Consumption finance intelligent risk control method and system, computer equipment and storage medium

Abstract

The invention provides an intelligent risk control method and system for consumer finance, computer equipment and a storage medium, and the method comprises the steps: deploying a block chain alliance chain and supply chain financial platform, a credit investigation mechanism and an asset auditing platform, building an encrypted data channel, and setting daily update data; after the trading day ends, obtaining core enterprise supply chain bill data to construct an enterprise supply chain atlas, and obtaining associated personnel behaviors and biological characteristic data to construct a personnel supply chain atlas; integrating an enterprise credit report of a credit investigation institution, an enterprise asset proof of an asset auditing platform and credit investigation and asset files of associated personnel, and constructing an enterprise and personnel financial asset atlas; updating the enterprise atlas based on the personnel financial asset atlas, and evaluating the consumer financial risk level in combination with the supply chain data; and configuring a next-day supervision process according to the risk level, and implementing dynamic supervision on the next trading day. By adopting the method, reasonable allocation of risk control resources can be realized, and meanwhile, the pertinence and effectiveness of consumption financial risk control on enterprises are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent financial management, and more specifically, to an intelligent risk control method, system, computer equipment, and storage medium for consumer finance. Background Technology

[0002] Against the backdrop of the rapid development of the consumer finance industry, traditional risk control models, which rely excessively on single-dimensional corporate financial data and credit records, struggle to effectively identify potential risks in complex transaction scenarios. With the intensifying risk transmission effects of core enterprises in the supply chain finance ecosystem and the gradual application of blockchain technology in the financial sector, how to deeply integrate multi-source heterogeneous data to achieve dynamic risk management across the entire supply chain has become an industry challenge. While some existing solutions attempt to integrate multi-source data through big data analytics or utilize blockchain for data storage and sharing of invoices, they still suffer from problems such as limited data integration dimensions and insufficient analysis of risk transmission paths, failing to systematically correlate data on corporate supply chain relationships, personnel behavior characteristics, and financial asset status.

[0003] Among the closest existing technologies, some solutions calculate credit scores by collecting data such as corporate financial statements and bank statements, or identify fraud risks by analyzing user operation patterns. However, they generally lack the ability to explore the inherent correlations between data, and cannot accurately assess the risk transmission effect of core enterprises in the supply chain network. In addition, blockchain-based risk control solutions mostly remain at the level of data storage, failing to fully utilize the characteristics of smart contracts and distributed consensus to automate and coordinate risk control processes, resulting in blind spots in risk identification and delayed responses.

[0004] The shortcomings of these existing technologies make them difficult to adapt to the complex and ever-changing risk profiles in consumer finance scenarios: on the one hand, the isolated analysis of single-dimensional data makes risk assessment static and unable to capture the dynamic changes in business operations and market environment in real time, which in turn prevents intelligent risk control in consumer finance from achieving optimal results and reduces the effectiveness of financial risk control; on the other hand, the lack of differentiated regulatory strategies based on the risk characteristics of enterprises increases the financing costs of low-risk enterprises and fails to effectively constrain high-risk enterprises, which in turn leads to an unreasonable allocation of risk control resources. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a consumer finance intelligent risk control method, system, computer equipment and storage medium to achieve reasonable allocation of risk control resources, while improving the pertinence and effectiveness of consumer finance risk control for enterprises.

[0006] In a first aspect, embodiments of this application provide a smart risk control method for consumer finance, the method comprising:

[0007] Deploy a blockchain consortium blockchain and establish encrypted data channels with supply chain finance platforms, credit reporting agencies, and asset verification platforms. At the same time, set the data update frequency of the blockchain consortium blockchain to be updated once per transaction day.

[0008] After the end of each trading day, the supply chain bill data of the core enterprise uploaded by the supply chain finance platform is obtained through the blockchain consortium chain, and the enterprise supply chain map of the core enterprise is constructed based on the supply chain bill data.

[0009] The blockchain consortium blockchain is used to obtain user behavior data and biometric data of the core enterprise’s related personnel uploaded to the supply chain finance platform. Based on the user behavior data and biometric data of each related person, a personnel supply chain map of each related person is constructed.

[0010] The core enterprise's corporate credit report and the personnel credit reports of each associated person are obtained through the blockchain consortium chain; the core enterprise's corporate asset certification documents and the personnel asset certification documents of each associated person are obtained through the asset verification platform.

[0011] The core enterprise's financial asset map is constructed based on the enterprise's credit report and asset certification documents, and the related personnel's financial asset map is constructed based on the related personnel's credit reports and asset certification documents.

[0012] The enterprise financial asset map is updated based on the personnel financial asset map of each related person, and the consumer finance risk level of the core enterprise is determined based on the updated enterprise financial asset map and the enterprise supply chain map.

[0013] Based on the aforementioned consumer finance risk level, the consumer finance regulatory process for the next trading day is configured for the core enterprise, and on the next trading day, consumer finance supervision of the core enterprise is carried out based on the consumer finance regulatory process of that trading day.

[0014] Optionally, constructing the enterprise supply chain map of the core enterprise based on the supply chain invoice data includes:

[0015] The supply chain bill data is parsed to obtain basic bill information, circulation records, and payment data;

[0016] Extract the enterprise entity, supplier entity, and bill entity from the basic information of the bill, the circulation record, and the payment data, as well as the opening relationship, circulation relationship, and payment relationship between the entities;

[0017] A graph model is constructed using the triplet pattern, with enterprise entities, supplier entities, and bill entities as nodes, and opening relationships, circulation relationships, and payment relationships as edges. The attributes of the edges include relationship type and timestamp, which are stored in the Neo4j graph database to obtain the initial supply chain graph.

[0018] The enterprise supply chain map is obtained by using a graph neural network to complete the missing relationships in the initial supply chain map.

[0019] Optionally, the step of constructing a personnel supply chain map for each associated person based on their user behavior data and biometric data includes:

[0020] The user behavior data is parsed to extract transaction operation information, and the biometric data is encrypted using SHA-256 to generate a feature hash value.

[0021] The identity entities of each associated person are determined based on the feature hash value, and the interaction relationships between each associated person are determined based on the transaction operation information and the identity entities of each associated person.

[0022] A directed graph model is used to construct nodes and edges. Each associated person entity is used as a node, its identity entity is used as a node attribute, and the interaction relationship between each associated person is used as a directed edge to construct an initial personnel supply chain graph.

[0023] An LSTM neural network is used to analyze the time series of user behavior to identify abnormal edges in the initial personnel supply chain map. After correcting the abnormal edges using a graph neural network, the personnel supply chain map is obtained.

[0024] Optionally, constructing the corporate financial asset map of the core enterprise based on the corporate credit report and the corporate asset certification documents includes:

[0025] The enterprise credit report and the enterprise asset certification documents are parsed to extract the core enterprise's historical behavior data and enterprise asset status data, respectively.

[0026] From the enterprise's historical behavior data and the enterprise's asset status data, identify core entities, including core enterprises and cooperative institutions, as well as asset entities, including fixed assets and current assets, and establish business relationships between core enterprises and cooperative institutions, as well as ownership relationships between core enterprises and different asset entities.

[0027] An attribute graph model is used to construct nodes and edges, with core entities and asset entities as nodes. Credit records and performance capabilities in the historical behavior data are used as attributes of core entity nodes, and asset composition and value assessment in the asset status data are used as attributes of asset entity nodes. Business relationships and ownership relationships are used as edges, and weights are set to construct an initial corporate financial asset graph.

[0028] The authenticity of the data in the initial corporate financial asset graph is verified by using blockchain notarization technology, and the implicit relationships in the initial corporate financial asset graph are inferred by using a knowledge graph completion algorithm to construct the corporate financial asset graph.

[0029] Optionally, the construction of the financial asset map of each associated person based on their credit reports and asset certificates includes:

[0030] The personnel credit reports and personnel asset certificates are analyzed to extract the historical credit data and asset holding data of each related person.

[0031] From the historical credit data of the personnel and the asset holding data of the personnel, identify the identity entities and asset entities of related personnel, establish the direct holding relationship between the identity entities and asset entities of related personnel, as well as the indirect relationship between related personnel and other related personnel;

[0032] A heterogeneous graph model is used to construct nodes and edges, with related personnel identity entities and asset entities as different types of nodes. Default records and credit scores in the historical credit data are used as attributes of related personnel nodes, and asset size and liquidity in the asset holding data are used as attributes of asset nodes. Direct holding relationships and indirect relationships are used as different types of edges, and weights are set to construct an initial personnel financial asset graph.

[0033] The authenticity of the data in the initial personnel financial asset map is verified by using blockchain notarization technology, and the implicit relationships in the initial personnel financial asset map are completed by using knowledge reasoning algorithms to construct the personnel financial asset map.

