Enterprise customer risk assessment method, apparatus, device, medium, and program product

Abstract

The application provides an enterprise customer risk assessment method, device, equipment, medium and program product, which can be applied to the field of big data technology. The method comprises the following steps: acquiring transaction flow data of a target enterprise customer and associated enterprise customers; taking the target enterprise customer and the associated enterprise customers as nodes, determining a transaction relationship between the nodes according to the transaction flow data, and constructing a transaction knowledge graph with the transaction relationship as an edge; dividing the target enterprise customer and the associated enterprise customers into multiple clusters according to the transaction relationship between the nodes in the transaction knowledge graph; acquiring risk event information of each node; and calculating a risk score of the target enterprise customer according to the risk event information of the nodes in each cluster and the weight corresponding to each cluster, wherein the risk score represents the degree of influence of the associated risk of the target enterprise customer caused by the risk events of the associated enterprise customers.

Description

Technical Field

[0001] This application relates to the field of big data technology, specifically to a method, apparatus, equipment, medium, and program product for enterprise customer risk assessment. Background Technology

[0002] In related technologies, risk assessment of corporate clients relies on static relationships such as equity and guarantees, making it difficult to identify implicit connections formed through transaction flows, resulting in omissions in risk transmission paths and insufficient assessment accuracy. Summary of the Invention

[0003] In view of the above problems, embodiments of this application provide a method, apparatus, device, medium, and program product for enterprise customer risk assessment.

[0004] According to a first aspect of this application, a method for assessing enterprise customer risk is provided, comprising: acquiring transaction flow data of a target enterprise customer and its related enterprise customers; determining the transaction relationships between nodes based on the transaction flow data, using the target enterprise customer and related enterprise customers as nodes, and constructing a transaction knowledge graph with the transaction relationships as edges; dividing the target enterprise customer and related enterprise customers into multiple clusters based on the transaction relationships between nodes in the transaction knowledge graph, with different clusters corresponding to different degrees of transaction relationship closeness, the target enterprise customer belonging to the first cluster, and the transaction relationship closeness between nodes in the first cluster and the target enterprise customer being higher than the transaction relationship closeness between nodes in other clusters and the target enterprise customer; acquiring risk event information of each node; and calculating a risk score for the target enterprise customer based on the risk event information of nodes in each cluster and the weights corresponding to each cluster, the risk score representing the degree of impact of associated risks suffered by the target enterprise customer due to risk events of related enterprise customers.

[0005] According to an embodiment of this application, based on the transaction relationships between nodes in the transaction knowledge graph, target enterprise customers and related enterprise customers are divided into multiple clusters, including: extracting transaction pairs whose bidirectional transaction frequencies both exceed a preset threshold based on the transaction relationships between nodes in the transaction knowledge graph, determining multiple node sets, and connecting the nodes in each node set to each other through transaction pairs; determining the target node set where the target enterprise customer is located from the multiple node sets; and dividing the target enterprise customer and related enterprise customers into multiple clusters based on the closeness of their transaction relationships with the target node set.

[0006] According to embodiments of this application, based on the closeness of the transaction relationship with the target node set, the target enterprise customers and associated enterprise customers are divided into multiple clusters, including: dividing nodes in the target node set whose transaction relationship closeness is greater than or equal to a first threshold into a first cluster, the first cluster containing the target enterprise customers; dividing nodes in the transaction knowledge graph that have a direct transaction relationship with nodes in the first cluster and whose transaction relationship closeness with nodes in the first cluster is less than the first threshold but greater than or equal to a second threshold into a second cluster, the first threshold being greater than the second threshold; and dividing nodes in the transaction knowledge graph that have a direct transaction relationship with nodes in the second cluster and are only indirectly associated with the first cluster through nodes in the second cluster into a third cluster.

[0007] According to embodiments of this application, obtaining risk event information for each node includes: obtaining the original text of risk events related to each node; performing structured parsing of the original text of risk events using a large language model to obtain structured risk event information; and associating the structured risk event information with the corresponding node.

[0008] According to an embodiment of this application, the structured risk event information includes the subject identifier of the risk event; associating the structured risk event information with the corresponding node includes: matching the subject identifier with nodes in the transaction knowledge graph, and associating the structured risk event information with the successfully matched node.

[0009] According to an embodiment of this application, a risk score for a target enterprise customer is calculated based on the risk event information of nodes in each cluster and the weights corresponding to each cluster, including: obtaining risk event information of nodes in each cluster within a preset time range; and determining the risk score for the target enterprise customer based on the risk event information of nodes in each cluster within the preset time range and the weights corresponding to each cluster.

