, obtaining a post-loan risk assessment model, inputting a multi-dimensional feature vector, and generating a post-loan risk score, a risk contribution degree weight and a post-loan risk type; and the post-loan risk management and control module is used for matching disposal strategies according to post-loan risk types when the post-loan risk score is greater than a dynamic threshold value, performing priority ranking on the disposal strategies based on the risk contribution degree weights, generating a risk disposal scheme and performing real-time adjustment according to the disposal feedback of the risk disposal scheme. According to the invention, accurate evaluation and efficient management and control of post-loan risks can be realized, and the risk management capability of financial institutions is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of risk management and data processing, and in particular to a data management and control system for post-loan risk monitoring. Background Art

[0002] Traditional post-loan risk monitoring systems rely on a single data source, lacking real-time and multi-dimensional data analysis capabilities, making it difficult to accurately capture changes in borrower risk. For example, they focus solely on financial data while ignoring external correlates such as social and consumer information, resulting in an overly one-sided post-loan risk assessment. Furthermore, post-loan risk data management models are rigid and unable to dynamically adapt to complex risk situations. This leads to delayed adjustments to management strategies, impacting the efficiency and effectiveness of risk management and making them unable to meet the demands of modern post-loan risk monitoring.

[0003] Therefore, it is urgent to provide a technical solution to solve the above problems. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides a data management and control system for post-loan risk monitoring.

[0005] In a first aspect, the present invention provides a data management and control system for post-loan risk monitoring, the technical solution of which is as follows: A multi-source data fusion module is used to obtain the borrower's post-loan financial information and external related information in real time, perform data standardization and fusion, and generate a multi-dimensional risk feature vector; wherein the external related information includes: social relationship map, career stability network and consumption behavior link; A post-loan risk assessment module is used to construct and train an integrated learning model including a feature decoupling layer, a base model layer and a meta-model layer to obtain a post-loan risk assessment model, input the multi-dimensional risk feature vector into the post-loan risk assessment model, and generate a post-loan risk assessment result including a post-loan risk score, a risk contribution weight and a post-loan risk type; wherein the feature decoupling layer is used to split the multi-dimensional risk feature vector into risk monitoring structured data, social relationship topology features and consumption behavior time series features; the base model layer is used to generate a financial risk probability, a consumption fluctuation probability and a social risk probability based on the risk monitoring structured data, the social relationship topology features and the consumption behavior time series features; the meta-model layer is used to generate the post-loan risk score, the risk contribution weight and the post-loan risk type based on the financial risk probability, the consumption fluctuation probability and the social risk probability; The post-loan risk management and control module is used to match the disposal strategies in the disposal solution knowledge graph according to the post-loan risk type when the post-loan risk score is greater than the dynamic risk threshold, and prioritize all matched disposal strategies based on the size of the risk contribution weight, generate a risk disposal solution, and adjust the management and control strategy in real time according to the disposal feedback data of the risk disposal solution.

[0006] The beneficial effects of the data management and control system for post-loan risk monitoring of the present invention are as follows: The system of the present invention can achieve accurate assessment and efficient control of post-loan risks through multi-source data fusion and combination with advanced integrated learning models and intelligent management and control strategies, effectively enhance the risk management capabilities of financial institutions, and reduce potential risks.

[0007] Based on the above solution, the data management and control system for post-loan risk monitoring of the present invention can also be improved as follows.

[0008] In an optional manner, the multi-source data fusion module is specifically used to: Obtain the borrower's structured post-loan financial information in real time through a distributed API interface; Utilizing a graph neural network to analyze the borrower's communication records and social platform data, and generate the social relationship graph representing the borrower's social topological relationships; Based on the social security payment record and employer association graph of the borrower, constructing the occupational stability network that characterizes the structural characteristics of the borrower's occupational status; The UnionPay transaction flow of the borrower is analyzed using a temporal convolutional network to obtain the consumption behavior link that characterizes the temporal fluctuation characteristics of the borrower's consumption behavior.

[0009] In an optional manner, the multi-source data fusion module is further configured to: The post-loan financial information, the social relationship graph, the career stability network and the consumer behavior link are respectively subjected to feature alignment, data cleaning and normalization processing, and the processed feature vectors are sequentially subjected to dimensionality reduction processing, feature fusion and feature screening to obtain the multidimensional risk feature vector.

[0010] In an optional manner, the base model layer includes: a gradient boosting tree model, a temporal convolutional network, and a graph attention network; the base model layer is specifically used to: The risk monitoring structured data is input into the gradient boosting tree model to obtain the financial risk probability, the consumption behavior time series features are input into the time series convolutional network to obtain the consumption fluctuation probability, and the social relationship topology features are input into the graph attention network to obtain the social risk probability.