[0034] Optionally, the enterprise financial asset map is updated based on the personnel financial asset map of each related person, and the consumer finance risk level of the core enterprise is determined based on the updated enterprise financial asset map and the enterprise supply chain map, including:

[0035] Extract the relationship between the related personnel and the core enterprise from the personnel financial asset map of each related person, map the relationship to the corresponding node of the enterprise financial asset map, and update the credit attribute and asset association attribute of the related personnel of the core enterprise node.

[0036] A graph matching algorithm is used to identify common nodes in the financial asset graphs of individuals and enterprises, and the node attributes of the enterprise financial asset graph are dynamically updated based on the asset change records of related individuals.

[0037] The updated corporate financial asset map and corporate supply chain map are fused together. Centered on the core enterprise node, the transaction relationships in the supply chain map and the credit asset relationships in the financial asset map are integrated to form a multi-dimensional risk network.

[0038] Based on the fused multidimensional risk network, the analytic hierarchy process (AHP) is used to determine each risk factor and its weight. According to the actual value and weight of each risk factor, a comprehensive risk score is calculated using a preset risk assessment function. The comprehensive risk score is then compared with a preset risk grading standard to determine the consumer finance risk level.

[0039] Optionally, configuring the consumer finance regulatory process for the core enterprise based on the consumer finance risk level for the next trading day includes:

[0040] The corresponding regulatory intensity coefficient is determined based on the aforementioned consumer finance risk level, wherein the regulatory intensity coefficient is positively correlated with the risk level;

[0041] The consumer finance regulatory process for the core enterprise on the next trading day is determined based on the regulatory intensity coefficient.

[0042] Secondly, embodiments of this application provide a consumer finance intelligent risk control system, the system comprising:

[0043] The blockchain consortium blockchain deployment module is used to deploy the blockchain consortium blockchain, establish encrypted data channels with supply chain finance platforms, credit reporting agencies, and asset verification platforms, and set the data update frequency of the blockchain consortium blockchain to be updated once per transaction day.

[0044] The enterprise supply chain map construction module is used to obtain the supply chain invoice data of the core enterprise uploaded by the supply chain finance platform through the blockchain consortium chain after the end of each trading day, and construct the enterprise supply chain map of the core enterprise based on the supply chain invoice data.

[0045] The personnel supply chain map construction module is used to obtain user behavior data and biometric data of the core enterprise's related personnel uploaded by the supply chain finance platform through the blockchain consortium chain, and construct personnel supply chain maps of each related personnel based on the user behavior data and biometric data of each related personnel.

[0046] The enterprise personnel report document acquisition module is used to acquire the enterprise credit report of the core enterprise and the personnel credit reports of each related person uploaded by the credit reporting agency through the blockchain consortium chain, and to acquire the enterprise asset certificate documents of the core enterprise and the personnel asset certificate documents of each related person uploaded by the asset verification platform.

[0047] The financial asset mapping module is used to construct the corporate financial asset mapping of the core enterprise based on the enterprise credit report and the enterprise asset certification documents, and to construct the personnel financial asset mapping of each related person based on the personnel credit reports and personnel asset certification documents of each related person.

[0048] The financial risk level determination module is used to update the enterprise's financial asset map based on the personnel financial asset map of each related person, and to determine the consumer finance risk level of the core enterprise based on the updated enterprise financial asset map and the enterprise supply chain map.

[0049] The consumer finance supervision module is used to configure the consumer finance supervision process for the core enterprise for the next trading day based on the consumer finance risk level, and to conduct consumer finance supervision of the core enterprise based on the consumer finance supervision process of the next trading day.

[0050] Optionally, constructing the enterprise supply chain map of the core enterprise based on the supply chain invoice data includes:

[0051] The supply chain bill data is parsed to obtain basic bill information, circulation records, and payment data;

[0052] Extract the enterprise entity, supplier entity, and bill entity from the basic information of the bill, the circulation record, and the payment data, as well as the opening relationship, circulation relationship, and payment relationship between the entities;

[0053] A graph model is constructed using the triplet pattern, with enterprise entities, supplier entities, and bill entities as nodes, and opening relationships, circulation relationships, and payment relationships as edges. The attributes of the edges include relationship type and timestamp, which are stored in the Neo4j graph database to obtain the initial supply chain graph.

[0054] The enterprise supply chain map is obtained by using a graph neural network to complete the missing relationships in the initial supply chain map.

[0055] Optionally, the step of constructing a personnel supply chain map for each associated person based on their user behavior data and biometric data includes:

[0056] The user behavior data is parsed to extract transaction operation information, and the biometric data is encrypted using SHA-256 to generate a feature hash value.

[0057] The identity entities of each associated person are determined based on the feature hash value, and the interaction relationships between each associated person are determined based on the transaction operation information and the identity entities of each associated person.

[0058] A directed graph model is used to construct nodes and edges. Each associated person entity is used as a node, its identity entity is used as a node attribute, and the interaction relationship between each associated person is used as a directed edge to construct an initial personnel supply chain graph.

[0059] An LSTM neural network is used to analyze the time series of user behavior to identify abnormal edges in the initial personnel supply chain map. After correcting the abnormal edges using a graph neural network, the personnel supply chain map is obtained.

[0060] Optionally, constructing the corporate financial asset map of the core enterprise based on the corporate credit report and the corporate asset certification documents includes:

[0061] The enterprise credit report and the enterprise asset certification documents are parsed to extract the core enterprise's historical behavior data and enterprise asset status data, respectively.

[0062] From the enterprise's historical behavior data and the enterprise's asset status data, identify core entities, including core enterprises and cooperative institutions, as well as asset entities, including fixed assets and current assets, and establish business relationships between core enterprises and cooperative institutions, as well as ownership relationships between core enterprises and different asset entities.

[0063] An attribute graph model is used to construct nodes and edges, with core entities and asset entities as nodes. Credit records and performance capabilities in the historical behavior data are used as attributes of core entity nodes, and asset composition and value assessment in the asset status data are used as attributes of asset entity nodes. Business relationships and ownership relationships are used as edges, and weights are set to construct an initial corporate financial asset graph.

[0064] The authenticity of the data in the initial corporate financial asset graph is verified by using blockchain notarization technology, and the implicit relationships in the initial corporate financial asset graph are inferred by using a knowledge graph completion algorithm to construct the corporate financial asset graph.

[0065] Optionally, the construction of the financial asset map of each associated person based on their credit reports and asset certificates includes:

[0066] The personnel credit reports and personnel asset certificates are analyzed to extract the historical credit data and asset holding data of each related person.

[0067] From the historical credit data of the personnel and the asset holding data of the personnel, identify the identity entities and asset entities of related personnel, establish the direct holding relationship between the identity entities and asset entities of related personnel, as well as the indirect relationship between related personnel and other related personnel;

[0068] A heterogeneous graph model is used to construct nodes and edges, with related personnel identity entities and asset entities as different types of nodes. Default records and credit scores in the historical credit data are used as attributes of related personnel nodes, and asset size and liquidity in the asset holding data are used as attributes of asset nodes. Direct holding relationships and indirect relationships are used as different types of edges, and weights are set to construct an initial personnel financial asset graph.

[0069] The authenticity of the data in the initial personnel financial asset map is verified by using blockchain notarization technology, and the implicit relationships in the initial personnel financial asset map are completed by using knowledge reasoning algorithms to construct the personnel financial asset map.

[0070] Optionally, the enterprise financial asset map is updated based on the personnel financial asset map of each related person, and the consumer finance risk level of the core enterprise is determined based on the updated enterprise financial asset map and the enterprise supply chain map, including:

[0071] Extract the relationship between the related personnel and the core enterprise from the personnel financial asset map of each related person, map the relationship to the corresponding node of the enterprise financial asset map, and update the credit attribute and asset association attribute of the related personnel of the core enterprise node.

[0072] A graph matching algorithm is used to identify common nodes in the financial asset graphs of individuals and enterprises, and the node attributes of the enterprise financial asset graph are dynamically updated based on the asset change records of related individuals.

[0073] The updated corporate financial asset map and corporate supply chain map are fused together. Centered on the core enterprise node, the transaction relationships in the supply chain map and the credit asset relationships in the financial asset map are integrated to form a multi-dimensional risk network.

[0074] Based on the fused multidimensional risk network, the analytic hierarchy process (AHP) is used to determine each risk factor and its weight. According to the actual value and weight of each risk factor, a comprehensive risk score is calculated using a preset risk assessment function. The comprehensive risk score is then compared with a preset risk grading standard to determine the consumer finance risk level.

[0075] Optionally, configuring the consumer finance regulatory process for the core enterprise based on the consumer finance risk level for the next trading day includes:

[0076] The corresponding regulatory intensity coefficient is determined based on the aforementioned consumer finance risk level, wherein the regulatory intensity coefficient is positively correlated with the risk level;

[0077] The consumer finance regulatory process for the core enterprise on the next trading day is determined based on the regulatory intensity coefficient.

[0078] Thirdly, embodiments of this application provide a computer device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, the steps of the consumer finance intelligent risk control method described in any of the optional embodiments of the first aspect are performed.

[0079] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the consumer finance intelligent risk control method described in any of the optional embodiments of the first aspect.