[0010] According to an embodiment of this application, risk event information includes the number of times risk events occur, their types, and the repetition of similar risk events on the same node. Based on the risk event information of nodes in each cluster within a preset time range and the weights corresponding to each cluster, a risk score for the target enterprise customer is determined. This includes: determining the risk contribution value of each cluster based on the number of times risk events occur on nodes in each cluster within a preset time range, the weights corresponding to different types of risk events, and the enhancement factor corresponding to multiple occurrences of the same type of risk event on the same node; and determining the risk score for the target enterprise customer based on the risk contribution value of each cluster and the weights corresponding to each cluster. The weight corresponding to the first cluster is greater than the weights corresponding to other clusters.

[0011] According to a second aspect of this application, a corporate customer risk assessment device is provided, comprising: a first acquisition module for acquiring transaction flow data of a target corporate customer and its related corporate customers; a construction module for determining the transaction relationships between nodes based on the transaction flow data, using the target corporate customer and related corporate customers as nodes, and constructing a transaction knowledge graph with the transaction relationships as edges; a partitioning module for partitioning the target corporate customer and related corporate customers into multiple clusters based on the transaction relationships between nodes in the transaction knowledge graph, with different clusters corresponding to different degrees of transaction relationship closeness, the target corporate customer belonging to the first cluster, and the transaction relationship closeness between the nodes in the first cluster and the target corporate customer being higher than the transaction relationship closeness between the nodes in other clusters and the target corporate customer; a second acquisition module for acquiring risk event information of each node; and a calculation module for calculating a risk score for the target corporate customer based on the risk event information of the nodes in each cluster and the weights corresponding to each cluster, the risk score representing the degree of impact of associated risks suffered by the target corporate customer due to risk events of related corporate customers.

[0012] According to a third aspect of this application, an electronic device is provided, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method described above.

[0013] According to a fourth aspect of this application, a computer-readable storage medium is also provided, on which a computer program or instructions are stored, wherein the computer program or instructions, when executed by a processor, implement the steps of the above-described method.

[0014] According to a fifth aspect of this application, a computer program product is also provided, including a computer program or instructions that, when executed by a processor, implement the steps of the above-described method. Attached Figure Description

[0015] The above-mentioned contents, other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0016] Figure 1 The illustrations depict application scenarios of enterprise customer risk assessment methods, apparatus, devices, media, and program products according to embodiments of this application.

[0017] Figure 2 A flowchart illustrating an enterprise customer risk assessment method according to an embodiment of this application is shown schematically.

[0018] Figure 3 A flowchart illustrating a corporate customer risk assessment method according to another embodiment of this application is shown schematically;

[0019] Figure 4 A schematic diagram illustrating the principle of an enterprise customer risk assessment method according to an embodiment of this application is shown.

[0020] Figure 5 This schematic diagram illustrates a structural block diagram of an enterprise customer risk assessment apparatus according to an embodiment of this application;

[0021] Figure 6 A block diagram schematically illustrates an electronic device suitable for implementing a risk assessment method for enterprise customers according to an embodiment of this application. Detailed Implementation

[0022] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0023] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0024] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0025] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0026] In the technical solution of this application, the user information (including but not limited to user personal information, user image information, user device information, such as location information) and data (including but not limited to data used for analysis, stored data, and displayed data) involved are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of related data all comply with relevant laws, regulations, and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse.

[0027] In scenarios where personal information is used for automated decision-making, the methods, devices, and systems provided in this application all provide users with corresponding operation entry points for users to choose to agree to or reject the automated decision results; if the user chooses to reject, the process enters the expert decision-making process.

[0028] As used in this paper, the term "model" refers to a model that learns the relationship between inputs and outputs from training data, enabling it to generate corresponding outputs for a given input after training. Model generation can be based on machine learning techniques. Deep learning is a machine learning algorithm that processes inputs and provides corresponding outputs using multiple layers of processing units. A neural network model is an example of a deep learning-based model. In this paper, "model" may also be referred to as a "machine learning model," "learning model," "machine learning network," or "learning network," and these terms are used interchangeably.

[0029] Figure 1 The illustration schematically depicts application scenarios of enterprise customer risk assessment methods, apparatuses, devices, media, and program products according to embodiments of this application. For example... Figure 1 As shown, application scenario 100 according to an embodiment of this application may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 serves as a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables. For example, a user can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send information, etc.