[0011] In an optional manner, the metamodel layer is specifically used to: The financial risk probability, the consumption fluctuation probability, and the social risk probability are input into a first preset formula to obtain a linear weighted result and mapped to the post-loan risk score; wherein the first preset formula is: ; is the linear weighted result, is the first weight coefficient corresponding to the financial risk probability, is the second weight coefficient corresponding to the consumption fluctuation probability, is the third weight coefficient corresponding to the social risk probability, is the financial risk probability, is the consumption fluctuation probability, is the social risk probability; According to a second preset formula, and in combination with the first weight coefficient, the second weight coefficient, and the third weight coefficient, the risk contribution weight corresponding to each weight coefficient is calculated respectively; wherein the second preset formula is: ; , ; For the The risk contribution weight corresponding to the weight coefficient is: For the weight coefficients, For the Weight coefficients; The post-loan risk type is determined based on the probability sizes of the financial risk probability, the consumption fluctuation probability and the social risk probability and the corresponding probability thresholds.

[0012] In an optional manner, the metamodel layer is specifically used to: When the financial risk probability is greater than a first probability threshold and the risk contribution weight corresponding to the first weight coefficient is greater than a first weight value, determining that the post-loan risk type is an income interruption type; When the consumption fluctuation probability is greater than a second probability threshold and the risk contribution weight corresponding to the second weight coefficient is greater than a second weight value, determining that the post-loan risk type is an over-indebtedness type; When the social risk probability is greater than the third probability threshold or all risk probabilities are greater than the fourth probability threshold, the post-loan risk type is determined to be a suspected fraud type.

[0013] In an optional manner, the post-loan risk management module is specifically used to: Using the post-loan risk type as a query condition, matching multiple corresponding disposal strategies from the disposal solution knowledge graph; Calculate the priority score of any disposal strategy based on the timeliness coefficient and multiple risk correction scores corresponding to any disposal strategy, and combine each risk contribution weight, until the priority score of each disposal strategy is obtained; The risk disposal plan is generated according to the priority score.

[0014] In a second aspect, the present invention provides a data management and control method for post-loan risk monitoring, the technical solution of which is as follows: Obtain the borrower's post-loan financial information and external related information in real time and perform data standardization and integration to generate a multi-dimensional risk feature vector; wherein the external related information includes: social relationship map, career stability network and consumption behavior link; An integrated learning model including a feature decoupling layer, a base model layer and a meta-model layer is constructed and trained to obtain a post-loan risk assessment model, and the multi-dimensional risk feature vector is input into the post-loan risk assessment model to generate a post-loan risk assessment result including a post-loan risk score, a risk contribution weight and a post-loan risk type; wherein the feature decoupling layer is used to split the multi-dimensional risk feature vector into risk monitoring structured data, social relationship topology features and consumption behavior time series features; the base model layer is used to generate a financial risk probability, a consumption fluctuation probability and a social risk probability based on the risk monitoring structured data, the social relationship topology features and the consumption behavior time series features; the meta-model layer is used to generate the post-loan risk score, the risk contribution weight and the post-loan risk type based on the financial risk probability, the consumption fluctuation probability and the social risk probability; When the post-loan risk score is greater than the dynamic risk threshold, the disposal strategies in the disposal solution knowledge graph are matched according to the post-loan risk type, and all matched disposal strategies are prioritized based on the size of the risk contribution weight, and a risk disposal solution is generated. The management and control strategy is adjusted in real time according to the disposal feedback data of the risk disposal solution.

[0015] The beneficial effects of the data management and control method for post-loan risk monitoring of the present invention are as follows: The method of the present invention, through the fusion of multi-source data and the combination of advanced integrated learning models and intelligent management and control strategies, can achieve accurate assessment and efficient management of post-loan risks, effectively enhance the risk management capabilities of financial institutions, and reduce potential risks.

[0016] In a third aspect, the technical solution of an electronic device of the present invention is as follows: The invention comprises a memory, a processor and a program stored in the memory and running on the processor. When the processor executes the program, the steps of the data control method for post-loan risk monitoring of the present invention are realized.

[0017] In a fourth aspect, the present invention provides a computer-readable storage medium having the following technical solution: Instructions are stored in the computer-readable storage medium. When the computer-readable storage medium reads the instructions, the computer-readable storage medium executes the steps of the data control method for post-loan risk monitoring of the present invention.

[0018] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are only used to illustrate the embodiments and are not to be considered as limiting the present invention. In addition, the same reference symbols are used to represent the same components throughout the drawings. In the drawings: Figure 1 This is a schematic structural diagram of an embodiment of a data management and control system for post-loan risk monitoring according to the present invention; Figure 2 This is a structural diagram of the post-loan risk assessment model; Figure 3 Schematic diagram of a flow chart of an embodiment of a data management and control method for post-loan risk monitoring according to the present invention; Figure 4 The figure is a schematic structural diagram of an embodiment of an electronic device of the present invention. DETAILED DESCRIPTION

[0020] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0021] Figure 1 FIG. 1 shows a schematic diagram of a data management system 100 for post-loan risk monitoring according to an embodiment of the present invention. Figure 1 As shown, a data management and control system 100 for post-loan risk monitoring includes: a multi-source data fusion module 110, a post-loan risk assessment module 120 and a post-loan risk management and control module 130; The multi-source data fusion module 110 is used to obtain the borrower's post-loan financial information and external related information in real time and perform data standardization and fusion to generate a multi-dimensional risk feature vector.

[0022] In this embodiment, the borrower is assumed to be an individual. External related information includes: social relationship graph, career stability network and consumption behavior link, specifically: 1) Post-loan financial information refers to data related to the borrower's financial status generated after obtaining a loan, which is used to assess the borrower's repayment ability and credit risk. This data includes income data, debt status, repayment history, and asset information. Data sources for this information include bank account statements, credit reports, and third-party credit rating agencies.