[0080] The technical solution provided in this application includes, but is not limited to, the following beneficial effects:

[0081] A blockchain consortium blockchain is deployed, establishing encrypted data channels with supply chain finance platforms, credit reporting agencies, and asset verification platforms. The blockchain consortium blockchain is configured to update data once per transaction day. This step, through the distributed storage and encrypted consensus mechanism of blockchain, ensures the security and reliability of data transmission and storage across multiple platforms, preventing data tampering or leakage. Daily scheduled updates ensure that the data remains timely without overburdening the system with excessively frequent updates, providing a reliable data foundation for subsequent risk control analysis.

[0082] At the end of each trading day, the supply chain invoice data of the core enterprise uploaded by the supply chain finance platform is obtained through the blockchain consortium chain. Based on the supply chain invoice data, a corporate supply chain graph of the core enterprise is constructed. This process transforms the core enterprise's supply chain invoice data into a visualized graph structure, which can intuitively present the transaction relationship network between the enterprise and its upstream and downstream partners. Through graph analysis, potential transaction anomalies or risk transmission paths can be discovered, helping risk control personnel quickly locate risk nodes and enhance their ability to perceive supply chain transaction risks.

[0083] The blockchain consortium blockchain acquires user behavior data and biometric data of related personnel of the core enterprise uploaded to the supply chain finance platform. Based on the user behavior data and biometric data of each related person, a personnel supply chain map is constructed. This step starts from the dimensions of personnel behavior and biometrics to build a network of interaction relationships between personnel. Combined with time series analysis, it can identify abnormal operating behaviors, extending the scope of risk control from the enterprise level to related personnel, filling the gap in the insufficient consideration of personnel factors in traditional risk control, and realizing multi-dimensional risk monitoring.

[0084] By acquiring the core enterprise's credit report and the personnel credit reports of each associated individual uploaded by the credit reporting agency through the blockchain consortium blockchain, and by acquiring the core enterprise's asset verification documents and the personnel asset verification documents of each associated individual uploaded by the asset verification platform, this step achieves comprehensive collection of enterprise and personnel credit and asset data. Blockchain notarization ensures the authenticity and reliability of the data, providing a rich and accurate information source for the subsequent construction of a financial asset map, enabling risk control analysis to be conducted from a more comprehensive perspective.

[0085] Based on the enterprise's credit report and asset verification documents, a financial asset map of the core enterprise is constructed. Similarly, a financial asset map of each related individual is constructed based on their credit reports and asset verification documents. This financial asset map presents the credit status and asset relationships of enterprises and individuals in a structured manner. By integrating multi-faceted attribute information through attribute graphs and heterogeneous graph models, it enables in-depth analysis of the asset health and credit risk of enterprises and individuals, providing a quantitative basis for risk assessment.

[0086] The enterprise's financial asset map is updated based on the personnel financial asset maps of each related individual. Based on the updated enterprise financial asset map and the enterprise supply chain map, the consumer finance risk level of the core enterprise is determined. This step, through the coordinated updating of the related personnel and enterprise maps, transmits personnel risk to the enterprise level. Combined with the supply chain map, it achieves multi-map fusion analysis, comprehensively assessing the enterprise's consumer finance risk level. This overcomes the limitations of traditional single-dimensional assessments, making risk level determination more accurate and scientific.

[0087] Based on the aforementioned consumer finance risk level, the consumer finance regulatory process for the next trading day is configured for the core enterprise. On the next trading day, consumer finance supervision of the core enterprise is conducted based on the consumer finance regulatory process of that trading day. Dynamically configuring the regulatory process according to the risk level enables the implementation of differentiated control strategies for enterprises with different risk levels. This strengthens monitoring and restrictions on high-risk enterprises and reduces unnecessary regulatory intervention for low-risk enterprises, achieving a rational allocation of risk control resources and improving the targeting and effectiveness of consumer finance risk control for enterprises.

[0088] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0089] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0090] Figure 1 The flowchart of a consumer finance intelligent risk control method provided in Embodiment 1 of the present invention is shown;

[0091] Figure 2 The flowchart of a neural network model training method provided in Embodiment 1 of the present invention is shown;

[0092] Figure 3 The flowchart of a method for determining a target neural network model provided in Embodiment 1 of the present invention is shown;

[0093] Figure 4 The flowchart of a method for constructing a corporate financial asset map provided in Embodiment 1 of the present invention is shown;

[0094] Figure 5 The flowchart of a method for constructing a personnel financial asset map provided in Embodiment 1 of the present invention is shown;

[0095] Figure 6 The flowchart of a consumer finance risk rating method provided in Embodiment 1 of the present invention is shown;

[0096] Figure 7 This diagram shows a flowchart of a consumer finance regulatory process configuration method provided in Embodiment 1 of the present invention;

[0097] Figure 8 This diagram illustrates the structure of a consumer finance intelligent risk control system provided in Embodiment 2 of the present invention.

[0098] Figure 9 A schematic diagram of the structure of a computer device provided in Embodiment 3 of the present invention is shown. Detailed Implementation

[0099] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0100] Example 1

[0101] To facilitate understanding of this application, the following is combined with... Figure 1 The flowchart of the intelligent risk control method for consumer finance provided in Embodiment 1 of the present invention describes the content of Embodiment 1 in detail.

[0102] See Figure 1 As shown, Figure 1 The flowchart of a consumer finance intelligent risk control method provided in Embodiment 1 of the present invention is shown, wherein the method includes steps S101 to S107:

[0103] S101: Deploy a blockchain consortium chain to establish encrypted data channels with supply chain finance platforms, credit reporting agencies, and asset verification platforms, and set the data update frequency of the blockchain consortium chain to be once per transaction day.

[0104] Specifically, a consortium blockchain network is built using the Hyperledger Fabric framework. Nodes include a supply chain finance platform (data provider), credit reporting agencies (credit data source), an asset verification platform (asset verifier), and regulatory nodes (data auditors). A secure communication channel between nodes is established using an asymmetric encryption algorithm (RSA-2048), and data updates are scheduled to be fully synchronized within one hour of the end of each trading day. Blockchain notarization technology (Merkle tree structure) ensures data immutability, with each data block containing a timestamp and digital signature. For example, supply chain invoice data must be verified by the core enterprise's private key signature before being uploaded to the blockchain. The blockchain network automatically triggers the data verification process through smart contracts, and invalid data is automatically filtered out.

[0105] S102: After the end of each trading day, obtain the supply chain invoice data of the core enterprise uploaded by the supply chain finance platform through the blockchain consortium chain, and construct the enterprise supply chain map of the core enterprise based on the supply chain invoice data.

[0106] Specifically, supply chain invoice data (including fields such as issuer, payee, amount, and due date) obtained from the blockchain is parsed using NLP to extract entities (enterprises, suppliers, and invoices) and relationships (issuance, circulation, and payment). This data is stored in the Neo4j graph database. After constructing an initial graph, implicit relationships (such as undocumented second-tier suppliers) are completed using a graph neural network (GNN). For example, the invoice circulation records of a core enterprise can be traced back to five levels of suppliers. The dynamic graph update mechanism is based on transaction frequency (e.g., daily new transactions trigger incremental graph updates).

[0107] S103: Obtain user behavior data and biometric data of the core enterprise's related personnel uploaded to the supply chain finance platform through the blockchain consortium chain, and construct a personnel supply chain map of each related person based on the user behavior data and biometric data of each related person.

[0108] Specifically, user behavior data (login IP, transaction time) is anonymized and input into an LSTM model to analyze time-series features and identify abnormal patterns (such as high-frequency transfers in the early morning). Biometric data (fingerprint / face) is hashed using SHA-256 to generate unique feature vectors, which are then associated with relevant individuals (relatives, guarantors). When constructing the directed graph, the weights of interaction edges are set (number of joint transactions × 0.6 + geographical similarity × 0.4), and the GCN algorithm is used to correct logically contradictory edges (such as A transferring money to B but B having no corresponding receipt record). Finally, a dynamic network of interpersonal relationships containing several nodes is generated.

[0109] S104: Obtain the corporate credit report of the core enterprise and the personnel credit reports of each related person uploaded by the credit reporting agency through the blockchain consortium chain, and obtain the corporate asset certification documents of the core enterprise and the personnel asset certification documents of each related person uploaded by the asset verification platform.

[0110] Specifically, data acquisition is automatically triggered through smart contracts, including: credit data – calling the credit reporting agency's API to obtain the enterprise's credit score (central bank credit score) and related guarantee relationships, with the data transmitted using the national cryptographic standard SM4; asset verification – verifying the real estate certificate uploaded by the asset verification platform (comparing the on-chain hash value with the offline original) and accounts receivable pledge registration information; data storage – the blockchain records data access logs (e.g., an IP reading an enterprise credit report at 15:03:22), and data version control ensures that historical records are traceable.

[0111] S105: Construct the corporate financial asset map of the core enterprise based on the corporate credit report and the corporate asset certification documents, and construct the personnel financial asset map of each related person based on the personnel credit report and personnel asset certification documents of each related person.

