Rural credit risk early warning method and device, electronic equipment and storage medium

Abstract

The present disclosure provides a rural credit risk early warning method and device, an electronic device and a storage medium, relates to the technical field of credit risk assessment, and through acquiring multi-source heterogeneous data related to a borrower, generating a dynamic credit rating changing over time based on the data, generating risk early warning signals of different levels in combination with the dynamic credit rating and its change trend, executing matched intervention measures in response to the early warning signals and feeding back execution effect data to the credit rating model to realize model optimization, therefore, the problems of risk identification lag, insufficient intervention accuracy and information asymmetry caused by single credit evaluation dimension, static risk assessment, disconnection between early warning and intervention and lack of self-optimization capability in the prior art can be solved, and the technical effects of comprehensive and accurate evaluation of rural credit risk, early warning of potential risks, implementation of accurate and flexible intervention and continuous improvement of risk control efficiency are achieved.

Description

Technical Field

[0001] This disclosure relates to the field of credit risk assessment technology, and in particular to a rural credit risk early warning method and device, electronic equipment and storage medium. Background Technology

[0002] Rural credit, as a core pillar of the inclusive finance system, plays an irreplaceable role in promoting agricultural industrialization and supporting rural revitalization. To ensure the sound operation of credit business, a credit assessment system centered on pre-loan approval has been established in related technical fields, covering all key aspects of the process from data collection and information processing to risk decision-making. Specifically, this includes core operations such as credit data integration, financial indicator analysis, and account status monitoring.

[0003] At the technical implementation level, traditional rural credit risk assessment methods exhibit a clear path dependency: in terms of data sources, they mainly focus on structured data such as central bank credit reports and bank statements, with insufficient integration of diverse data scattered across different departments and scenarios; in terms of model construction, they mostly use ensemble learning algorithms to complete the initial credit score, integrating information from various dimensions through a fixed weighting method to form a single and fixed credit score result, which serves as the core basis for determining credit limits and interest rates; in terms of business application, such scoring results only primarily serve the pre-loan approval stage, providing weak support for post-loan management.

[0004] However, existing technical solutions have many key limitations and are difficult to adapt to the complex needs of rural credit scenarios: First, the scoring mechanism is static and lacks a dynamic credit rating system. It lacks both a systematic adjustment design for indicator weights over time and a dedicated model to capture the evolution of borrowers' credit status, resulting in credit assessment results that cannot promptly reflect fluctuations in borrowers' credit levels caused by changes in business conditions and the external environment. Second, data processing capabilities are significantly weak. On the one hand, it cannot continuously collect and process environmental data such as soil moisture and temperature recorded by IoT sensors, and on the other hand, it cannot handle data retained by e-commerce platforms. On the one hand, the existing technology suffers from unstructured and dynamic data such as sales rate and sales volume. On the other hand, it fails to break down data barriers between different entities such as the government, e-commerce, and cooperatives, and lacks integration of scattered data on agricultural production, subsidy distribution, and consumer behavior, making it difficult to build a comprehensive and three-dimensional credit profile of borrowers. Thirdly, the post-loan risk management mechanism is imperfect. The existing system can only trigger warnings based on simple fixed thresholds such as insufficient account balances, which cannot respond quickly to emergencies such as sudden natural disasters or market price crashes, nor has it formed a closed-loop system covering the entire process of "data collection - dynamic assessment - risk warning - precise intervention - feedback optimization." These deficiencies directly lead to a significant lag in risk identification in existing technologies, and the intervention measures lack specificity and foresight. This makes it difficult to effectively solve the risk management challenges in rural credit business caused by information asymmetry and complex and ever-changing risk factors, and fails to fully meet the actual needs of high-quality development of rural credit business. Summary of the Invention

[0005] This disclosure provides a method, apparatus, electronic device, and storage medium for early warning of rural credit risk. Its main purpose is to at least partially address one of the technical problems in related technologies.

[0006] According to the first aspect of this disclosure, a method for early warning of rural credit risk is provided, comprising:

[0007] Acquire multi-source heterogeneous data related to borrowers from multiple heterogeneous data sources; Based on the acquired multi-source heterogeneous data, a dynamic credit rating that changes over time is generated through a credit rating model. Based on the dynamic credit rating and its changing trend, it is compared with the preset early warning rules to generate risk warning signals of different levels; In response to the aforementioned risk warning signal, implement intervention measures commensurate with the warning level; Collect data on the effectiveness of the intervention measures and feed the data back to the credit rating model to optimize the credit rating model.

[0008] Optionally, the acquisition of multi-source heterogeneous data related to the borrower from multiple heterogeneous data sources includes: Data on the borrower's production and operating environment is collected through IoT devices; The borrower's transaction behavior data is obtained from an external commercial platform via an application programming interface; Obtain relevant policies, subsidies, and registration information for the borrower from a pre-set data platform.