[0023] 2) The social relationship graph refers to a network diagram constructed based on the borrower's social interaction data, used to identify risks such as hidden joint debt and fraud rings. The data components of the social relationship graph include: ① Nodes: Borrowers and their social contacts (mobile phone numbers, social media accounts). ② Edges: Call frequency, social platform interaction intensity (number of likes, comments). ③ Community characteristics: Community size, proportion of high-risk nodes (e.g., percentage of blacklisted users). Data sources for the social relationship graph include: telecom operator call records, social media APIs (e.g., WeChat, Weibo), and device sharing information.

[0024] 3) The Career Stability Network is a network constructed by integrating career-related data to assess the sustainability and reliability of a borrower's income source. The Career Stability Network's data consists of career history, social security records, and employer connections. Data sources for the Career Stability Network include Social Security Bureau data, tax filing records, and business information databases. The Consumption Behavior Link refers to the borrower's consumption patterns and time-series data, used to identify risks such as irrational consumption and capital chain disruptions.

[0025] 4) The data components of the consumer behavior chain include consumption patterns, time series characteristics, and abnormal transactions. Data sources for the consumer behavior chain include UnionPay transaction records, Alipay / WeChat payment records, and e-commerce platform consumption data.

[0026] Among them, the multidimensional risk feature vector is a feature vector generated by fusing data containing multiple different dimensions. The specific dimensions are set according to actual needs and there is no restriction here.

[0027] In an optional manner, the multi-source data fusion module 110 is specifically configured to: Through a distributed API interface, the structured post-loan financial information of the borrower is obtained in real time; the communication records and social platform data of the borrower are analyzed using a graph neural network to generate the social relationship graph that represents the social topological relationship of the borrower; based on the social security payment records and employer association graph of the borrower, the career stability network that represents the structured characteristics of the borrower's career status is constructed; the UnionPay transaction flow of the borrower is analyzed using a temporal convolutional network to obtain the consumption behavior link that represents the temporal fluctuation characteristics of the borrower's consumption behavior.

[0028] The distributed API interface adopts a microservice architecture, deploys multiple API service nodes, and achieves high-concurrency request processing through load balancing, enabling real-time access to post-loan financial information. Graph neural networks (GNNs) refer to deep learning models used to process graph-structured data and capture relationships between nodes. Using the GraphSAGE algorithm, node vectors are generated through neighbor sampling and aggregation, thereby generating a social relationship graph. The employer association graph refers to obtaining the borrower's corporate equity structure from the industrial and commercial information API, and using the Node2Vec algorithm to generate corporate node embeddings to characterize industry risk correlations. Structural features of occupational status include: structural features (months of continuous social security payments) and temporal features (such as the frequency of job changes). The temporal convolutional network uses TCN temporal modeling to capture the long-term dependencies of consumer behavior, generating consumer behavior links through time slicing and sliding windows.

[0029] In an optional manner, the multi-source data fusion module 110 is further configured to: The post-loan financial information, the social relationship graph, the career stability network and the consumer behavior link are respectively subjected to feature alignment, data cleaning and normalization processing, and the processed feature vectors are sequentially subjected to dimensionality reduction processing, feature fusion and feature screening to obtain the multidimensional risk feature vector.

[0030] Feature alignment methods include time alignment (e.g., aligning consumption time series features with monthly financial indicators and interpolating missing periods). Data cleaning methods include, but are not limited to, filling in missing values ​​and correcting outliers. Normalization includes, but is not limited to, Z-score normalization for numerical fields (e.g., monthly income, debt-to-equity ratio), and one-hot encoding for categorical fields (e.g., occupation type). Dimensionality reduction involves using principal component analysis to compress feature dimensions (e.g., compressing a social relationship graph from 64 to 16 dimensions). This reduction preserves core information and reduces computational complexity. Feature fusion involves horizontally concatenating all feature vectors. Feature screening involves performing Min-Max normalization on the concatenated multidimensional vectors, calculating feature importance using random forests, and removing features with low importance scores (e.g., less than 0.01).

[0031] The post-loan risk assessment module 120 is used to construct and train an integrated learning model including a feature decoupling layer, a base model layer and a meta-model layer to obtain a post-loan risk assessment model, input the multi-dimensional risk feature vector into the post-loan risk assessment model, and generate a post-loan risk assessment result including a post-loan risk score, a risk contribution weight and a post-loan risk type.

[0032] Among them, such as Figure 2 As shown in the figure, the post-loan risk assessment model includes a feature decoupling layer, a base model layer, and a meta-model layer, which are connected in sequence. The feature decoupling layer is used to split the multidimensional risk feature vector into risk monitoring structured data, social relationship topology features, and consumption behavior time series features. The base model layer is used to generate financial risk probability, consumption fluctuation probability, and social risk probability based on risk monitoring structured data, social relationship topology features, and consumption behavior time series features. The meta-model layer is used to generate post-loan risk scores, risk contribution weights, and post-loan risk types based on financial risk probability, consumption fluctuation probability, and social risk probability.