[0112] Specifically, the corporate financial asset graph integrates corporate credit reports, asset certificates, and related personnel data. Node attributes in the corporate financial asset graph include fixed asset valuation (on-chain evidence-based assessment reports) and accounts receivable aging (automatically calculated overdue rate). Edge weights in the corporate financial asset graph include the strength of cooperative relationships, calculated as: Annual Transaction Amount / Industry Average × 0.7 + Years of Cooperation × 0.3. Finally, an RDF inference engine is used to infer implicit relationships within the corporate financial asset graph (e.g., identifying implicit related parties through holding companies). The personnel graph integrates social security payment records and stock holdings (real-time market value obtained via API), forming a multi-dimensional asset network.

[0113] S106: Update the enterprise's financial asset map based on the personnel financial asset map of each related person, and determine the consumer finance risk level of the core enterprise based on the updated enterprise financial asset map and the enterprise supply chain map.

[0114] Specifically, the graph fusion phase uses a graph attention network (GAT) to dynamically adjust node importance. For example, if a related person's credit card delinquency triggers a +15% risk weight for the core enterprise node. The multidimensional risk network includes six dimensions (supply chain stability, asset liquidity, etc.). After determining the weights using the analytic hierarchy process (AHP), a comprehensive score is calculated using a logistic regression model. Risk level classification standards: Low and low-to-medium risk [0-40 points]: routine monitoring. Medium and high-to-medium risk [40-80 points]: manual review. High risk [80-100 points]: transaction circuit breaker. The scoring results are stored on the blockchain and synchronized to the regulatory node.

[0115] S107: Based on the aforementioned consumer finance risk level, configure the consumer finance regulatory process for the next trading day for the core enterprise, and conduct consumer finance supervision of the core enterprise based on the consumer finance regulatory process of that trading day on the next trading day.

[0116] Specifically, regulatory strategies can be dynamically adjusted based on risk levels. One approach includes:

[0117] Low and low-to-medium risk: Automatic approval, only transaction logs are recorded (logs are compressed and retained for 6 months); Medium and high-to-medium risk: OCR verification of document authenticity and manual review are triggered (completed within 2 hours); High risk: Account freeze + biometric verification (voiceprint + facial recognition dual-factor authentication). On the next trading day, consumer finance supervision of core enterprises will be carried out based on the consumer finance regulatory process determined in S107.

[0118] In an optional implementation, see Figure 2 As shown, Figure 2The flowchart of a neural network model training method provided in Embodiment 1 of the present invention is shown, wherein the step of constructing the enterprise supply chain map of the core enterprise based on the supply chain invoice data includes steps S201 to S204:

[0119] S201: Parse the supply chain bill data to obtain basic bill information, circulation records and payment data.

[0120] Specifically, parsing supply chain invoice data is fundamental to constructing an enterprise supply chain map. First, OCR technology is used to recognize text in invoices in PDF and scanned formats, extracting key fields including invoice number, issuance date, amount, circulation record, and payment status. For example, tables and handwritten signature areas in PDF invoices require accurate recognition through image preprocessing (binarization and noise reduction) and character segmentation techniques. Then, a rule engine is used to verify data integrity; for example, amount fields must be numeric and greater than zero, and timestamps must conform to ISO8601 format. The parsed data is stored in a structured format, providing input for subsequent entity relationship extraction.

[0121] S202: Extract the enterprise entity, supplier entity, and bill entity from the basic information of the bill, the circulation record, and the payment data, as well as the opening relationship, circulation relationship, and payment relationship between the entities.

[0122] Specifically, entity recognition is performed from the parsed structured data based on a domain dictionary and a BiLSTM-CRF model. For example, second-tier suppliers are identified through contextual semantic analysis (e.g., Company B in "manufactured by Company A on behalf of Company B"), and three types of entities—enterprises, suppliers, and bills—are extracted using rule templates (e.g., regular expressions matching keywords such as "liquidator" and "acceptor"). Relationship extraction employs a hybrid approach: for issuance relationships, the bill issuance time is matched with the signature position (e.g., "this bill was issued by XX Company"); for circulation relationships, a chain structure based on endorsers is constructed (A→B→C constitutes continuous circulation); and for payment relationships, the payment status field is matched with the acceptor's account. The final output is a timestamped entity relationship triplet, providing high-quality input for graph construction.

[0123] S203: The graph model is constructed using the triplet pattern, with enterprise entities, supplier entities, and bill entities as nodes, and opening relationships, circulation relationships, and payment relationships as edges. The attributes of the edges include relationship type and timestamp, which are stored in the Neo4j graph database to obtain the initial supply chain graph.

[0124] Specifically, an initial supply chain graph is constructed in the Neo4j graph database. Nodes are categorized into three types: enterprises, suppliers, and invoices. Edges represent the issuance, circulation, and payment relationships. Node attributes include the unified social credit code and invoice amount, while edge attributes record timestamps, transaction amounts, etc. For example, the node attributes for an enterprise issuing invoices include invoice type (electronic / paper) and acceptance method, while the attributes of the circulation edge include the number of endorsements and the average circulation interval. Batch import is achieved using the APOC library, and indexes are built to optimize queries (e.g., querying records for the past 3 months by timestamp range).

[0125] S204: Use a graph neural network to complete the missing relationships in the initial supply chain graph to obtain the enterprise supply chain graph.

[0126] Specifically, the GraphSAGE algorithm is used to complete the initial graph, addressing the data sparsity problem. First, feature encoding is performed through three GraphConv layers, with the hidden layer dimension progressively reduced (256→128→64) to capture local node features. Then, second-level neighbor information is aggregated (e.g., inferring potential suppliers of node A from supplier C of node B), and an attention mechanism is introduced to dynamically adjust neighbor weights. For example, if a supplier has a high transaction frequency, its influence weight is increased to 0.6. The completion process includes missing edge detection (identifying isolated nodes by using graph embedding similarity cosine values ​​<0.3) and dynamic updates (daily incremental updates of 20% of newly added relationships from the past 7 days).

[0127] In an optional implementation, see Figure 3 As shown, Figure 3 The flowchart of a method for determining a target neural network model provided in Embodiment 1 of the present invention is shown, wherein the step of constructing a personnel supply chain map of each associated person based on user behavior data and biometric data includes steps S301 to S304:

[0128] S301: The user behavior data is parsed to extract transaction operation information, and the biometric data is encrypted using SHA-256 to generate a feature hash value.

[0129] Specifically, user behavior data (such as transaction time, amount, and counterparty) is parsed in real time using the Apache Kafka streaming platform to extract structured fields (such as transaction type and geographic location). Biometric data (fingerprints / faces) is preprocessed using OpenCV (denoising and normalization), and then a 64-bit feature hash value is generated using the SHA-256 algorithm. For example, a user's fingerprint image is preprocessed into a 256-dimensional vector, and then a unique identifier (such as a3f5d2...b8c9) is generated using SHA-256 hashing, ensuring that the biometric data cannot be irreversibly restored. The parsed data is aggregated according to a time window (1 hour) and stored in an HBase distributed database to provide time-series input for subsequent analysis.

[0130] S302; Based on the feature hash value, determine the identity entity of each associated person, and based on the transaction operation information and the identity entity of each associated person, determine the interaction relationship between each associated person.

[0131] Specifically, the feature hash value is matched against the identity database using a Bloom filter to identify related individuals (such as relatives or guarantors). Transaction operation information (such as co-trading parties and fund flows) is combined with a graph embedding algorithm (Node2Vec) to calculate node similarity and construct interaction relationships. For example, if users A and B complete 5 transactions through the same device within 3 days, relationship attributes such as interaction frequency = 5 and device fingerprint similarity = 0.87 are generated. A rule engine is used to define thresholds (such as frequency > 3 triggering relationship establishment), and finally, weighted interaction relationship triples are output (such as (user A) - [collaboration] {frequency: 5} - (user B)).

[0132] S303: A directed graph model is used to construct nodes and edges. Each associated person entity is used as a node, its identity entity is used as a node attribute, and the interaction relationship between each associated person is used as a directed edge to construct the initial personnel supply chain graph.

[0133] Specifically, a dynamic directed graph is constructed in Neo4j. Nodes include: Personnel entities: name, ID number (de-identified as hash value), role label (e.g., "core enterprise supplier"); Identity entities: hash value, associated device ID, geographical location. Edge attributes include: Interaction type: transaction / communication / co-renting; Time decay factor: e^(-λt) (λ = 0.1, t is the time difference in hours).

[0134] Batch imports are achieved using the APOC library, and composite indexes are built to optimize queries (such as searching by hash value + time range). For example, five financial transactions between a guarantor and a core enterprise form a time-weighted directed edge, where edge weight = transaction amount × time decay factor.

[0135] S304: LSTM neural network is used to analyze the user behavior time series, identify abnormal edges in the initial personnel supply chain map, and the abnormal edges are corrected by graph neural network to obtain the personnel supply chain map.