[0009] Optionally, the credit rating model is a model that integrates an initial static score and a dynamic trend score; wherein the dynamic trend score is obtained by analyzing the borrower's time-series behavioral data; The generation of a dynamic credit rating that changes over time includes: Standardize multi-source heterogeneous data from multiple heterogeneous data sources to eliminate dimensional differences; A weighted calculation method is used to integrate the processed multi-dimensional indicator data into a single credit assessment index, wherein the weights of at least some indicators are dynamically adjusted according to the time series characteristics of the data.

[0010] Optionally, the step of comparing the dynamic credit rating and its changing trend with preset early warning rules to generate risk early warning signals of different levels includes: Set up a warning rule base with multiple risk levels; When the decline in the dynamic credit rating or the negative slope of the change trend exceeds the threshold of the corresponding level, a risk warning signal for the corresponding level is triggered, and key risk indicators that lead to the rating change are identified as risk triggers.

[0011] Optionally, the step of responding to the risk warning signal and implementing intervention measures matching the warning level includes: Based on the level of the risk warning signal and the identified risk triggers, the corresponding intervention strategy is matched from the preset strategy library; The intervention strategies include at least one of sending reminder messages to borrowers, adjusting repayment plans, or triggering a manual verification process.

[0012] Optionally, the step of collecting data on the effectiveness of the intervention measures and feeding this data back to the credit rating model to optimize the model includes: The results of interventions that can mitigate risk factors are used as a positive feedback feature and input into the credit rating model. Based on the positive feedback characteristics, the relevant weight parameters in the model are adaptively adjusted to optimize subsequent rating calculations.

[0013] According to a second aspect of this disclosure, a rural credit risk early warning device is provided, comprising: The acquisition unit is used to acquire multi-source heterogeneous data related to the borrower from multiple heterogeneous data sources. The calculation unit is used to calculate, based on the acquired multi-source heterogeneous data, using a credit rating model to generate a dynamic credit rating that changes over time. The comparison unit is used to compare the dynamic credit rating and its changing trend with the preset early warning rules to generate risk warning signals of different levels. An execution unit is used to respond to the risk warning signal and execute intervention measures that match the warning level. The collection unit is used to collect data on the effectiveness of the intervention measures and feed the data back to the credit rating model to optimize the credit rating model.

[0014] Optionally, the acquisition unit is also used for: Data on the borrower's production and operating environment is collected through IoT devices; The borrower's transaction behavior data is obtained from an external commercial platform via an application programming interface; Obtain relevant policies, subsidies, and registration information for the borrower from a pre-set data platform.

[0015] Optionally, the credit rating model is a model that integrates an initial static score and a dynamic trend score; wherein the dynamic trend score is obtained by analyzing the borrower's time-series behavioral data; The computing unit is also used for: Standardize multi-source heterogeneous data from multiple heterogeneous data sources to eliminate dimensional differences; A weighted calculation method is used to integrate the processed multi-dimensional indicator data into a single credit assessment index, wherein the weights of at least some indicators are dynamically adjusted according to the time series characteristics of the data.

[0016] Optionally, the comparison unit is also used for: Set up a warning rule base with multiple risk levels; When the decline in the dynamic credit rating or the negative slope of the change trend exceeds the threshold of the corresponding level, a risk warning signal for the corresponding level is triggered, and key risk indicators that lead to the rating change are identified as risk triggers.

[0017] Optionally, the execution unit is also used for: Based on the level of the risk warning signal and the identified risk triggers, the corresponding intervention strategy is matched from the preset strategy library; The intervention strategies include at least one of sending reminder messages to borrowers, adjusting repayment plans, or triggering a manual verification process.

[0018] Optionally, the collection unit is also used for: The results of interventions that can mitigate risk factors are used as a positive feedback feature and input into the credit rating model. Based on the positive feedback characteristics, the relevant weight parameters in the model are adaptively adjusted to optimize subsequent rating calculations.

[0019] According to a third aspect of this disclosure, an electronic device is provided, comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method described in the first aspect above.

[0020] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are configured to cause the computer to perform the method described in the first aspect above.

[0021] According to a fifth aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method described in the first aspect above.

[0022] The rural credit risk early warning method, device, electronic equipment, and storage medium disclosed herein acquire multi-source heterogeneous data related to borrowers, generate a dynamic credit rating that changes over time based on this data, and generate risk early warning signals of different levels by combining the dynamic credit rating and its changing trend. In response to the early warning signals, matching intervention measures are implemented, and the implementation effect data is fed back to the credit rating model for model optimization. Therefore, it can solve the problems of risk identification lag, insufficient intervention accuracy, and information asymmetry caused by the single dimension of credit assessment, static risk assessment, disconnect between early warning and intervention, and lack of self-optimization capability in existing technologies. It achieves the technical effects of comprehensively and accurately assessing rural credit risk, providing early warning of potential risks, implementing precise and flexible interventions, and continuously improving the efficiency of risk management.