[0033] The feature decoupling layer works as follows: Based on the structured data index (which records the position index of financial and career stability-related features in the fusion vector), the structured data in the multidimensional risk feature vector is extracted and converted into raw dimensions that can be interpreted by subsequent models, thereby generating risk monitoring structured data. Based on the social topology index (which marks the feature positions after dimensionality reduction and the corresponding raw graph metadata paths), the graph embedding vector in the multidimensional risk feature vector is extracted and loaded with the graph metadata template to restore the graph structure and generate social relationship topology features. Based on the consumption time series index (which records the time window correspondence between time series features in the fusion vector), the time series vector in the multidimensional risk feature vector is extracted and sequence reconstructed to generate consumption behavior time series features. It should be noted that if there are index misalignments in the multidimensional risk feature vector, the indexes are matched based on feature distribution similarity. If the graph data or time series data in the multidimensional risk feature vector is damaged, a generative adversarial network (GAN) is used to complete the missing parts.

[0034] In an optional manner, the base model layer includes: a gradient boosting tree model, a temporal convolutional network, and a graph attention network; the base model layer is specifically used to: The risk monitoring structured data is input into the gradient boosting tree model to obtain the financial risk probability, the consumption behavior time series features are input into the time series convolutional network to obtain the consumption fluctuation probability, and the social relationship topology features are input into the graph attention network to obtain the social risk probability.

[0035] Among them, the risk monitoring structured data is a multi-dimensional numerical feature, including asset-liability ratio, income volatility coefficient, social security payment continuity score, etc. The gradient boosting tree model gradually optimizes the prediction results by integrating multiple decision trees; each tree is trained based on the residual of the previous tree, and finally generates the prediction probability through weighted voting or averaging. The input of the gradient boosting tree model is risk monitoring structured data, and the output is the financial risk probability. The specific processing steps include: ① Each decision tree selects the best split point based on the information gain (such as the Gini coefficient) of the risk monitoring structured data; ② The subsequent decision tree focuses on predicting the prediction error of the previous tree; ③ The output of all decision trees is mapped to financial risk probability through the Sigmoid function. For example, the risk monitoring structured data is input into the gradient boosting tree model. The model uses multiple trees to judge the importance of features such as debt-to-asset ratio and income fluctuation, and outputs .

[0036] Among them, the time series characteristics of consumption behavior are multidimensional time series, such as consumption amount, consumption entropy, and proportion of nighttime consumption over several weeks. The Time Series Convolutional Network (TCN) captures the long-term dependencies and periodic patterns in consumption data through causal convolution and dilated convolution. The input of the Time Series Convolutional Network is the time series characteristics of consumption behavior, and the output is the probability of consumption fluctuation. For example, if it is detected that the proportion of nighttime consumption of a borrower has increased sharply in the past four weeks, the Time Series Convolutional Network outputs the consumption fluctuation probability. .

[0037] Among them, the topological features of social relationships are: the graph structure composed of the adjacency matrix and nodes, as well as the node features. The Graph Attention Network (GAT) dynamically weights and aggregates neighbor node information through the attention mechanism to identify high-risk communities and transmission paths. The input of the Graph Attention Network is the topological features of social relationships, and the output is the social risk probability. The specific processing steps include: ① Calculation of attention coefficient; ② Weighted summation of neighbor node features and updating of the central node representation; ③ Mapping the graph embedding into the social contribution probability through the fully connected layer. .

[0038] In an optional manner, the metamodel layer is specifically used to: The financial risk probability, consumption fluctuation probability and social risk probability are input into the first preset formula to obtain a linear weighted result and mapped into a post-loan risk score.

[0039] Among them, the first preset formula is: ; is the linear weighted result, is the first weight coefficient corresponding to the financial risk probability, is the second weight coefficient corresponding to the probability of consumption fluctuation, is the third weight coefficient corresponding to the social risk probability, is the probability of financial risk, is the probability of consumption fluctuation, is the probability of social risk.

[0040] Among them, the meta-model layer in this embodiment is a logistic regression model, and the first weight coefficient , the second weight coefficient and the third weight coefficient This can be determined by iteratively training the logistic regression model using training data. For example, 、 、 .

[0041] The linear weighted result has a value range of 0-1, which is mapped to a 0-100 post-loan risk score through the Sigmoid function. , the post-loan risk score obtained after mapping is 75.

[0042] According to the second preset formula, and in combination with the first weight coefficient, the second weight coefficient and the third weight coefficient, the risk contribution weight corresponding to each weight coefficient is calculated respectively.

[0043] Among them, the second preset formula is: ; , ; For the The risk contribution weight corresponding to the weight coefficient is: For the weight coefficients, For the A weight coefficient.

[0044] It's important to note that the risk contribution weight is a metric used to quantify the relative importance of different risk dimensions (such as financial, consumer, and social) to the post-loan risk score. The weight coefficient determines the direct impact of each characteristic on the score (which can be positive or negative). The risk contribution weight explains the relative importance of each risk dimension in the decision-making process (on a scale of 0-1).

[0045] The post-loan risk type is determined based on the probability of financial risk, consumption fluctuation probability and social risk probability and the corresponding probability thresholds.