[0030] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be electronic devices such as smartphones, wearable devices, personal computers, intelligent voice interaction devices, smart home appliances, intelligent vehicles, in-vehicle terminals, aircraft, unmanned vending terminals, and extended reality devices. Extended reality devices can include virtual reality devices, augmented reality devices, and mixed reality devices. A client application for the target application can be installed and run on the terminal devices. This target application can include, but is not limited to, financial transaction applications, payment applications, shopping applications, web browser applications, search applications, instant messaging tools, email clients, and social media platform software (these are just examples). Furthermore, this application embodiment does not limit the form of the target application, and it can include, but is not limited to, applications, mini-programs, etc., installed on the terminal devices, and can also be in the form of web pages.

[0031] Server 105 can be a server providing various services, such as a backend management server supporting websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (this is just an example). The backend management server can analyze and process received user requests and other data, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services such as cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks, and basic cloud computing services such as big data. The server can be the backend server of the aforementioned target application, used to provide backend services to the clients of the target application.

[0032] It should be noted that the enterprise customer risk assessment method provided in this application embodiment can generally be executed by server 105 and / or terminal devices 101-103. Accordingly, the enterprise customer risk assessment device provided in this application embodiment can generally be set in server 105 and / or terminal devices 101-103.

[0033] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0034] Figure 2 A flowchart illustrating an enterprise customer risk assessment method according to an embodiment of this application is shown schematically. Figure 2 As shown, the enterprise customer risk assessment method 200 according to the embodiments of this application may include steps S210 to S250.

[0035] In step S210, transaction flow data of the target enterprise customer and its related enterprise customers are obtained.

[0036] In the embodiments of this application, the target enterprise customer refers to the enterprise whose risk status is to be assessed, such as a company engaged in commodity trading (denoted as Company A). Related enterprise customers refer to other enterprises that have transactions with the target enterprise customer, such as suppliers, distributors, or partners of Company A (denoted as Company B, Company C, etc.). Transaction flow data refers to the details of inter-enterprise transfer transactions recorded in the banking system, including information such as the payer, payee, transaction amount, and transaction time.

[0037] For example, if Company A made 100 payments to Company B in the past year, each payment being 10 yuan, these records constitute transaction flow data.

[0038] In the embodiments of this application, transaction flow data of the target enterprise customer and its related enterprise customers are obtained. First, the target enterprise customer to be evaluated is identified, such as Company X, which manufactures electronic products. Then, the transaction flow data of Company X and all its known counterparties (i.e., related enterprise customers) over a past period are obtained. This data includes fields such as the payer's account, payee's account, transaction amount, and transaction time for each transaction. For example, the transaction records between Company X and companies Y and Z over the past month, totaling 10,000 transaction flow records, are obtained.

[0039] In step S220, the target enterprise customer and related enterprise customers are used as nodes. The transaction relationship between the nodes is determined based on the transaction flow data, and a transaction knowledge graph is constructed with the transaction relationship as the edge.

[0040] In the embodiments of this application, a node refers to an abstract point representing each enterprise in the transaction knowledge graph, such as company A corresponding to node A and company B corresponding to node B. A transaction relationship refers to the financial connection between two nodes determined based on transaction flow data, such as a transaction relationship between node A and node B, which may include attributes such as transaction frequency and total amount. An edge is a line segment connecting two nodes, used to visually represent the transaction relationship, such as a directed edge between node A and node B, with the arrow pointing from the payer to the payee.

[0041] In the embodiments of this application, the transaction knowledge graph is a network structure composed of nodes and edges, used to comprehensively reflect the transaction relationships between enterprises. For example, using companies A, B, C, etc. as nodes and their transaction relationships as edges, a graph containing all transaction information is constructed.

[0042] In the embodiments of this application, target enterprise customers and related enterprise customers are used as nodes. Transaction relationships between nodes are determined based on transaction flow data, and a transaction knowledge graph is constructed using these transaction relationships as edges. Company X and its related enterprise customers are abstracted as nodes, with each node corresponding to one enterprise. A transaction relationship between two nodes is determined based on the payer and payee of each transaction in the transaction flow data.

[0043] For example, if company X transfers money to company Y 10 times, there is a transaction relationship between node X and node Y. This relationship can record attributes such as the total transaction amount and the transaction frequency.

[0044] In the embodiments of this application, nodes represent enterprises and edges represent transaction relationships, constructing a complete transaction knowledge graph. For example, the constructed graph contains 51 nodes and hundreds of edges, each edge carrying attributes such as transaction frequency and amount.