[0023] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0024] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein: Figure 1A flowchart illustrating a rural credit risk early warning method provided in this embodiment of the disclosure; Figure 2 This is a schematic diagram of the structure of a rural credit risk early warning device provided in an embodiment of the present disclosure; Figure 3 A schematic block diagram of an example electronic device provided for embodiments of this disclosure. Detailed Implementation

[0025] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0026] The following description, with reference to the accompanying drawings, outlines an embodiment of a rural credit risk early warning method, apparatus, electronic device, and storage medium.

[0027] Figure 1 This is a flowchart illustrating a rural credit risk early warning method provided in an embodiment of this disclosure.

[0028] like Figure 1 As shown, the method includes the following steps: Step 101: Obtain multi-source heterogeneous data related to the borrower from multiple heterogeneous data sources.

[0029] In the embodiments of this disclosure, to overcome the technical bottlenecks of limited data sources and difficulty in integrating scattered data in rural credit risk assessment, this step utilizes a pre-set data collection mechanism to comprehensively acquire various types of data related to borrowers from multiple heterogeneous data sources with different types and structures. This data covers both structured and unstructured information that characterizes core assessment dimensions such as the borrower's individual credit status, production and operational dynamics, and external environmental impacts. This achieves the systematic aggregation of scattered data across scenarios and departments, providing comprehensive and diverse basic data support for subsequent dynamic credit rating. As one implementation method, the heterogeneous data sources may include historical credit data from financial institutions, registration and subsidy data from relevant government departments, operational and consumption data from commercial platforms, environmental and operational data collected by IoT sensing devices, and publicly released market and meteorological data, etc.

[0030] By integrating multi-source heterogeneous data, the limitations of data silos in traditional rural credit assessment are effectively broken, the data dimensions of credit assessment are enriched, and a solid data foundation is laid for building a comprehensive and accurate credit profile of borrowers. This solves the problem of information asymmetry caused by the reliance on single structured data in traditional assessment from the source.

[0031] Step 102: Based on the acquired multi-source heterogeneous data, a dynamic credit rating that changes over time is generated by calculating using a credit rating model.

[0032] In the embodiments of this disclosure, to overcome the technical limitations of traditional static ratings in reflecting real-time changes in borrowers' credit status, this step utilizes a pre-set credit rating model (i.e., an assessment model capable of integrating multi-dimensional data and capturing the evolutionary patterns over time) to systematically process and deeply analyze the acquired multi-source heterogeneous data. By mining the core features related to borrowers' credit in the data and their changing patterns over time, the model can output a credit rating result that is dynamically updated over time. This result can reflect in real time the fluctuations in the borrower's credit level caused by factors such as changes in their own business operations and adjustments in the external environment, completely breaking free from the static constraints of traditional fixed ratings. As one implementation method, the credit rating model can combine an ensemble learning algorithm to complete the initial credit assessment and introduce a time series analysis model to capture the evolutionary trend of credit status, continuously updating the rating result according to a preset period.

[0033] By generating dynamic credit ratings, the problem of lagging response to changes in credit status in traditional static ratings is effectively solved. This enables continuous tracking and accurate characterization of borrowers' credit status, providing a timely and reliable core basis for subsequent risk warnings, and significantly improving the dynamic adaptability and accuracy of rural credit risk assessment.

[0034] Step 103: Based on the dynamic credit rating and its changing trend, compare it with the preset early warning rules to generate risk early warning signals of different levels.

[0035] In the embodiments of this disclosure, to address the technical shortcomings of traditional rural credit early warning systems that rely solely on fixed, simple thresholds and cannot adapt to dynamic changes in credit status, this step, based on pre-defined early warning rules adapted to the characteristics of rural credit risk (i.e., a rule system covering multiple dimensions of judgment criteria such as credit rating level, magnitude of change, and rate of change), systematically compares the aforementioned generated dynamic credit rating results with their changing trends over time. By identifying abnormal changes in credit status and risk evolution trends, different levels of risk early warning signals corresponding to the severity of risk are generated, ensuring that risks can be accurately identified and differentiated according to severity. As one implementation method, the early warning rules may include thresholds for the decline in dynamic credit ratings within a set period, thresholds for abnormal fluctuations in key operational data, etc., and the corresponding early warning signals can be divided into different response levels to match differentiated handling needs.

[0036] By combining dynamic credit ratings and their changing trends with appropriate early warning rules, the problem of lagging and insufficient targeting of traditional early warning methods is effectively solved. This enables precise tiered early warning of rural credit risks, providing accurate decision-making basis for subsequent implementation of differentiated intervention measures.

[0037] Step 104: In response to the risk warning signal, implement intervention measures that match the warning level.