[0136] Specifically, user behavior time series data is input into an LSTM model (64 hidden units, Dropout = 0.2) to capture abnormal patterns, including: extracting time-series features such as transaction frequency and device switching frequency within a sliding window (30 minutes); and determining anomalies by reconstructing an error threshold using an autoencoder. A Graph Neural Network (GAT) corrects abnormal edges, including: aggregating features of neighboring nodes (such as transaction patterns of associated individuals); calculating edge importance weights (e.g., anomaly edge weight decay coefficient = 0.5); and generating a corrected graph using the Graph SAGE algorithm. For example, if a user's high-frequency transfers in the early morning are marked as abnormal by the LSTM, the GNN can reduce the weight of the abnormal edges by analyzing the normal sleep patterns of their associated individuals, thus retaining reasonable transaction behavior.

[0137] In an optional implementation, see Figure 4 As shown, Figure 4 The flowchart illustrates a method for constructing a corporate financial asset map according to Embodiment 1 of the present invention, wherein the step of constructing the corporate financial asset map of the core enterprise based on the corporate credit report and the corporate asset certification documents includes steps S401 to S403:

[0138] S401: Analyze the enterprise credit report and the enterprise asset certification documents to extract the core enterprise's historical behavior data and enterprise asset status data, respectively.

[0139] Specifically, enterprise credit reports (PDF / structured data) are processed using OCR technology to extract key fields, including credit behavior data such as credit records, external guarantees, and default history. Asset certificates (such as property ownership certificates and equity pledge registrations) are parsed using Natural Language Processing (NLP) to extract information such as asset type, valuation, and ownership status. For example, parsing a property ownership certificate requires identifying fields such as the property owner, real estate registration number, and appraised value, and using regular expressions to validate the data format (e.g., the appraised value must be a number and greater than zero). The data cleaning stage uses a rule engine to filter invalid records (such as expired mortgage information), ultimately generating a structured data table.

[0140] S402: From the enterprise's historical behavior data and the enterprise's asset status data, identify core entities, including core enterprises and cooperative institutions, as well as asset entities, including fixed assets and current assets, and establish business relationships between core enterprises and cooperative institutions, as well as ownership relationships between core enterprises and different asset entities.

[0141] Specifically, based on the parsed structured data, three types of entities are identified using the BiLSTM-CRF model. Core entities: core enterprise name (Unified Social Credit Code), holding subsidiaries (equity ratio ≥ 50%), and related guarantors. Asset entities: fixed assets (real estate / equity, with GPS coordinates) and current assets (accounts receivable / inventory, with transaction contract numbers). Relationship extraction: business relationships (holding, supply and sales agreements) are extracted through dependency parsing, and ownership relationships (mortgage registration number, pledgee) are matched using rule templates. For example, the equity pledge record of a holding subsidiary A of a core enterprise can be linked to A's fixed asset list.

[0142] S403: The attribute graph model is used to construct nodes and edges, with core entities and asset entities as nodes. Credit records and performance capabilities in the historical behavior data are used as attributes of core entity nodes, and asset composition and value assessment in the asset status data are used as attributes of asset entity nodes. Business relationships and ownership relationships are used as edges and weights are set to construct the initial enterprise financial asset graph.

[0143] Specifically, a hierarchical graph is constructed in Neo4j. Node attributes include core enterprise nodes: debt-to-asset ratio (dynamically calculated), industry risk rating (central bank classification); asset nodes: residual value rate (annual depreciation calculation), liquidity level (current assets / total assets). Edge attributes include business relationship edges: transaction frequency (monthly average), contract fulfillment rate (historical fulfillment count / total count); ownership relationship edges: collateral ratio (debt amount / asset valuation), registration status (valid / invalid).

[0144] The weighting is calculated using a dynamic formula, for example, the weight of controlling relationship = equity ratio × 0.6 + control stability × 0.4. Batch import is achieved through the APOC library, and composite indexes are built to optimize queries (such as searching by asset type + region).

[0145] S404: Utilize blockchain notarization technology to verify the authenticity of the data in the initial corporate financial asset graph, infer the implicit relationships in the initial corporate financial asset graph through a knowledge graph completion algorithm, and construct the corporate financial asset graph.

[0146] Specifically, blockchain verification includes: using the national cryptographic algorithm SM3 to generate hash values ​​for key data in the blockchain graph (such as asset valuation and mortgage status), and uploading these hash values ​​to the consortium blockchain's storage nodes to ensure the data is tamper-proof. For example, when real estate mortgage registration information is uploaded to the blockchain, a blockchain timestamp and a CA digital certificate are recorded simultaneously.

[0147] Implicit relationship completion includes: analyzing node relationships based on graph attention networks (GAT), for example: inferring implicit supply chain financial risks through the related guarantee relationships of core enterprise suppliers; and using RDF inference engines (such as Apache Jena) to complete asset relationships that are not explicitly recorded (such as equipment of Company A being held by Company B on its behalf).

[0148] In an optional implementation, see Figure 5 As shown, Figure 5 The flowchart illustrates a method for constructing a personnel financial asset map according to Embodiment 1 of the present invention. The method for constructing the personnel financial asset map based on the personnel credit reports and asset certificates of each associated person includes steps S501 to S504:

[0149] S501: Analyze the personnel credit reports and personnel asset certificates to extract the personnel historical credit data and personnel asset holding data of each related personnel.

[0150] Specifically, personnel credit reports (PDF / structured data) are processed using OCR technology to extract key fields, including credit behavior data such as credit records, external guarantees, and default history. Asset certificates (such as property ownership certificates and equity pledge registrations) are parsed using Natural Language Processing (NLP) to extract information such as asset type, valuation, and ownership status. For example, parsing a property ownership certificate requires identifying fields such as the property owner, real estate registration number, and appraised value, and using regular expressions to validate the data format (e.g., the appraised value must be a number and greater than zero). The data cleaning stage uses a rule engine to filter invalid records (such as expired mortgage information), ultimately generating a structured data table.

[0151] S502; Identify the related personnel identity entities and asset entities from the personnel historical credit data and the personnel asset holding data, establish the direct holding relationship between the related personnel identity entities and asset entities, as well as the indirect relationship between the related personnel and other related personnel.

[0152] Specifically, based on the parsed structured data, a BiLSTM-CRF model is used to identify three types of entities: Personnel entities: name, ID number (de-identified as hash value), kinship (e.g., spouse, parents); Asset entities: fixed assets (real estate / vehicles, with GPS coordinates), financial assets (stocks / funds, with transaction records). Relationship extraction: direct holding relationships (property registration holder), indirect related relationships (joint guarantors, family business equity penetration). For example, the beneficiary information of multiple insurance policies for a certain person can be used to infer their family relationship network.

[0153] S503: A heterogeneous graph model is used to construct nodes and edges, with related personnel identity entities and asset entities as different types of nodes. Default records and credit scores in the historical credit data are used as attributes of related personnel nodes, and asset size and liquidity in the asset holding data are used as attributes of asset nodes. Direct holding relationships and indirect related relationships are used as different types of edges, and weights are set to construct the initial personnel financial asset graph.

[0154] Specifically, a hierarchical heterogeneous graph is constructed in Neo4j, with nodes categorized into individuals and assets, and edges representing holding or association relationships. Node attributes include: Individual nodes: credit score (central bank credit score), debt ratio (total debt / income); Asset nodes: residual value rate (annual depreciation calculation), liquidity level (current assets / total assets). Edge attributes include: Direct holding edges: property registration status (valid / invalid), mortgage ratio (debt amount / asset valuation); Indirect association edges: joint investment ratio, kinship strength (blood distance × 0.6 + social association × 0.4). Batch import is achieved through the APOC library, and a composite index is built to optimize queries (e.g., searching by asset type + region).

[0155] S504: Utilize blockchain notarization technology to verify the authenticity of the data in the initial personnel financial asset map, and use knowledge reasoning algorithms to complete the implicit relationships in the initial personnel financial asset map to construct the personnel financial asset map.

[0156] Specifically, the national cryptographic algorithm SM3 is used to generate hash values ​​for key graph data (such as asset valuation and kinship), which are then uploaded to the consortium blockchain's evidence storage nodes to ensure data immutability. For example, when real estate mortgage registration information is uploaded to the blockchain, the blockchain timestamp and CA digital certificate are recorded simultaneously. Node relationships are analyzed based on Graph Attention Network (GAT), for example, by inferring the potential joint liability of A and C through the kinship between A's guarantor B and C. RDF inference engines (such as Apache Jena) are used to supplement unrecorded asset holding arrangements (such as equipment of Company D being held by individual E).

[0157] In an optional implementation, see Figure 6 As shown, Figure 6 The flowchart of a consumer finance risk level method provided in Embodiment 1 of the present invention is shown. The method involves updating the enterprise's financial asset map based on the personnel financial asset map of each associated individual, and determining the consumer finance risk level of the core enterprise based on the updated enterprise financial asset map and the enterprise supply chain map. This includes steps S601 to S604.