[0046] Specifically, when the financial risk probability is greater than a first probability threshold and the risk contribution weight corresponding to the first weight coefficient is greater than a first weight value, the post-loan risk type is determined to be an income interruption type.

[0047] The first probability threshold is 0.7 by default, and the first weight value is 0.5 by default.

[0048] When the consumption fluctuation probability is greater than the second probability threshold and the risk contribution weight corresponding to the second weight coefficient is greater than the second weight value, the post-loan risk type is determined to be over-indebtedness type.

[0049] The second probability threshold is 0.6 by default, and the second weight value is 0.4 by default.

[0050] When the social risk probability is greater than the third probability threshold or all risk probabilities are greater than the fourth probability threshold, the post-loan risk type is determined to be a suspected fraud type.

[0051] The third probability threshold is 0.8 by default, and the fourth probability threshold is 0.5 by default.

[0052] It should be noted that the above thresholds can be adjusted based on actual circumstances and are not limited here. In addition to the above post-loan risk types, situations where all risk probabilities are less than the threshold and the post-loan risk score is greater than or equal to the dynamic risk threshold are determined to be other risk types. Situations where all risk probabilities are less than the threshold and the post-loan risk score is less than the dynamic risk threshold are determined to be normal risk types.

[0053] The post-loan risk management and control module 130 is used to match the disposal strategies in the disposal solution knowledge graph according to the post-loan risk type when the post-loan risk score is greater than the dynamic risk threshold, and prioritize all matched disposal strategies based on the size of the risk contribution weight, generate a risk disposal solution, and adjust the management and control strategy in real time according to the disposal feedback data of the risk disposal solution.

[0054] The dynamic risk threshold is adjusted primarily based on market fluctuations (e.g., if the industry default rate increases by 5%, the dynamic risk threshold is lowered by 3 points). The default baseline value for the dynamic risk threshold is 70, determined based on the statistical quantile of historical default data. Upper and lower limits are set for the dynamic risk threshold (the upper limit defaults to 85, and the lower limit defaults to 60). Specifically, a Q-learning algorithm is used to update the dynamic risk threshold at preset intervals. For example, during an economic downturn, the dynamic risk threshold can be lowered from 70 to 65 to expand risk coverage.

[0055] Among them, the graph structure of the disposal plan knowledge graph includes: nodes and edges; nodes include: risk type, disposal strategy (strategy ID, applicable conditions, expected effects) and execution effect (repayment rate increase, risk reduction slope); edges include: the relationship between risk type and disposal strategy, and the relationship between disposal strategy and execution effect.

[0056] In an optional manner, the post-loan risk management module 130 is specifically used to: Taking the post-loan risk type as a query condition, a plurality of corresponding disposal strategies are matched from the disposal solution knowledge graph.

[0057] According to the timeliness coefficient and multiple risk correction scores corresponding to any disposal strategy, and combined with each risk contribution weight, the priority score of any disposal strategy is calculated until the priority score of each disposal strategy is obtained.

[0058] The priority score of any of the handling strategies is calculated according to the third preset formula: ; For the The priority score of each disposal strategy, For the The corresponding treatment strategy The risk-adjusted score for each risk, is the timeliness coefficient corresponding to the disposal strategy. When the disposal strategy is immediate strategy, ; When the processing strategy is delay strategy, The first risk refers to financial risk, the second risk refers to consumption risk, and the third risk refers to social risk.

[0059] It should be noted that the risk adjustment score is determined by the following steps: obtaining historical execution data, expert experience rules and test results; ; is the weight of historical data, , the default value is 0.8 when there is sufficient data, and the default value for new processing strategies is 0.2. It is the historical effect average, which refers to the disposal strategy in the past N executions. Risk The average revision rate (if the financial risk decreases by 15%, is 0.15). It is the expert prediction value, specifically the effect estimated by the expert based on the business rules (the value is 0-1). After the disposal strategy is executed, it is updated according to the actual effect. , and get the risk-adjusted score ; is the revised risk correction score, The historical score attenuation coefficient is 0.7 by default. Afterwards, Determined as a risk-modified score.

[0060] The risk disposal plan is generated according to the priority score.

[0061] Among them, if the matched treatment strategies include more than three treatment strategies, only the treatment strategies with the top three priority scores are selected to generate the risk treatment plan.

[0062] In the post-loan risk management module 130, the process of adjusting the management strategy in real time based on the feedback data of the risk treatment plan includes: ① quantifying the feedback data; evaluating the strategic effect of the risk treatment plan; if the actual correction rate is lower than expected, then reducing the priority score of the treatment plan; , To measure the actual effect of the treatment strategy, For the expected effect of the disposal strategy, =(1-0.2×(1-0.5)=0.84) / (1-0.2×(1-0.5)=0.84), lowering the priority of the proposed solution. ② Adjust the dynamic risk threshold based on market fluctuations. ③ Retrain the risk contribution weights dynamically through logistic regression to adapt to new risk patterns.

[0063] The technical solution of this embodiment can achieve accurate assessment and efficient control of post-loan risks through multi-source data fusion and combination with advanced integrated learning models and intelligent management and control strategies, effectively enhance the risk management capabilities of financial institutions, and reduce potential risks.