[0038] In the embodiments of this disclosure, to overcome the technical shortcomings of traditional rural credit intervention measures being passive, lagging, and lacking specificity, this step, upon receiving a risk warning signal, automatically or semi-automatically executes differentiated intervention measures adapted to the warning level based on a preset intervention strategy system. These intervention measures need to be flexibly adapted to the severity of the risk, the risk triggers, and the borrower's actual business situation, ensuring both the controllability of credit business risks and taking into account the borrower's actual needs, avoiding the negative effects of a one-size-fits-all, rigid approach. As one implementation method, for low-level risk warnings, caring interventions such as business guidance and market information dissemination can be implemented; for medium-level risk warnings, supportive interventions such as repayment plan adjustments can be provided; and for high-level risk warnings, risk mitigation interventions such as credit limit control and post-loan on-site verification can be adopted.

[0039] By establishing a precise matching mechanism between early warning levels and intervention measures, the problems of insufficient targeting and passive risk response in traditional intervention methods have been effectively solved. This has enabled flexible and differentiated handling of rural credit risks, which has not only improved the timeliness and effectiveness of risk intervention but also optimized the borrower service experience, thus helping to reduce the non-performing loan rate.

[0040] Step 105: Collect data on the effectiveness of the intervention measures and feed the data back to the credit rating model to optimize the credit rating model.

[0041] In the embodiments of this disclosure, to address the technical shortcomings of traditional credit rating models—such as the lack of a continuous optimization mechanism and difficulty in adapting to the dynamic changes in rural credit scenarios—this step systematically acquires relevant effect data after the implementation of intervention measures through a pre-set data collection mechanism. This data reflects the actual effect of the intervention on risk mitigation and the borrower's feedback. This effect data is then fed back to the credit rating model, providing a direct basis for adjusting model parameters, optimizing feature weights, or iterating judgment logic. This allows the model to continuously upgrade based on actual business practices and adapt to the dynamic evolution of rural credit risk. As one implementation method, data such as the borrower's response to intervention measures and the subsequent improvement of risk status can be collected and fed back to the model to optimize the weight allocation of rating indicators or credit trend judgment rules.

[0042] By constructing a closed-loop mechanism of "intervention effect - data feedback - model optimization", the problems of fixed and insufficient adaptability of traditional credit rating models are effectively solved, the accuracy of the model's assessment of rural credit risk and dynamic response capability are continuously improved, and the self-evolution capability of the risk management system is further strengthened.

[0043] The rural credit risk early warning method disclosed herein acquires multi-source heterogeneous data related to borrowers, generates a dynamic credit rating that changes over time based on this data, and generates risk early warning signals of different levels by combining the dynamic credit rating and its changing trend. It then responds to the early warning signals by implementing matching intervention measures and feeds the implementation effect data back to the credit rating model for model optimization. Therefore, it can solve the problems of risk identification lag, insufficient intervention accuracy, and information asymmetry caused by the single dimension of credit assessment, static risk assessment, disconnect between early warning and intervention, and lack of self-optimization capability in existing technologies. This achieves the technical effects of comprehensively and accurately assessing rural credit risk, providing early warning of potential risks, implementing precise and flexible interventions, and continuously improving the efficiency of risk management.

[0044] As a specific implementation of this disclosure, based on the basic solution, the acquisition of multi-source heterogeneous data related to the borrower from multiple heterogeneous data sources is further defined as follows: collecting the borrower's production and operation environment data through IoT devices; acquiring the borrower's transaction behavior data from an external commercial platform through an application programming interface; and acquiring policy, subsidy, and registration information related to the borrower from a preset data platform.

[0045] Specifically, the acquisition methods for multi-source heterogeneous data are precisely implemented: For borrowers' production and operation environment data, IoT sensors deployed in production scenarios such as farmland, greenhouses, and breeding farms, as well as agricultural machinery equipped with positioning and trajectory recording functions, are used to collect dynamic data such as soil moisture, ambient temperature, crop growth images, breeding environment parameters, and agricultural machinery operation trajectories in real time; For borrowers' transaction behavior data, application programming interfaces (APIs) that are pre-connected to e-commerce platforms, logistics service providers, and third-party payment institutions are used to obtain borrowers' agricultural input purchase details, agricultural product sales flow, logistics and distribution records, and daily consumption payment data in batches, achieving full-dimensional capture of transaction behavior; and For policies, subsidies, and registration information related to borrowers, they are obtained from pre-set government department data platforms (including the Ministry of Agriculture and Rural Affairs, social security departments, etc.), specifically covering authoritative data such as land ownership registration information, agricultural machinery purchase subsidy issuance records, new agricultural business entity registration data, and social security payment information.