[0158] S601: Extract the relationship between the related personnel and the core enterprise from the personnel financial asset map of each related person, map the relationship to the corresponding node of the enterprise financial asset map, and update the credit attribute and asset association attribute of the related personnel of the core enterprise node.

[0159] Specifically, the direct / indirect relationships (such as guarantees and equity penetration) between related personnel and core enterprises are extracted from the personnel financial asset graph, and node mapping is achieved through Neo4j's Cypher query language. For example, if related personnel A provides joint liability guarantees for core enterprises, the guarantee coverage ratio attribute of the core enterprise node in the enterprise graph is updated (calculation formula: guarantee amount / net assets × 100%). At the same time, based on personnel asset change records (such as real estate appreciation and stock reduction), the implicit debt ratio of core enterprises is dynamically adjusted (formula: implicit debt = related assets × risk exposure coefficient, the coefficient is dynamically assigned according to the relationship type [0.3-0.7]).

[0160] S602: Use graph matching algorithm to identify common nodes in the personnel financial asset graph and the enterprise financial asset graph, and trigger dynamic updates of node attributes in the enterprise financial asset graph based on the asset change records of related personnel.

[0161] Specifically, graph isomorphism detection algorithms (such as VF2) are used to identify common nodes in the personnel and enterprise graphs (such as the same property / equity), triggering bidirectional attribute updates. For example, if the market value of personnel B's property drops by 10%, the residual value rate of all collateral under B's name in the enterprise graph is updated simultaneously (formula: residual value rate = current valuation / original valuation × 100%). For abnormal transactions (such as sudden interruptions) between upstream and downstream enterprises in the supply chain, the node similarity is calculated using a graph embedding algorithm (Node2Vec). If the similarity is lower than a threshold (such as 0.4), the supply chain stability score of the enterprise node is downgraded (downgrade amount = 1 - similarity).

[0162] S603: The updated corporate financial asset map and corporate supply chain map are fused together. Centered on the core enterprise node, the transaction relationships in the supply chain map and the credit asset relationships in the financial asset map are integrated to form a multi-dimensional risk network.

[0163] Specifically, centered on the core enterprise, the supply chain graph (nodes including suppliers and logistics providers) and the financial asset graph (nodes including banks and guarantee institutions) are fused. Graph fusion includes node merging strategies: identical entities (such as "XX Logistics Company") are aligned using a unified social credit code, and conflicting attributes are resolved through weighted averaging (weight = data source credibility). It also includes edge relationship enhancement: supply chain transaction relationships (such as "purchasing raw materials from Company A") and financial asset relationships (such as "Credit Line of Company A") are combined into composite edges, with attributes including transaction frequency, credit coverage, etc. Finally, risk propagation modeling: risk transmission paths are simulated based on graph convolutional networks (GCNs), for example, the probability P of supplier default propagating to the core enterprise through supply chain relationships. 传播 The calculation formula is:

[0164]

[0165] Where, p k Let w be the probability of default for the k-th associated node. k Let n be the edge weight and n be the number of nodes.

[0166] S604: Based on the fused multidimensional risk network, the analytic hierarchy process is used to determine each risk factor and its weight. According to the actual value and weight of each risk factor, a comprehensive risk score is calculated through a preset risk assessment function. The comprehensive risk score is compared with a preset risk grading standard to determine the consumer finance risk level.

[0167] Specifically, from the integrated multidimensional risk network, primary risk factors (credit risk, liquidity risk, operational risk, and market risk) are extracted, and further broken down into secondary sub-factors (such as credit risk including supplier fulfillment rate and guarantee coverage rate). The weights of the primary factors are determined using the analytic hierarchy process (e.g., 40% for credit risk, 25% for liquidity risk), while the weights of the secondary sub-factors are dynamically calculated using the entropy weight method. This utilizes the dispersion of each sub-factor's data to reflect its degree of impact on risk, avoiding subjective weighting bias.

[0168] Based on the multidimensional risk network after graph fusion, factors strongly correlated with consumer finance risk are screened. For example, "supplier fulfillment rate" is extracted from supply chain transaction relationships, and "guarantee coverage rate" is extracted from financial asset relationships, ensuring that the factors cover both business operations (supply chain) and financial attributes (assets, credit).

[0169] The data for each secondary sub-factor are standardized and uniformly mapped to a score range of 0-100. Taking "supplier fulfillment rate" as an example, if the historical fulfillment rate follows a normal distribution, the deviation of the actual value from the mean and standard deviation is calculated and converted into a standardized score. "Guarantee coverage rate" is calculated using the original formula (guarantee amount / net assets × 100%) and then normalized to 0-100 points using min-max normalization to eliminate dimensional differences.

[0170] A comprehensive scoring model is constructed to calculate the comprehensive risk score:

[0171]

[0172] Among them, W i For factor weights (first- and second-level factor weights are calculated in a nested manner; for example, under a credit risk weight of 40%, the supplier fulfillment rate sub-weight is obtained using the entropy weight method and then multiplied by 40%), S i The scores are standardized (e.g., supplier fulfillment rates are converted to a normal distribution and scored from 0 to 100). Through stratified weighting, the importance differences of macro-level risk categories (primary factors) are reflected, while the impact of micro-level sub-factors (secondary factors) is refined, making the scores more closely reflect the actual risk structure. m represents the number of factors.

[0173] The system establishes a predefined correspondence between score ranges and risk levels (e.g., [0,20] represents low risk, [80,100] represents high risk), and the range division is validated using industry risk thresholds and historical data. For example, by retrospectively analyzing risk events of core enterprises and statistically analyzing the distribution of comprehensive scores under different risk levels, the range boundaries are adjusted to ensure that the level determination effectively distinguishes the degree of risk.

[0174] This application provides an achievable mapping relationship between consumer finance risk levels and score ranges, as shown in the table below:

[0175] Consumer finance risk level Score range Low risk [0,20) Low to medium risk [20,40) Medium risk [40,60) Medium and high risk [60,80) High risk [80,100]

[0176] This application also employs a dynamic adjustment mechanism, regularly (e.g., quarterly) updating the risk factor system and score ranges. When the market environment changes (e.g., an increase in industry default rates) or the company's business model is adjusted (e.g., the supply chain expands to new regions), the factor weights are recalculated and the standardization method is optimized to ensure that the risk level assessment continuously adapts to the actual scenario and avoids misjudgments due to model rigidity.

[0177] In an optional implementation, see Figure 7 As shown, Figure 7 The flowchart illustrates a consumer finance regulatory process configuration method provided in Embodiment 1 of the present invention, wherein configuring the consumer finance regulatory process for the core enterprise for the next trading day based on the consumer finance risk level includes steps S701 to S702:

[0178] S701: Determine the corresponding regulatory intensity coefficient based on the aforementioned consumer finance risk level, wherein the regulatory intensity coefficient is positively correlated with the risk level.

[0179] Specifically, according to the "Supervisory Rating Measures for Consumer Finance Companies," consumer finance risk levels are divided into low risk, low-to-medium risk, medium risk, medium-to-high risk, and high risk, with a basic regulatory intensity coefficient (K). base The risk level of consumer finance is positively correlated with the risk level of consumer finance, and the specific mapping relationship is shown in the table below:

[0180] Consumer finance risk level <![CDATA[Basic supervision intensity coefficient (K base )]]> Low risk 0.5 Low to medium risk 0.7 Medium risk 0.9 Medium and high risk 1.1 High risk 1.3

[0181] It also employs a dynamic adjustment mechanism, using a rule engine (such as Drools) to monitor real-time risk events (such as related-party asset defaults or supply chain disruptions) and automatically trigger a coefficient jump. For example, when the guarantee coverage ratio in the enterprise graph decreases by 10% or the implicit debt ratio exceeds the threshold, K... base The value rose from 0.7 (low to medium risk) to 0.9 (medium risk).

[0182] According to the composite calculation model K final =K base +ΔK adjust Determine the final regulatory intensity coefficient K final , where ΔK adjust The risk classification results are dynamically adjusted based on the quarterly risk classification results (such as the frequency of abnormal transactions and the rate of increase in the complaint rate).

[0183] S702: Determine the consumer finance regulatory process for the core enterprise on the next trading day based on the regulatory intensity coefficient.

[0184] Specifically, K final When the value is ≤0.7, the consumer finance regulatory process for the next trading day will be primarily based on off-site monitoring. 0.7<K final When K < 1.3, the consumer finance regulatory process configuration for the next trading day includes: increasing the frequency of window guidance and limiting the proportion of high-risk business to the first preset threshold. 0.9 < K final When the threshold is <1.3, the regulatory process for consumer finance on the next trading day will include: initiating a special audit and raising the capital adequacy ratio requirement to the second preset threshold. When the threshold is <1.1, the regulatory process for consumer finance on the next trading day will include: initiating a special audit and raising the capital adequacy ratio requirement to the second preset threshold. final When the threshold is <1.3, the regulatory process for consumer finance on the next trading day will include restricting shareholder dividends and dispatching a regulatory working group. finall When the value is ≥1.3, the consumer finance regulatory process for the next trading day will be configured as follows: initiate takeover or market exit procedures.