[0045] In step S230, based on the transaction relationships between nodes in the transaction knowledge graph, the target enterprise customer and related enterprise customers are divided into multiple clusters. Different clusters correspond to different degrees of transaction relationship closeness. The target enterprise customer belongs to the first cluster. The transaction relationship closeness between the nodes in the first cluster and the target enterprise customer is higher than that between the nodes in other clusters and the target enterprise customer.

[0046] In the embodiments of this application, a cluster refers to different groups of nodes divided according to the closeness of their transaction relationships. For example, the companies that trade most frequently and have the closest financial dealings with Company A are grouped into the first cluster, and the companies with the next closest transactions are grouped into the second cluster. The closeness of the transaction relationship refers to a quantitative indicator that measures the frequency and scale of transactions between nodes, such as the frequency of two-way transactions and the total transaction amount. The first cluster is the cluster that includes the target company's customers and has the closest transaction relationships among its internal nodes. Other clusters refer to clusters outside the first cluster, whose nodes have a lower degree of transaction relationship with the target company's customers than the nodes in the first cluster.

[0047] In the embodiments of this application, based on the transaction relationships (such as bidirectional transaction frequency and transaction amount) between nodes in the knowledge graph, graph clustering or threshold partitioning methods are used to divide all enterprises into multiple hierarchical clusters. The transaction relationships between nodes in the graph are analyzed, and the closeness of the transaction relationship between each node and the target enterprise customer is calculated. The closeness can be comprehensively measured by indicators such as bidirectional transaction frequency and transaction amount ratio. The target customer is always located in the first cluster, and the closeness of the other clusters decreases sequentially.

[0048] For example, if Company X and Company Y exchange funds more than 100 times per month, and the amount accounts for 30% of X's total transaction volume, then X and Y are considered to have a very high degree of closeness; while Company X and Company W only have occasional small transactions, so their closeness is very low. Based on the degree of closeness, all nodes are divided into multiple clusters.

[0049] In step S240, risk event information for each node is obtained.

[0050] In the embodiments of this application, risk event information refers to negative event data related to the enterprise, such as administrative penalties, environmental penalties, and operational abnormalities.

[0051] In the embodiments of this application, risk event information for each node is obtained. Raw risk event data for all nodes (including target customers and associated customers) can be obtained through an external data interface. This data can be in text format, such as penalty documents, announcements, etc.

[0052] In step S250, based on the risk event information of nodes in each cluster and the weights corresponding to each cluster, the risk score of the target enterprise customer is calculated. The risk score represents the degree of impact of associated risks on the target enterprise customer due to the risk events of associated enterprise customers.

[0053] In the embodiments of this application, weight refers to a pre-set coefficient for different clusters, used to reflect the magnitude of the impact of risk events within that cluster on the target enterprise customer. Typically, the first cluster has the highest weight. The risk score is a numerical value used to quantify the degree of associated risk impact suffered by the target enterprise customer due to risk events of related enterprise customers. The degree of associated risk impact refers to the magnitude of the risk that may be transmitted to the target enterprise customer through the transaction chain due to a risk event occurring in a related enterprise.

[0054] In the embodiments of this application, a risk score for the target enterprise customer is calculated based on the risk event information of nodes in each cluster and the corresponding weight of each cluster. The risk event information (such as frequency and type) occurring on all nodes within each cluster is statistically analyzed, and a weighted calculation is performed based on the weights of each cluster to obtain the risk score for the target customer. The risk score reflects the degree of impact on the target customer due to the transmission of risk events from related customers.

[0055] Through the embodiments of this application, a knowledge graph is constructed using transaction flow data to accurately segment clusters of varying degrees of closeness. By integrating risk event information and differentiated weights of each cluster node, in-depth identification and quantitative assessment of implicitly related risks are achieved. The method of this embodiment can effectively discover transaction interest communities that cannot be captured by traditional static relationships, significantly improving the accuracy of risk transmission assessment and the timeliness of early warning.

[0056] Figure 3A flowchart illustrating a corporate customer risk assessment method according to another embodiment of this application is shown schematically. Figure 3 As shown in the embodiments of this application, the target enterprise customer and related enterprise customers are divided into multiple clusters according to the transaction relationship between nodes in the transaction knowledge graph, which may include steps S310 to S330.

[0057] In step S310, based on the transaction relationships between nodes in the transaction knowledge graph, transaction pairs whose bidirectional transaction frequencies both exceed a preset threshold are extracted, and multiple node sets are determined. The nodes in each node set are interconnected through transaction pairs.