[0046] By clearly defining the targeted acquisition paths for three types of key data, we have achieved real-time perception of dynamic data in the production environment, accurately captured the real trajectory of business transactions, and incorporated authoritative policy and registration data. This has further broken down data barriers in different scenarios, making the collected data more aligned with the needs of rural credit assessment. It has provided more timely, authentic, and comprehensive data support for subsequent dynamic credit rating, significantly improving the accuracy of credit profiling.

[0047] As a specific implementation of this disclosure, based on the basic scheme, the credit rating model is further defined as a model that integrates an initial static score and a dynamic trend score; wherein, the dynamic trend score is obtained by analyzing the borrower's time-series behavioral data; the generation of a dynamic credit rating that changes over time includes: standardizing multi-source heterogeneous data from multiple heterogeneous data sources to eliminate differences in dimensions; and using a weighted calculation method to integrate the processed multi-dimensional indicator data into a single credit assessment index, wherein the weights of at least some indicators are dynamically adjusted according to the time-series characteristics of the data.

[0048] Specifically, the first step involves standardizing the collected multi-source heterogeneous data (covering multiple dimensions such as borrower individual characteristics, production and operation, and external environment). Logarithmic transformation is used to eliminate dimensional differences between different indicators, avoiding assessment bias caused by excessively large data ranges. Then, based on the standardized data, an initial static scoring model is constructed using an ensemble learning algorithm to calculate a static score reflecting the borrower's basic credit level. Simultaneously, a time-series analysis model continuously learns from the borrower's time-series behavioral data (including continuous periodic business transaction data, production activity data, and repayment performance data) to capture the data's evolution over time and output a dynamic trend score representing the direction and magnitude of changes in credit status. In the final credit rating generation stage, a weighted calculation method is used to merge the initial static score and the dynamic trend score into a single credit assessment index. The weights of indicators related to business dynamics and market changes are dynamically adjusted based on their time-series characteristics. By using a preset weight decay coefficient combined with the indicator's current performance, differentiated weights are assigned to data at different time points, making the impact of recent data on the rating results more significant. Finally, a dynamic credit rating updated according to a preset period (e.g., monthly) is output.

[0049] Standardized processing ensures the comparability of multi-dimensional indicators. By integrating static and dynamic scoring and adjusting the dynamic weights of some indicators, it not only maintains a stable assessment of basic credit levels but also accurately captures the dynamic trends of credit status. This effectively solves the problem that traditional fixed-weight scoring cannot adapt to changes in the time dimension, making credit rating results more consistent with the borrower's real-time credit status and significantly improving the accuracy and dynamic adaptability of rural credit assessment.

[0050] As a specific implementation of this disclosure, based on the basic scheme, the method of comparing the dynamic credit rating and its changing trend with preset early warning rules to generate risk early warning signals of different levels is further defined, including: setting up an early warning rule library with multiple risk levels; when the decline of the dynamic credit rating or the negative slope of the changing trend exceeds the threshold of the corresponding level, triggering the risk early warning signal of the corresponding level, and identifying the key risk indicators that cause the rating change as risk triggers.

[0051] Specifically, an early warning rule base is constructed, comprising multiple risk levels (e.g., three levels: attention, warning, and circuit breaker). Each level has a pre-defined trigger threshold. The threshold settings are based on common risk evolution characteristics in rural credit scenarios, including the threshold for the decline of dynamic credit rating within a preset period (e.g., a decline of 2 levels in two consecutive months) and the threshold for the negative slope of the credit status change trend curve (e.g., a monthly rating decline rate exceeding 1 level / month). Different risk levels correspond to differentiated threshold standards, with higher levels having stricter thresholds. During the early warning determination process, the actual decline of dynamic credit rating and the negative slope of the change trend are compared in real time with the thresholds of each level in the early warning rule base. When any indicator reaches or exceeds the corresponding level threshold, a risk warning signal for that level is immediately triggered. Simultaneously, through feature importance analysis algorithms, key indicators that contribute the most to the current credit rating change are selected from multiple dimensions such as individual farmer characteristics, production and operation fundamentals, and external environment (e.g., severe weather data during the growth period of specific crops, month-on-month sharp drop in e-commerce sales, abnormal loan fund flows, etc.). These are clearly identified as risk triggers and output synchronously with the early warning signal.

[0052] By combining tiered early warning rules with precise cause identification, we can achieve tiered and refined early warning of risks, avoiding the lag or misjudgment caused by a single threshold. We can also provide clear targeting basis for subsequent intervention measures, solving the problem that traditional early warning only knows that the risk exists but does not know the root cause of the risk. This significantly improves the practicality of risk early warning and the pertinence of subsequent intervention.

[0053] As a specific implementation of this disclosure, based on the basic solution, the response to the risk warning signal and the execution of intervention measures matching the warning level are further defined, including: matching a corresponding intervention strategy from a preset strategy library according to the level of the risk warning signal and the identified risk triggers; the intervention strategy includes at least one of sending a reminder message to the borrower, adjusting the repayment plan, or triggering a manual verification process.