[0185] Example 2

[0186] See Figure 8 As shown, Figure 8 This diagram illustrates the structure of a consumer finance intelligent risk control system according to Embodiment 2 of the present invention, wherein the system includes:

[0187] The blockchain consortium blockchain deployment module 801 is used to deploy the blockchain consortium blockchain, establish an encrypted data channel with the supply chain finance platform, credit reporting agency and asset verification platform, and set the data update frequency of the blockchain consortium blockchain to be updated once per transaction day.

[0188] The enterprise supply chain map construction module 802 is used to obtain the supply chain invoice data of the core enterprise uploaded by the supply chain finance platform through the blockchain consortium chain after the end of each trading day, and construct the enterprise supply chain map of the core enterprise based on the supply chain invoice data.

[0189] The personnel supply chain graph construction module 803 is used to obtain user behavior data and biometric data of the core enterprise's related personnel uploaded by the supply chain finance platform through the blockchain consortium chain, and construct personnel supply chain graphs of each related personnel based on the user behavior data and biometric data of each related personnel.

[0190] The enterprise personnel report document acquisition module 804 is used to acquire the enterprise credit report of the core enterprise and the personnel credit reports of each related person uploaded by the credit reporting agency through the blockchain consortium chain, and to acquire the enterprise asset certificate documents of the core enterprise and the personnel asset certificate documents of each related person uploaded by the asset verification platform.

[0191] The financial asset mapping module 805 is used to construct the corporate financial asset mapping of the core enterprise based on the corporate credit report and the corporate asset certification documents, and to construct the personnel financial asset mapping of each related person based on the personnel credit report and personnel asset certification documents of each related person.

[0192] The financial risk level determination module 806 is used to update the enterprise's financial asset map based on the personnel financial asset map of each related person, and to determine the consumer finance risk level of the core enterprise based on the updated enterprise financial asset map and the enterprise supply chain map.

[0193] The consumer finance supervision module 807 is used to configure the consumer finance supervision process for the core enterprise for the next trading day based on the consumer finance risk level, and to conduct consumer finance supervision of the core enterprise based on the consumer finance supervision process of the next trading day.

[0194] In an optional implementation, constructing the enterprise supply chain map of the core enterprise based on the supply chain invoice data includes:

[0195] The supply chain bill data is parsed to obtain basic bill information, circulation records, and payment data;

[0196] Extract the enterprise entity, supplier entity, and bill entity from the basic information of the bill, the circulation record, and the payment data, as well as the opening relationship, circulation relationship, and payment relationship between the entities;

[0197] A graph model is constructed using the triplet pattern, with enterprise entities, supplier entities, and bill entities as nodes, and opening relationships, circulation relationships, and payment relationships as edges. The attributes of the edges include relationship type and timestamp, which are stored in the Neo4j graph database to obtain the initial supply chain graph.

[0198] The enterprise supply chain map is obtained by using a graph neural network to complete the missing relationships in the initial supply chain map.

[0199] In an optional implementation, the construction of the personnel supply chain map for each associated person based on their user behavior data and biometric data includes:

[0200] The user behavior data is parsed to extract transaction operation information, and the biometric data is encrypted using SHA-256 to generate a feature hash value.

[0201] The identity entities of each associated person are determined based on the feature hash value, and the interaction relationships between each associated person are determined based on the transaction operation information and the identity entities of each associated person.

[0202] A directed graph model is used to construct nodes and edges. Each associated person entity is used as a node, its identity entity is used as a node attribute, and the interaction relationship between each associated person is used as a directed edge to construct an initial personnel supply chain graph.

[0203] An LSTM neural network is used to analyze the time series of user behavior to identify abnormal edges in the initial personnel supply chain map. After correcting the abnormal edges using a graph neural network, the personnel supply chain map is obtained.

[0204] In an optional implementation, constructing the core enterprise's financial asset map based on the enterprise credit report and the enterprise asset documentation includes:

[0205] The enterprise credit report and the enterprise asset certification documents are parsed to extract the core enterprise's historical behavior data and enterprise asset status data, respectively.

[0206] From the enterprise's historical behavior data and the enterprise's asset status data, identify core entities, including core enterprises and cooperative institutions, as well as asset entities, including fixed assets and current assets, and establish business relationships between core enterprises and cooperative institutions, as well as ownership relationships between core enterprises and different asset entities.

[0207] An attribute graph model is used to construct nodes and edges, with core entities and asset entities as nodes. Credit records and performance capabilities in the historical behavior data are used as attributes of core entity nodes, and asset composition and value assessment in the asset status data are used as attributes of asset entity nodes. Business relationships and ownership relationships are used as edges, and weights are set to construct an initial corporate financial asset graph.

[0208] The authenticity of the data in the initial corporate financial asset graph is verified by using blockchain notarization technology, and the implicit relationships in the initial corporate financial asset graph are inferred by using a knowledge graph completion algorithm to construct the corporate financial asset graph.

[0209] In an optional implementation, the construction of the financial asset map of each associated person based on their credit reports and asset certificates includes:

[0210] The personnel credit reports and personnel asset certificates are analyzed to extract the historical credit data and asset holding data of each related person.

[0211] From the historical credit data of the personnel and the asset holding data of the personnel, identify the identity entities and asset entities of related personnel, establish the direct holding relationship between the identity entities and asset entities of related personnel, as well as the indirect relationship between related personnel and other related personnel;

[0212] A heterogeneous graph model is used to construct nodes and edges, with related personnel identity entities and asset entities as different types of nodes. Default records and credit scores in the historical credit data are used as attributes of related personnel nodes, and asset size and liquidity in the asset holding data are used as attributes of asset nodes. Direct holding relationships and indirect relationships are used as different types of edges, and weights are set to construct an initial personnel financial asset graph.

[0213] The authenticity of the data in the initial personnel financial asset map is verified by using blockchain notarization technology, and the implicit relationships in the initial personnel financial asset map are completed by using knowledge reasoning algorithms to construct the personnel financial asset map.

[0214] In an optional implementation, the updating of the enterprise financial asset map based on the personnel financial asset map of each associated person, and the determination of the consumer finance risk level of the core enterprise based on the updated enterprise financial asset map and the enterprise supply chain map, includes:

[0215] Extract the relationship between the related personnel and the core enterprise from the personnel financial asset map of each related person, map the relationship to the corresponding node of the enterprise financial asset map, and update the credit attribute and asset association attribute of the related personnel of the core enterprise node.

[0216] A graph matching algorithm is used to identify common nodes in the financial asset graphs of individuals and enterprises, and the node attributes of the enterprise financial asset graph are dynamically updated based on the asset change records of related individuals.

[0217] The updated corporate financial asset map and corporate supply chain map are fused together. Centered on the core enterprise node, the transaction relationships in the supply chain map and the credit asset relationships in the financial asset map are integrated to form a multi-dimensional risk network.

[0218] Based on the fused multidimensional risk network, the analytic hierarchy process (AHP) is used to determine each risk factor and its weight. According to the actual value and weight of each risk factor, a comprehensive risk score is calculated using a preset risk assessment function. The comprehensive risk score is then compared with a preset risk grading standard to determine the consumer finance risk level.

[0219] In an optional implementation, configuring the consumer finance regulatory process for the core enterprise for the next trading day based on the consumer finance risk level includes:

[0220] The corresponding regulatory intensity coefficient is determined based on the aforementioned consumer finance risk level, wherein the regulatory intensity coefficient is positively correlated with the risk level;

[0221] The consumer finance regulatory process for the core enterprise on the next trading day is determined based on the regulatory intensity coefficient.

[0222] Example 3

[0223] Based on the same application concept, see [link / reference] Figure 9 As shown, Figure 9 A schematic diagram of the structure of a computer device provided in Embodiment 3 of the present invention is shown, wherein, as Figure 9 As shown, the computer device 900 provided in Embodiment 3 of this application includes:

[0224] The computer device 900 includes a processor 901, a memory 902, and a bus 903. The memory 902 stores machine-readable instructions that can be executed by the processor 901. When the computer device 900 is running, the processor 901 communicates with the memory 902 through the bus 903. When the machine-readable instructions are executed by the processor 901, the steps of the consumer finance intelligent risk control method shown in Embodiment 1 are performed.

[0225] Example 4

[0226] Based on the same concept, this application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the consumer finance intelligent risk control method described in any of the above embodiments.

[0227] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and apparatus described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0228] The computer program product for intelligent risk control in consumer finance provided in this embodiment of the invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0229] The intelligent risk control system for consumer finance provided in this invention can be specific hardware on a device or software or firmware installed on the device. The system provided in this invention has the same implementation principle and technical effects as the aforementioned method embodiments. For the sake of brevity, any parts not mentioned in the system embodiments can be referred to the corresponding content in the aforementioned method embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can all be referred to the corresponding processes in the above method embodiments, and will not be repeated here.