[0058] In the embodiments of this application, the bidirectional transaction frequency refers to the number of times two nodes initiate transfers to each other. For example, if company A transfers funds to company B 10 times, and company B transfers funds to company A 8 times, then the bidirectional transaction frequencies are 10 and 8, respectively.

[0059] In the embodiments of this application, the preset threshold refers to a pre-set minimum transaction frequency value used to filter node pairs with frequent transactions. For example, the threshold is set to 5 times / month.

[0060] In the embodiments of this application, a transaction pair refers to a binary tuple consisting of two nodes that have a bidirectional transaction relationship. A node set refers to a group of nodes that are interconnected through transaction pairs. For example, if there is a bidirectional transaction pair between AB and BC, then A, B, and C belong to the same node set.

[0061] For example, in a transaction knowledge graph, if company X and company Y have transferred funds to each other 12 and 10 times respectively in the past six months, both exceeding the threshold of 5 times, then X and Y constitute a transaction pair. Similarly, X and Z, and Y and Z also constitute transaction pairs, and X, Y, and Z are interconnected through these transaction pairs, forming a set of nodes.

[0062] In the embodiments of this application, based on the transaction relationships between nodes in the transaction knowledge graph, transaction pairs whose bidirectional transaction frequencies both exceed a preset threshold are extracted to determine multiple node sets. All node pairs whose bidirectional transaction frequencies reach the preset threshold are filtered from the knowledge graph, and these node pairs are combined into multiple independent node sets according to their connectivity. Nodes within each set form a tight local network through bidirectional transaction pairs, while different sets have no or only sparse connections.

[0063] In step S320, the target node set where the target enterprise customer is located is determined from multiple node sets.

[0064] In the embodiments of this application, the target node set refers to the set of nodes that contains the target enterprise customer.

[0065] In embodiments of this application, the target node set containing the target enterprise customer is determined from multiple node sets. All node sets are traversed to find the unique set containing the target enterprise customer, which serves as the basis for subsequent partitioning.

[0066] For example, after searching, it was found that company X is located in set S, which contains 5 companies including X, Y, and Z.

[0067] In step S330, the target enterprise customers and related enterprise customers are divided into multiple clusters based on the closeness of their transaction relationship with the target node set.

[0068] In the embodiments of this application, the closeness of the transaction relationship is an indicator that measures the frequency of transactions and the degree of financial dependence between nodes, and can be determined by a combination of factors such as transaction frequency, amount, and bidirectionality. For example, the average daily transaction amount, the number of monthly transactions, etc.

[0069] In the embodiments of this application, target enterprise customers and associated enterprise customers are divided into multiple clusters based on the closeness of their transaction relationships with the target node set. Within and around the target node set, the target enterprise customers and all associated enterprise customers are divided into multiple clusters of different levels based on the closeness of their transaction relationships with the target enterprise customers (or with the core nodes within the target node set). The first cluster contains the nodes with the closest transaction relationships with the target customers, and the closeness of subsequent clusters decreases sequentially.

[0070] For example, the set S and other nodes that have transactions with it can be divided into three clusters: the first cluster (X, Y, Z), the second cluster (W, U), and the third cluster (V, T).

[0071] Through the embodiments of this application, by filtering through the two-way transaction frequency threshold, it is possible to accurately identify implicitly related groups that have frequent two-way transactions with target customers, filter out low-frequency noise transactions, provide a reliable data foundation for subsequent multi-level circle division, and ensure the accuracy and effectiveness of cluster division.

[0072] In some embodiments, target enterprise customers and associated enterprise customers are divided into multiple clusters based on the closeness of their transaction relationships with the target node set, including: classifying nodes in the target node set whose transaction relationship closeness is greater than or equal to a first threshold as a first cluster, the first cluster containing target enterprise customers; classifying nodes in the transaction knowledge graph that have direct transaction relationships with nodes in the first cluster and whose transaction relationship closeness with nodes in the first cluster is less than the first threshold but greater than or equal to a second threshold as a second cluster, the first threshold being greater than the second threshold; and classifying nodes in the transaction knowledge graph that have direct transaction relationships with nodes in the second cluster and are only indirectly associated with the first cluster through nodes in the second cluster as a third cluster.

[0073] In the embodiments of this application, the first threshold is a standard for defining the highest degree of closeness of the core trading circle, representing extremely frequent transactions and extremely high depth of capital flow.