[0054] Specifically, the pre-set strategy library stores a set of intervention strategies corresponding to different warning levels (such as attention, warning, and circuit breaker) and various risk triggers (such as natural disasters, business downturns, and abnormal cash flow). The system first analyzes the level information and identified key risk triggers in the risk warning signal, and then uses a strategy matching algorithm to accurately retrieve the appropriate intervention plan from the library. Sending reminder information to borrowers is specifically manifested in pushing targeted content via SMS platform, such as risk prevention suggestions, agricultural technical guidance, or market information. Adjusting repayment plans provides flexible adjustment options such as deferred repayment and installment repayment based on the borrower's actual business situation and the degree of risk impact. The system interface is linked with the bank's core business system to complete the rule configuration. When the manual verification process is triggered, the system will generate a task work order containing borrower information, risk triggers, and verification focus on the account manager's workbench, and simultaneously push risk warning details to guide the account manager to conduct on-site visits or telephone follow-ups, and at least one of the above intervention strategies or a combination thereof must be selected for execution.

[0055] By employing a dual matching logic of "level + incentive," intervention strategies become more targeted, avoiding the blindness of single-minded intervention. At the same time, diversified intervention methods take into account both risk control and the actual needs of borrowers. This achieves flexible intervention at the initial stage of risk while ensuring the handling of high-risk scenarios through manual verification, effectively improving the execution of intervention measures, helping to reduce the non-performing loan rate and optimize the borrower service experience.

[0056] As a specific implementation of this disclosure, based on the basic scheme, the method further specifies the collection of data on the effectiveness of the intervention measures and the feedback of the data on the effectiveness to the credit rating model for optimization. This includes: inputting the results of the intervention measures that can alleviate risk factors as a positive feedback feature into the credit rating model; and adaptively adjusting the relevant weight parameters in the model based on the positive feedback feature to optimize subsequent rating calculations.

[0057] Specifically, the process for collecting data on the effectiveness of intervention measures and optimizing the model is as follows: First, key outcome data after the implementation of intervention measures are systematically collected, with a focus on screening positive outcomes that can directly alleviate risk triggers. Examples include borrowers completing agricultural insurance for natural disaster risks, achieving a recovery in production and sales rates after adjusting production strategies according to business guidance, and fulfilling repayment obligations according to negotiated plans. These outcomes are quantified into standardized positive feedback features (such as "risk trigger mitigation status indicators" and "intervention response effectiveness scores"). Subsequently, these positive feedback features are input into the credit rating model through the model interface. Based on preset parameter adjustment rules, the model identifies assessment indicators related to the risk trigger (such as secondary indicators corresponding to natural disaster response capabilities and business risk resistance capabilities). The weight parameters of these indicators in the rating calculation are adaptively adjusted so that for borrowers who have taken effective intervention measures and whose risks have been mitigated, the influence of relevant indicators in subsequent dynamic credit rating calculations is more in line with the actual risk situation, avoiding excessive rating downgrades due to historical risk signals.

[0058] By clearly extracting positive feedback features and making targeted adjustments to model weights, the credit rating model can accurately learn the mitigating effect of effective intervention measures on risk, avoiding the limitations of traditional models that "only assess risk and not improvement." This significantly improves the model's dynamic adaptability to changes in borrowers' risk status, making subsequent credit rating results more consistent with reality and further strengthening the practical value of the "intervention-feedback-optimization" closed loop.

[0059] It should be noted that the embodiments of this disclosure may include multiple steps. For ease of description, these steps are numbered, but these numbers are not a limitation on the execution time slots or execution order between the steps; these steps can be implemented in any order, and the embodiments of this disclosure do not limit this.

[0060] Corresponding to the aforementioned rural credit risk early warning method, this disclosure also proposes a rural credit risk early warning device. Since the device embodiments of this disclosure correspond to the aforementioned method embodiments, details not disclosed in the device embodiments can be referred to the aforementioned method embodiments, and will not be repeated here.

[0061] Figure 2 This is a schematic diagram of the structure of a rural credit risk early warning device provided in an embodiment of the present disclosure, as shown below. Figure 2 As shown, it includes: Acquisition unit 21 is used to acquire multi-source heterogeneous data related to the borrower from multiple heterogeneous data sources; The calculation unit 22 is used to calculate, based on the acquired multi-source heterogeneous data, through a credit rating model to generate a dynamic credit rating that changes over time. The comparison unit 23 is used to compare the dynamic credit rating and its changing trend with the preset early warning rules to generate risk early warning signals of different levels. Execution unit 24 is used to respond to the risk warning signal and execute intervention measures that match the warning level; The collection unit 25 is used to collect data on the implementation effect of the intervention measures and feed the data back to the credit rating model to optimize the credit rating model.