[0230] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. The system embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some communication interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0231] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0232] In addition, the functional units in the embodiments provided by the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0233] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0234] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0235] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. All should be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A smart risk control method for consumer finance, characterized in that, The method includes: Deploy a blockchain consortium blockchain and establish encrypted data channels with supply chain finance platforms, credit reporting agencies, and asset verification platforms. At the same time, set the data update frequency of the blockchain consortium blockchain to be updated once per transaction day. After the end of each trading day, the supply chain bill data of the core enterprise uploaded by the supply chain finance platform is obtained through the blockchain consortium chain, and the enterprise supply chain map of the core enterprise is constructed based on the supply chain bill data. The blockchain consortium blockchain is used to obtain user behavior data and biometric data of the core enterprise’s related personnel uploaded to the supply chain finance platform. Based on the user behavior data and biometric data of each related person, a personnel supply chain map of each related person is constructed. The core enterprise's corporate credit report and the personnel credit reports of each associated person are obtained through the blockchain consortium chain; the core enterprise's corporate asset certification documents and the personnel asset certification documents of each associated person are obtained through the asset verification platform. The core enterprise's financial asset map is constructed based on the enterprise's credit report and asset certification documents, and the related personnel's financial asset map is constructed based on the related personnel's credit reports and asset certification documents. The enterprise financial asset map is updated based on the personnel financial asset map of each related person, and the consumer finance risk level of the core enterprise is determined based on the updated enterprise financial asset map and the enterprise supply chain map. Based on the aforementioned consumer finance risk level, the consumer finance regulatory process for the next trading day is configured for the core enterprise, and on the next trading day, consumer finance supervision of the core enterprise is carried out based on the consumer finance regulatory process of that trading day.

2. The method according to claim 1, characterized in that, The construction of the core enterprise's supply chain map based on the supply chain invoice data includes: The supply chain bill data is parsed to obtain basic bill information, circulation records, and payment data; Extract the enterprise entity, supplier entity, and bill entity from the basic information of the bill, the circulation record, and the payment data, as well as the opening relationship, circulation relationship, and payment relationship between the entities; A graph model is constructed using the triplet pattern, with enterprise entities, supplier entities, and bill entities as nodes, and opening relationships, circulation relationships, and payment relationships as edges. The attributes of the edges include relationship type and timestamp, which are stored in the Neo4j graph database to obtain the initial supply chain graph. The enterprise supply chain map is obtained by using a graph neural network to complete the missing relationships in the initial supply chain map.

3. The method according to claim 1, characterized in that, The construction of the personnel supply chain map for each associated person based on user behavior data and biometric data includes: The user behavior data is parsed to extract transaction operation information, and the biometric data is encrypted using SHA-256 to generate a feature hash value. The identity entities of each associated person are determined based on the feature hash value, and the interaction relationships between each associated person are determined based on the transaction operation information and the identity entities of each associated person. A directed graph model is used to construct nodes and edges. Each associated person entity is used as a node, its identity entity is used as a node attribute, and the interaction relationship between each associated person is used as a directed edge to construct an initial personnel supply chain graph. An LSTM neural network is used to analyze the time series of user behavior to identify abnormal edges in the initial personnel supply chain map. After correcting the abnormal edges using a graph neural network, the personnel supply chain map is obtained.

4. The method according to claim 1, characterized in that, The construction of the core enterprise's financial asset map based on the enterprise's credit report and asset certification documents includes: The enterprise credit report and the enterprise asset certification documents are parsed to extract the core enterprise's historical behavior data and enterprise asset status data, respectively. From the enterprise's historical behavior data and the enterprise's asset status data, identify core entities, including core enterprises and cooperative institutions, as well as asset entities, including fixed assets and current assets, and establish business relationships between core enterprises and cooperative institutions, as well as ownership relationships between core enterprises and different asset entities. An attribute graph model is used to construct nodes and edges, with core entities and asset entities as nodes. Credit records and performance capabilities in the historical behavior data are used as attributes of core entity nodes, and asset composition and value assessment in the asset status data are used as attributes of asset entity nodes. Business relationships and ownership relationships are used as edges, and weights are set to construct an initial corporate financial asset graph. The authenticity of the data in the initial corporate financial asset graph is verified by using blockchain notarization technology, and the implicit relationships in the initial corporate financial asset graph are inferred by using a knowledge graph completion algorithm to construct the corporate financial asset graph.

5. The method according to claim 1, characterized in that, The construction of a financial asset map of each associated person based on their credit reports and asset certificates includes: The personnel credit reports and personnel asset certificates are analyzed to extract the historical credit data and asset holding data of each related person. From the historical credit data of the personnel and the asset holding data of the personnel, identify the identity entities and asset entities of related personnel, establish the direct holding relationship between the identity entities and asset entities of related personnel, as well as the indirect relationship between related personnel and other related personnel; A heterogeneous graph model is used to construct nodes and edges, with related personnel identity entities and asset entities as different types of nodes. Default records and credit scores in the historical credit data are used as attributes of related personnel nodes, and asset size and liquidity in the asset holding data are used as attributes of asset nodes. Direct holding relationships and indirect relationships are used as different types of edges, and weights are set to construct an initial personnel financial asset graph. The authenticity of the data in the initial personnel financial asset map is verified by using blockchain notarization technology, and the implicit relationships in the initial personnel financial asset map are completed by using knowledge reasoning algorithms to construct the personnel financial asset map.

6. The method according to claim 1, characterized in that, The enterprise financial asset map is updated based on the personnel financial asset map of each related person. Based on the updated enterprise financial asset map and the enterprise supply chain map, the consumer finance risk level of the core enterprise is determined, including: Extract the relationship between the related personnel and the core enterprise from the personnel financial asset map of each related person, map the relationship to the corresponding node of the enterprise financial asset map, and update the credit attribute and asset association attribute of the related personnel of the core enterprise node. A graph matching algorithm is used to identify common nodes in the financial asset graphs of individuals and enterprises, and the node attributes of the enterprise financial asset graph are dynamically updated based on the asset change records of related individuals. The updated corporate financial asset map and corporate supply chain map are fused together. Centered on the core enterprise node, the transaction relationships in the supply chain map and the credit asset relationships in the financial asset map are integrated to form a multi-dimensional risk network. Based on the fused multidimensional risk network, the analytic hierarchy process (AHP) is used to determine each risk factor and its weight. According to the actual value and weight of each risk factor, a comprehensive risk score is calculated using a preset risk assessment function. The comprehensive risk score is then compared with a preset risk grading standard to determine the consumer finance risk level.

7. The method according to claim 1, characterized in that, The process for configuring the consumer finance regulatory procedures for the core enterprise for the next trading day based on the consumer finance risk level includes: The corresponding regulatory intensity coefficient is determined based on the aforementioned consumer finance risk level, wherein the regulatory intensity coefficient is positively correlated with the risk level; The consumer finance regulatory process for the core enterprise on the next trading day is determined based on the regulatory intensity coefficient.

8. A consumer finance intelligent risk control system, characterized in that, The system includes: The blockchain consortium blockchain deployment module is used to deploy the blockchain consortium blockchain, establish encrypted data channels with supply chain finance platforms, credit reporting agencies, and asset verification platforms, and set the data update frequency of the blockchain consortium blockchain to be updated once per transaction day. The enterprise supply chain map construction module is used to obtain the supply chain invoice data of the core enterprise uploaded by the supply chain finance platform through the blockchain consortium chain after the end of each trading day, and construct the enterprise supply chain map of the core enterprise based on the supply chain invoice data. The personnel supply chain map construction module is used to obtain user behavior data and biometric data of the core enterprise's related personnel uploaded by the supply chain finance platform through the blockchain consortium chain, and construct personnel supply chain maps of each related personnel based on the user behavior data and biometric data of each related personnel. The enterprise personnel report document acquisition module is used to acquire the enterprise credit report of the core enterprise and the personnel credit reports of each related person uploaded by the credit reporting agency through the blockchain consortium chain, and to acquire the enterprise asset certificate documents of the core enterprise and the personnel asset certificate documents of each related person uploaded by the asset verification platform. The financial asset mapping module is used to construct the corporate financial asset mapping of the core enterprise based on the enterprise credit report and the enterprise asset certification documents, and to construct the personnel financial asset mapping of each related person based on the personnel credit reports and personnel asset certification documents of each related person. The financial risk level determination module is used to update the enterprise's financial asset map based on the personnel financial asset map of each related person, and to determine the consumer finance risk level of the core enterprise based on the updated enterprise financial asset map and the enterprise supply chain map. The consumer finance supervision module is used to configure the consumer finance supervision process for the core enterprise for the next trading day based on the consumer finance risk level, and to conduct consumer finance supervision of the core enterprise based on the consumer finance supervision process of the next trading day.

9. A computer device, characterized in that, include: The computer device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of the consumer finance intelligent risk control method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the consumer finance intelligent risk control method as described in any one of claims 1 to 7.

Images