[0064] Figure 3 The following is a flow chart showing an embodiment of a data control method for post-loan risk monitoring provided by the present invention. Figure 3 As shown, the following steps are included: S1. Obtain the borrower's post-loan financial information and external related information in real time and perform data standardization and integration to generate a multi-dimensional risk feature vector; wherein the external related information includes: social relationship map, career stability network and consumption behavior link; S2. Construct an integrated learning model including a feature decoupling layer, a base model layer and a meta-model layer and train it to obtain a post-loan risk assessment model, input the multi-dimensional risk feature vector into the post-loan risk assessment model, and generate a post-loan risk assessment result including a post-loan risk score, a risk contribution weight and a post-loan risk type; wherein the feature decoupling layer is used to split the multi-dimensional risk feature vector into risk monitoring structured data, social relationship topology features and consumption behavior time series features; the base model layer is used to generate a financial risk probability, a consumption fluctuation probability and a social risk probability according to the risk monitoring structured data, the social relationship topology features and the consumption behavior time series features; the meta-model layer is used to generate the post-loan risk score, the risk contribution weight and the post-loan risk type according to the financial risk probability, the consumption fluctuation probability and the social risk probability; S3. When the post-loan risk score is greater than the dynamic risk threshold, the disposal strategies in the disposal solution knowledge graph are matched according to the post-loan risk type, and all matched disposal strategies are prioritized based on the size of the risk contribution weight, a risk disposal solution is generated, and the management and control strategy is adjusted in real time according to the disposal feedback data of the risk disposal solution.

[0065] In an optional manner, the step of obtaining the borrower's post-loan financial information and external related information in real time includes: Obtain the borrower's structured post-loan financial information in real time through a distributed API interface; Utilizing a graph neural network to analyze the borrower's communication records and social platform data, and generate the social relationship graph representing the borrower's social topological relationships; Based on the social security payment record and employer association graph of the borrower, constructing the occupational stability network that characterizes the structural characteristics of the borrower's occupational status; The UnionPay transaction flow of the borrower is analyzed using a temporal convolutional network to obtain the consumption behavior link that characterizes the temporal fluctuation characteristics of the borrower's consumption behavior.

[0066] In an optional manner, the steps of performing data standardization and fusion on post-loan financial information and external related information to generate a multi-dimensional risk feature vector include: The post-loan financial information, the social relationship graph, the career stability network and the consumer behavior link are respectively subjected to feature alignment, data cleaning and normalization processing, and the processed feature vectors are sequentially subjected to dimensionality reduction processing, feature fusion and feature screening to obtain the multidimensional risk feature vector.

[0067] In an optional manner, the base model layer includes: a gradient boosting tree model, a temporal convolutional network, and a graph attention network; the base model layer is specifically used to: The risk monitoring structured data is input into the gradient boosting tree model to obtain the financial risk probability, the consumption behavior time series features are input into the time series convolutional network to obtain the consumption fluctuation probability, and the social relationship topology features are input into the graph attention network to obtain the social risk probability.

[0068] In an optional manner, the metamodel layer is specifically used to: The financial risk probability, the consumption fluctuation probability, and the social risk probability are input into a first preset formula to obtain a linear weighted result and mapped to the post-loan risk score; wherein the first preset formula is: ; is the linear weighted result, is the first weight coefficient corresponding to the financial risk probability, is the second weight coefficient corresponding to the consumption fluctuation probability, is the third weight coefficient corresponding to the social risk probability, is the financial risk probability, is the consumption fluctuation probability, is the social risk probability; According to a second preset formula, and in combination with the first weight coefficient, the second weight coefficient, and the third weight coefficient, the risk contribution weight corresponding to each weight coefficient is calculated respectively; wherein the second preset formula is: ; , ; For the The risk contribution weight corresponding to the weight coefficient is: For the weight coefficients, For the Weight coefficients; The post-loan risk type is determined based on the probability sizes of the financial risk probability, the consumption fluctuation probability and the social risk probability and the corresponding probability thresholds.

[0069] In an optional manner, the metamodel layer is specifically used to: When the financial risk probability is greater than a first probability threshold and the risk contribution weight corresponding to the first weight coefficient is greater than a first weight value, determining that the post-loan risk type is an income interruption type; When the consumption fluctuation probability is greater than a second probability threshold and the risk contribution weight corresponding to the second weight coefficient is greater than a second weight value, determining that the post-loan risk type is an over-indebtedness type; When the social risk probability is greater than the third probability threshold or all risk probabilities are greater than the fourth probability threshold, the post-loan risk type is determined to be a suspected fraud type.

[0070] In an optional manner, the steps of matching the disposal strategies in the disposal solution knowledge graph according to the post-loan risk type, and prioritizing all matched disposal strategies based on the risk contribution weights to generate a risk disposal solution include: Using the post-loan risk type as a query condition, matching multiple corresponding disposal strategies from the disposal solution knowledge graph; Calculate the priority score of any disposal strategy based on the timeliness coefficient and multiple risk correction scores corresponding to any disposal strategy, and combine each risk contribution weight, until the priority score of each disposal strategy is obtained; The risk disposal plan is generated according to the priority score.