[0062] The rural credit risk early warning device disclosed herein acquires multi-source heterogeneous data related to borrowers, generates a dynamic credit rating that changes over time based on this data, and generates risk early warning signals of different levels by combining the dynamic credit rating and its changing trend. It responds to the early warning signals by implementing matching intervention measures and feeds back the implementation effect data to the credit rating model to achieve model optimization. Therefore, it can solve the problems of risk identification lag, insufficient intervention accuracy, and information asymmetry caused by the single dimension of credit assessment, static risk assessment, disconnect between early warning and intervention and lack of self-optimization ability in the existing technology. It achieves the technical effects of comprehensively and accurately assessing rural credit risk, providing early warning of potential risks, implementing precise and flexible intervention, and continuously improving the efficiency of risk management.

[0063] Furthermore, in one possible implementation of this embodiment, the acquisition unit 21 is also used for: Data on the borrower's production and operating environment is collected through IoT devices; The borrower's transaction behavior data is obtained from an external commercial platform via an application programming interface; Obtain relevant policies, subsidies, and registration information for the borrower from a pre-set data platform.

[0064] Furthermore, in one possible implementation of this embodiment, the credit rating model is a model that integrates an initial static score and a dynamic trend score; wherein, the dynamic trend score is obtained by analyzing the borrower's time-series behavioral data; The computing unit 22 is also used for: Standardize multi-source heterogeneous data from multiple heterogeneous data sources to eliminate dimensional differences; A weighted calculation method is used to integrate the processed multi-dimensional indicator data into a single credit assessment index, wherein the weights of at least some indicators are dynamically adjusted according to the time series characteristics of the data.

[0065] Furthermore, in one possible implementation of this embodiment, the comparison unit 23 is also used for: Set up a warning rule base with multiple risk levels; When the decline in the dynamic credit rating or the negative slope of the change trend exceeds the threshold of the corresponding level, a risk warning signal for the corresponding level is triggered, and key risk indicators that lead to the rating change are identified as risk triggers.

[0066] Furthermore, in one possible implementation of this embodiment, the execution unit 24 is also used for: Based on the level of the risk warning signal and the identified risk triggers, the corresponding intervention strategy is matched from the preset strategy library; The intervention strategies include at least one of sending reminder messages to borrowers, adjusting repayment plans, or triggering a manual verification process.

[0067] Furthermore, in one possible implementation of this embodiment, the collecting unit 25 is also used for: The results of interventions that can mitigate risk factors are used as a positive feedback feature and input into the credit rating model. Based on the positive feedback characteristics, the relevant weight parameters in the model are adaptively adjusted to optimize subsequent rating calculations.

[0068] It should be noted that the foregoing explanation of the method embodiments also applies to the apparatus of this embodiment, and the principle is the same, so it is not limited in this embodiment.

[0069] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0070] Figure 3 A schematic block diagram of an example electronic device 300 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0071] like Figure 3As shown, the electronic device 300 includes a computing unit 301, which can perform various appropriate actions and processes based on a computer program stored in ROM (Read-Only Memory) 302 or a computer program loaded from storage unit 308 into RAM (Random Access Memory) 303. The RAM 303 may also store various programs and data required for the operation of the electronic device 300. The computing unit 301, ROM 302, and RAM 303 are interconnected via a bus 304. An I / O (Input / Output) interface 305 is also connected to the bus 304.

[0072] Multiple components in electronic device 300 are connected to I / O interface 305, including: input unit 306, such as keyboard, mouse, etc.; output unit 307, such as various types of displays, speakers, etc.; storage unit 308, such as disk, optical disk, etc.; and communication unit 309, such as network card, modem, wireless transceiver, etc. Communication unit 309 allows electronic device 300 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0073] The computing unit 301 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 301 include, but are not limited to, CPUs (Central Processing Units), GPUs (Graphics Processing Units), various special-purpose AI (Artificial Intelligence) computing chips, various computing units running machine learning model algorithms, DSPs (Digital Signal Processors), and any suitable processor, controller, microcontroller, etc. The computing unit 301 performs the various methods and processes described above, such as the rural credit risk early warning method. For example, in some embodiments, the rural credit risk early warning method can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as storage unit 308. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 300 via ROM 302 and / or communication unit 309. When the computer program is loaded into RAM 303 and executed by the computing unit 301, one or more steps of the methods described above can be performed. Alternatively, in other embodiments, the computing unit 301 may be configured to perform the aforementioned rural credit risk early warning method by any other suitable means (e.g., by means of firmware).

[0074] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, FPGAs (Field Programmable Gate Arrays), ASICs (Application-Specific Integrated Circuits), ASSPs (Application-Specific Standard Products), SOCs (System-on-Chips), CPLDs (Complex Programmable Logic Devices), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0075] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0076] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, RAM, ROM, EPROM (Electrically Programmable Read-Only Memory) or flash memory, optical fiber, CD-ROM (Compact Disc Read-Only Memory), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0077] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (Cathode-Ray Tube) or LCD (Liquid Crystal Display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0078] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include LANs (Local Area Networks), WANs (Wide Area Networks), the Internet, and blockchain networks.