[0074] In the embodiments of this application, nodes within the target node set whose transaction relationships are closer than or equal to a first threshold are classified into a first cluster. First, a screening process is performed within the already determined target node set to identify nodes with extremely frequent and large-scale transactions with each other. These nodes constitute the tightest defense circle or community of interests centered on the target enterprise, and are thus assigned to the first cluster.

[0075] Figure 4 A schematic diagram illustrating the principle of a business customer risk assessment method according to an embodiment of this application is shown.

[0076] In the embodiments of this application, it is assumed that "daily average fund transfers exceeding 10,000" is set as the first threshold. Figure 4 In the diagram, A, B, and C all belong to the target node set. Among them, there are "high-frequency two-way" transactions between A and B, and between A and C, and there are also "high-frequency two-way" transactions between B and C. Their density exceeds the first threshold. Therefore, A, B, and C are divided into the first cluster.

[0077] In the embodiments of this application, nodes in the transaction knowledge graph that have direct transaction relationships with nodes in the first cluster, and whose transaction relationship with nodes in the first cluster is less than a first threshold but greater than or equal to a second threshold, are classified into a second cluster. The first threshold is greater than the second threshold. Based on the first cluster, a layer is expanded outward to find nodes that have direct financial transactions with members of the first cluster. If the transaction closeness between these nodes and the first cluster is not as high as that within the cluster (below the first threshold), but still meets a certain standard (above the second threshold), they are considered important secondary related parties and are included in the second cluster.

[0078] For example, Figure 4 Nodes D and E are directly connected to the first cluster (A, B, C). D and E have direct transactions with nodes within the cluster. D and C have a direct transaction relationship, and E and B have a direct transaction relationship. The closeness of these transactions does not reach the first threshold, but exceeds the set second threshold (e.g., "average daily transaction amount greater than 5000"). Therefore, D and E are classified into the second cluster.

[0079] In the embodiments of this application, nodes in the transaction knowledge graph that have direct transaction relationships with nodes in the second cluster and are only indirectly associated with the first cluster through nodes in the second cluster are classified into a third cluster. The process continues to expand outwards to find nodes that directly transact with members of the second cluster. These nodes have no direct connection path to the first cluster.

[0080] For example, Figure 4 The nodes (F, G, and H) in the middle group have direct connections to the second cluster (D, E). F is connected to D, G is connected to E, and H is connected to F. F, G, and H do not have direct connections to the first cluster (A, B, C) and must reach A, B, and C through D or E. Therefore, F and G are classified as the third cluster.

[0081] Through the embodiments of this application, a concentric circle risk layer structure with the target customer as the core is constructed by setting a dual threshold progressive division, realizing refined risk classification and isolation from the core circle to the outer circle.

[0082] In some embodiments, obtaining risk event information for each node includes: obtaining the original text of risk events related to each node; performing structured parsing of the original text of risk events using a large language model to obtain structured risk event information; and associating the structured risk event information with the corresponding node.

[0083] In the embodiments of this application, the original text of a risk event refers to an unprocessed, unstructured document describing negative information about an enterprise, such as an administrative penalty decision, an environmental penalty announcement, or a notice of an abnormal business operation list.

[0084] In the embodiments of this application, the original text of risk events related to each node is obtained. Through an external data interface, the original documents of negative information related to all node enterprises are systematically obtained. The content of these documents is unstructured natural language.

[0085] In the embodiments of this application, a large language model is used to perform structured parsing of the original text of risk events to obtain structured risk event information. A large language model refers to a deep learning model that has been pre-trained on a large corpus and fine-tuned according to instructions, possessing text understanding and information extraction capabilities. Structured parsing refers to the process of converting unstructured text into formatted data with fixed fields. Structured risk event information refers to standardized record information containing fields such as subject identifier, event type, occurrence date, and severity.

[0086] In the embodiments of this application, a large language model, finely tuned with instructions, is used to extract information from the original text of each risk event. The model can understand the context, accurately extract key fields, and convert scattered text into machine-readable structured data. For example, when penalty text is input into the large language model, the model outputs structured fields.

[0087] In the embodiments of this application, entity linking technology can be used to precisely match the subject identifier (such as enterprise name, unified social credit code) in structured risk event information with nodes in the transaction knowledge graph. After a successful match, the structured information is stored as an additional attribute of the node, so that each node in the knowledge graph carries its associated risk event information.

[0088] In the embodiments of this application, structured risk event information is associated with corresponding nodes. For example, the entity name "XX Company" in the structured record can be matched with nodes in the knowledge graph. If the match is successful, the structured risk event information is associated with the corresponding node.