[0071] It should be noted that the beneficial effects of the data control method for post-loan risk monitoring provided by the above embodiment are the same as the beneficial effects of the above data control system 100 for post-loan risk monitoring, and will not be repeated here.

[0072] An electronic device according to an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, any one of the above-mentioned data control methods for post-loan risk monitoring is implemented. That is, an electronic device according to an embodiment of the present invention may include but is not limited to: a processor and a memory; the memory is used to store the computer program; the processor is used to execute the data control method for post-loan risk monitoring shown in any embodiment of the present invention by calling the computer program.

[0073] In an alternative embodiment, an electronic device is provided, such as Figure 4 As shown, Figure 4 The electronic device 4000 shown includes: a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, for example, via a bus 4002. Optionally, the electronic device 4000 may further include a transceiver 4004, which can be used for data exchange between the electronic device and other electronic devices, such as data transmission and / or data reception. It should be noted that in actual applications, the number of transceivers 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0074] Processor 4001 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the present disclosure. Processor 4001 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, or a combination of a DSP and a microprocessor.

[0075] Bus 4002 may include a path for transmitting information between the above components. Bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. Bus 4002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4In the figure, only one thick line is used to represent the bus 4002, but this does not mean that there is only one bus or one type of bus.

[0076] The memory 4003 may be a ROM (Read Only Memory) or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory) or other type of dynamic storage device that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0077] The memory 4003 is used to store application code (computer program) for executing the solution of the present invention, and is controlled by the processor 4001. The processor 4001 is used to execute the application code stored in the memory 4003 to implement the content shown in the above method embodiment.

[0078] Among them, the electronic device can also be a terminal device, and the terminal device can be any terminal device that can install applications and access web pages through applications, including at least one of a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, smart TV, and smart car-mounted device.

[0079] It should be noted that Figure 4 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0080] A computer-readable storage medium according to an embodiment of the present invention stores a computer program, which, when executed by a processor, implements any of the above-mentioned data management and control methods for post-loan risk monitoring.

[0081] Alternatively, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, or the like.

[0082] In an exemplary embodiment, a computer program product or computer program is also provided. The computer program product or computer program includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to implement the aforementioned data management and control method for post-loan risk monitoring.

[0083] Computer program code for performing the operations of the present invention may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0084] It should be understood that the flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the methods and computer program products according to various embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0085] The computer-readable storage medium provided in the embodiments of the present invention may be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0086] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device executes the method shown in the above embodiment.

[0087] The above description is merely an illustration of preferred embodiments of the present invention and the underlying technical principles. Those skilled in the art should understand that the scope of the present invention is not limited to technical solutions formed by specific combinations of the aforementioned technical features. It also encompasses other technical solutions formed by any combination of the aforementioned technical features or their equivalents, without departing from the aforementioned concepts. For example, a technical solution formed by replacing the aforementioned features with (but not limited to) technical features with similar functions disclosed in this invention.

[0088] It should be noted that the terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects and to define a specific order or precedence. Where appropriate, the order used for similar objects may be interchanged, such that the embodiments of the present application described herein can be implemented in an order other than the order shown or described.

[0089] Those skilled in the art will appreciate that the present invention may be implemented as a system, method, or computer program product. Therefore, the present invention may be implemented entirely in hardware, entirely in software (including firmware, resident software, microcode, etc.), or in a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, the present invention may be implemented as a computer program product embodied in one or more computer-readable media containing computer-readable program code.

[0090] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A data management and control system for post-loan risk monitoring, characterized in that: include: A multi-source data fusion module is used to obtain the borrower's post-loan financial information and external related information in real time, perform data standardization and fusion, and generate a multi-dimensional risk feature vector; wherein the external related information includes: social relationship map, career stability network and consumption behavior link; A post-loan risk assessment module is used to construct and train an integrated learning model including a feature decoupling layer, a base model layer and a meta-model layer to obtain a post-loan risk assessment model, input the multi-dimensional risk feature vector into the post-loan risk assessment model, and generate a post-loan risk assessment result including a post-loan risk score, a risk contribution weight and a post-loan risk type; wherein the feature decoupling layer is used to split the multi-dimensional risk feature vector into risk monitoring structured data, social relationship topology features and consumption behavior time series features; the base model layer is used to generate a financial risk probability, a consumption fluctuation probability and a social risk probability based on the risk monitoring structured data, the social relationship topology features and the consumption behavior time series features; the meta-model layer is used to generate the post-loan risk score, the risk contribution weight and the post-loan risk type based on the financial risk probability, the consumption fluctuation probability and the social risk probability; The post-loan risk management and control module is used to match the disposal strategies in the disposal solution knowledge graph according to the post-loan risk type when the post-loan risk score is greater than the dynamic risk threshold, and prioritize all matched disposal strategies based on the size of the risk contribution weight, generate a risk disposal solution, and adjust the management and control strategy in real time according to the disposal feedback data of the risk disposal solution.

2. The data management and control system for post-loan risk monitoring according to claim 1, characterized in that: The multi-source data fusion module is specifically used for: Obtain the borrower's structured post-loan financial information in real time through a distributed API interface; Utilizing a graph neural network to analyze the borrower's communication records and social platform data, and generate the social relationship graph representing the borrower's social topological relationships; Based on the social security payment record and employer association graph of the borrower, constructing the occupational stability network that characterizes the structural characteristics of the borrower's occupational status; The UnionPay transaction flow of the borrower is analyzed using a temporal convolutional network to obtain the consumption behavior link that characterizes the temporal fluctuation characteristics of the borrower's consumption behavior.