[0079] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. A server can be a cloud server, also known as a cloud computing server or cloud host, a hosting product within the cloud computing service system that addresses the shortcomings of traditional physical hosts and VPS (Virtual Private Server) services, such as high management difficulty and weak business scalability. Servers can also be servers for distributed systems or servers incorporating blockchain technology.

[0080] It's important to note that artificial intelligence (AI) is the study of enabling computers to simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, and planning). It encompasses both hardware and software technologies. AI hardware technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, and big data processing. AI software technologies primarily include computer vision, speech recognition, natural language processing, machine learning / deep learning, big data processing, and knowledge graph technologies.

[0081] The various numerical designations such as "first," "second," etc., used in this disclosure are merely for ease of description and are not intended to limit the scope of the embodiments of this disclosure, nor do they indicate a sequential order.

[0082] At least one of the features described in this disclosure can also be described as one or more, and multiple features can be two, three, four or more, and this disclosure does not impose any limitations. In the embodiments of this disclosure, for a technical feature, the technical features in that technical feature are distinguished by "first", "second", "third", "A", "B", "C" and "D", etc., and there is no sequential order or size order among the technical features described by "first", "second", "third", "A", "B", "C" and "D".

[0083] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0084] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for early warning of rural credit risk, characterized in that, include: Acquire multi-source heterogeneous data related to borrowers from multiple heterogeneous data sources; Based on the acquired multi-source heterogeneous data, a dynamic credit rating that changes over time is generated through a credit rating model. Based on the dynamic credit rating and its changing trend, it is compared with the preset early warning rules to generate risk warning signals of different levels; In response to the aforementioned risk warning signal, implement intervention measures commensurate with the warning level; Collect data on the effectiveness of the intervention measures and feed the data back to the credit rating model to optimize the credit rating model.

2. The method according to claim 1, characterized in that, The acquisition of multi-source heterogeneous data related to the borrower from multiple heterogeneous data sources includes: Data on the borrower's production and operating environment is collected through IoT devices; The borrower's transaction behavior data is obtained from an external commercial platform via an application programming interface; Obtain relevant policies, subsidies, and registration information for the borrower from a pre-set data platform.

3. The method according to claim 1, characterized in that, The credit rating model is a model that integrates an initial static score and a dynamic trend score; wherein, the dynamic trend score is obtained by analyzing the borrower's time-series behavioral data; The generation of a dynamic credit rating that changes over time includes: Standardize multi-source heterogeneous data from multiple heterogeneous data sources to eliminate dimensional differences; A weighted calculation method is used to integrate the processed multi-dimensional indicator data into a single credit assessment index, wherein the weights of at least some indicators are dynamically adjusted according to the time series characteristics of the data.

4. The method according to claim 1, characterized in that, The step of comparing the dynamic credit rating and its changing trend with preset early warning rules to generate risk warning signals of different levels includes: Set up a warning rule base with multiple risk levels; When the decline in the dynamic credit rating or the negative slope of the change trend exceeds the threshold of the corresponding level, a risk warning signal for the corresponding level is triggered, and key risk indicators that lead to the rating change are identified as risk triggers.

5. The method according to claim 1, characterized in that, The response to the risk warning signal, implementing intervention measures commensurate with the warning level, includes: Based on the level of the risk warning signal and the identified risk triggers, the corresponding intervention strategy is matched from the preset strategy library; The intervention strategies include at least one of sending reminder messages to borrowers, adjusting repayment plans, or triggering a manual verification process.

6. The method according to claim 1, characterized in that, The process of collecting data on the effectiveness of the intervention measures and feeding this data back to the credit rating model for optimization includes: The results of interventions that can mitigate risk factors are used as a positive feedback feature and input into the credit rating model. Based on the positive feedback characteristics, the relevant weight parameters in the model are adaptively adjusted to optimize subsequent rating calculations.

7. A rural credit risk early warning device, characterized in that, include: The acquisition unit is used to acquire multi-source heterogeneous data related to the borrower from multiple heterogeneous data sources. The calculation unit is used to calculate, based on the acquired multi-source heterogeneous data, using a credit rating model to generate a dynamic credit rating that changes over time. The comparison unit is used to compare the dynamic credit rating and its changing trend with the preset early warning rules to generate risk warning signals of different levels. An execution unit is used to respond to the risk warning signal and execute intervention measures that match the warning level. The collection unit is used to collect data on the effectiveness of the intervention measures and feed the data back to the credit rating model to optimize the credit rating model.

8. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-6.

9. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-6.

10. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method according to any one of claims 1-6.

Images