[0089] Through the embodiments of this application, unstructured risk event text is transformed into structured data using a large language model and accurately associated with corresponding nodes, thereby achieving in-depth mining and standardized processing of negative information and improving the reliability and accuracy of the enterprise customer risk assessment method of this embodiment.

[0090] In some embodiments, structured risk event information includes the subject identifier of the risk event; associating structured risk event information with corresponding nodes includes: matching the subject identifier with nodes in the transaction knowledge graph, and associating the structured risk event information with the successfully matched nodes.

[0091] In the embodiments of this application, the subject identifier refers to a key field used to uniquely identify the enterprise to which the risk event belongs, such as the enterprise name, unified social credit code, registration number, etc.

[0092] In the embodiments of this application, the subject identifier is matched with nodes in the transaction knowledge graph, and the structured risk event information is associated with the successfully matched nodes. The subject identifier (such as a company name or credit code) is extracted from the structured risk event information, and all nodes in the transaction knowledge graph are traversed to find nodes with the same identifier. If a unique matching node is found, the structured information is added to the risk event list of that node; if no match can be found, a new node may be created or marked as pending processing.

[0093] For example, matching is performed on 100 structured records one by one. Once a match is successful, the structured risk event information is stored as the attribute data of that node.

[0094] Through the embodiments of this application, by accurately matching the subject identifier, it is ensured that each risk event is accurately attributed to the corresponding enterprise node, avoiding association deviations caused by similar names or data errors, thereby improving the accuracy of subsequent risk assessment.

[0095] In some embodiments, the risk score of the target enterprise customer is calculated based on the risk event information of the nodes in each cluster and the weight corresponding to each cluster, including: obtaining the risk event information of the nodes in each cluster within a preset time range; and determining the risk score of the target enterprise customer based on the risk event information of the nodes in each cluster within the preset time range and the weight corresponding to each cluster.

[0096] In the embodiments of this application, the preset time range refers to a pre-defined retrospective period used to limit the timeliness of risk events. For example, the assessment baseline date is pushed back 30, 60, or 90 days. Risk event information refers to the structured negative event records associated with nodes, including event type, occurrence date, severity, etc.

[0097] In the embodiments of this application, risk event information of nodes in each cluster within a preset time range is obtained. Structured risk event information of all nodes within each cluster within a specified time window is retrieved from the transaction knowledge graph. Only risk events occurring within this time range are counted; older events outside the range are not included to ensure the timeliness of the scoring.

[0098] For example, the assessment baseline date is set as July 6, 2020, and the preset time range is 60 days (i.e., from May 7 to July 6, 2020). Risk event records for all nodes in the first cluster (X, Y, Z), the second cluster (W, U), and the third cluster (V, T) within this time range are retrieved from the knowledge graph.

[0099] In the embodiments of this application, the weight corresponding to a cluster refers to a pre-set risk transmission coefficient for different clusters, reflecting the degree of impact of the cluster's risk on the target customer. The first cluster has the highest weight, and subsequent clusters have progressively lower weights. The risk score is a numerical measure of the impact of associated risks on the target customer; a higher score indicates a greater risk.

[0100] In the embodiments of this application, a risk score for the target enterprise customer is determined based on risk event information of nodes in each cluster within a preset time range and the corresponding weights of each cluster. The risk event information of nodes within each cluster within the time window is aggregated (e.g., by counting the number of occurrences or calculating a severity weighted value). The aggregated result for each cluster is multiplied by its corresponding weight, and the weighted results of all clusters are summed to obtain the risk score for the target enterprise customer. This score comprehensively reflects the total impact of risk events from various levels of stakeholders on the target customer through different transmission intensities.

[0101] For example, the weight of the first cluster is 0.6, the weight of the second cluster is 0.3, and the weight of the third cluster is 0.1. Company Y in the first cluster received one administrative penalty (severity weight 1.0), and Company Z in the second cluster experienced one operational anomaly (severity weight 1.2). The risk score is then calculated as: (1 × 1.0 × 0.6) + (1 × 1.2 × 0.3) = 0.6 + 0.36 = 0.96.

[0102] Through the embodiments of this application, the timeliness of risk events is limited by a preset time range, avoiding interference from historical issues, allowing the scoring to focus on recent risk changes, and combining cluster-differentiated weights to accurately quantify the transmission impact of different levels of related risks on target customers, thereby improving the timeliness and accuracy of the scoring.