3. The data management and control system for post-loan risk monitoring according to claim 2, characterized in that: The multi-source data fusion module is further configured to: The post-loan financial information, the social relationship graph, the career stability network and the consumer behavior link are respectively subjected to feature alignment, data cleaning and normalization processing, and the processed feature vectors are sequentially subjected to dimensionality reduction processing, feature fusion and feature screening to obtain the multidimensional risk feature vector.

4. The data management and control system for post-loan risk monitoring according to claim 1, characterized in that: The base model layer includes: a gradient boosting tree model, a temporal convolutional network, and a graph attention network; the base model layer is specifically used to: The risk monitoring structured data is input into the gradient boosting tree model to obtain the financial risk probability, the consumption behavior time series features are input into the time series convolutional network to obtain the consumption fluctuation probability, and the social relationship topology features are input into the graph attention network to obtain the social risk probability.

5. The data management and control system for post-loan risk monitoring according to claim 1, characterized in that: The metamodel layer is specifically used for: The financial risk probability, the consumption fluctuation probability, and the social risk probability are input into a first preset formula to obtain a linear weighted result and mapped to the post-loan risk score; wherein the first preset formula is: ; is the linear weighted result, is the first weight coefficient corresponding to the financial risk probability, is the second weight coefficient corresponding to the consumption fluctuation probability, is the third weight coefficient corresponding to the social risk probability, is the financial risk probability, is the consumption fluctuation probability, is the social risk probability; According to a second preset formula, and in combination with the first weight coefficient, the second weight coefficient, and the third weight coefficient, the risk contribution weight corresponding to each weight coefficient is calculated respectively; wherein the second preset formula is: ; , ; For the The risk contribution weight corresponding to the weight coefficient is: For the weight coefficients, For the Weight coefficients; The post-loan risk type is determined based on the probability sizes of the financial risk probability, the consumption fluctuation probability and the social risk probability and the corresponding probability thresholds.

6. The data management and control system for post-loan risk monitoring according to claim 5, characterized in that: The metamodel layer is specifically used for: When the financial risk probability is greater than a first probability threshold and the risk contribution weight corresponding to the first weight coefficient is greater than a first weight value, determining that the post-loan risk type is an income interruption type; When the consumption fluctuation probability is greater than a second probability threshold and the risk contribution weight corresponding to the second weight coefficient is greater than a second weight value, determining that the post-loan risk type is an over-indebtedness type; When the social risk probability is greater than the third probability threshold or all risk probabilities are greater than the fourth probability threshold, the post-loan risk type is determined to be a suspected fraud type.

7. The data management and control system for post-loan risk monitoring according to claim 5 or 6, characterized in that: The post-loan risk management and control module is specifically used to: Using the post-loan risk type as a query condition, matching multiple corresponding disposal strategies from the disposal solution knowledge graph; Calculate the priority score of any disposal strategy based on the timeliness coefficient and multiple risk correction scores corresponding to any disposal strategy, and combine each risk contribution weight, until the priority score of each disposal strategy is obtained; The risk disposal plan is generated according to the priority score.

8. A data management and control method for post-loan risk monitoring, characterized in that: include: Obtain the borrower's post-loan financial information and external related information in real time and perform data standardization and integration to generate a multi-dimensional risk feature vector; wherein the external related information includes: social relationship map, career stability network and consumption behavior link; An integrated learning model including a feature decoupling layer, a base model layer and a meta-model layer is constructed and trained to obtain a post-loan risk assessment model, and the multi-dimensional risk feature vector is input into the post-loan risk assessment model to generate a post-loan risk assessment result including a post-loan risk score, a risk contribution weight and a post-loan risk type; wherein the feature decoupling layer is used to split the multi-dimensional risk feature vector into risk monitoring structured data, social relationship topology features and consumption behavior time series features; the base model layer is used to generate a financial risk probability, a consumption fluctuation probability and a social risk probability according to the risk monitoring structured data, the social relationship topology features and the consumption behavior time series features; the meta-model layer is used to generate the post-loan risk score, the risk contribution weight and the post-loan risk type according to the financial risk probability, the consumption fluctuation probability and the social risk probability; When the post-loan risk score is greater than the dynamic risk threshold, the disposal strategies in the disposal solution knowledge graph are matched according to the post-loan risk type, and all matched disposal strategies are prioritized based on the size of the risk contribution weight, and a risk disposal solution is generated. The management and control strategy is adjusted in real time according to the disposal feedback data of the risk disposal solution.

9. An electronic device, characterized in that: The electronic device includes a processor, which is coupled to a memory. The memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor so that the electronic device implements the data management and control method for post-loan risk monitoring as described in claim 8.

10. A computer-readable storage medium, characterized in that At least one computer program is stored in the computer-readable storage medium, and the at least one computer program is loaded and executed by the processor so that the computer-readable storage medium implements the data management and control method for post-loan risk monitoring as described in claim 8.

Images