[0103] In some embodiments, risk event information includes the number of times a risk event occurs, its type, and the repetition of the same type of risk event on the same node. Based on the risk event information of nodes in each cluster within a preset time range and the weights corresponding to each cluster, the risk score of the target enterprise customer is determined, including: determining the risk contribution value of each cluster based on the number of times a risk event occurs on nodes in each cluster within a preset time range, the weights corresponding to different types of risk events, and the enhancement factor corresponding to multiple occurrences of the same type of risk event on the same node; and determining the risk score of the target enterprise customer based on the risk contribution value of each cluster and the weights corresponding to each cluster; wherein the weight corresponding to the first cluster is greater than the weights corresponding to other clusters.

[0104] In the embodiments of this application, the repeated occurrence of the same type of risk event at the same node refers to the record of the same node being penalized multiple times for the same reason within a time window.

[0105] For example, if Company Y receives two administrative penalties within a 60-day window, both for "tax evasion," then the number of penalties is 2, the type is "administrative penalty," and the repetition is "the same reason repeated twice."

[0106] In the embodiments of this application, the weights corresponding to different types of risk events refer to the severity coefficients preset for each type of risk event, such as an administrative penalty weight of 1.0 and an environmental penalty weight of 1.2. The enhancement factor refers to the additional amplification coefficient added when the same node experiences multiple risk events for the same reason within a time window, for example, β=1.5. The risk contribution value of the cluster refers to the total impact value of all risk events of all nodes in the cluster after adjustment by the type weights and enhancement factors.

[0107] For example, if company Y in the first cluster receives two administrative penalties (type weight 1.0), triggering the enhancement factor of 1.5 repeatedly for the same reason, then Y's contribution value in this cluster is: 2 × 1.0 × 1.5 = 3.0. If there are other nodes in the cluster, the total risk contribution value of the cluster is obtained by summing them up.

[0108] In the embodiments of this application, the risk contribution value of each cluster is determined based on the number of times risk events occur within a preset time range for each node in the cluster, the weights corresponding to different types of risk events, and the enhancement factor corresponding to the same node experiencing the same type of risk event multiple times. For each cluster, firstly, the risk events of all nodes within the cluster within the time window are counted, then a type weight (different types have different severity) is applied to each event. If the node experiences the same event multiple times, it is additionally multiplied by an enhancement factor (increased penalty). Finally, the weighted values ​​of all events within the cluster are summed to obtain the risk contribution value of the cluster.

[0109] In the embodiments of this application, the weight corresponding to the cluster refers to the coefficient that reflects the strength of the risk transmission of the cluster to the target customer, with the first cluster having the highest weight and the other clusters decreasing in sequence.

[0110] In the embodiments of this application, the risk score of the target enterprise customer is determined based on the risk contribution value of each cluster and the corresponding weight of each cluster. The risk contribution value of each cluster is multiplied by the weight corresponding to that cluster, and then the products of all clusters are added together to obtain the risk score of the target customer. The first cluster has the largest weight, and the risk events within it have the most significant impact on the score.

[0111] For example, suppose the risk contribution value of the first cluster is 3.0 (weight 0.6), the contribution value of the second cluster is 1.2 (weight 0.3), and the contribution value of the third cluster is 0.5 (weight 0.1). Then the risk score = 3.0 × 0.6 + 1.2 × 0.3 + 0.5 × 0.1 = 1.8 + 0.36 + 0.05 = 2.21.

[0112] Through the embodiments of this application, the severity differences of negative events of different natures and the cumulative effect of repeated offenses are accurately quantified. Combined with the differentiated weights of different social circles, the accuracy and reliability of the determined risk scores are effectively improved.

[0113] Based on the above-described enterprise customer risk assessment method, embodiments of this application also provide an enterprise customer risk assessment device. The following will be combined with... Figure 5 The device is described in detail.

[0114] Figure 5 A schematic block diagram of a business customer risk assessment apparatus according to an embodiment of this application is shown.

[0115] like Figure 5 As shown, the enterprise customer risk assessment device 500 of this embodiment includes a first acquisition module 510, a construction module 520, a division module 530, a second acquisition module 540, and a calculation module 550.

[0116] The first acquisition module 510 is used to acquire transaction flow data of the target enterprise customer and its related enterprise customers. In one embodiment, the first acquisition module 510 can be used to execute step S210 described above, which will not be repeated here.

[0117] The construction module 520 is used to determine the transaction relationships between target enterprise customers and related enterprise customers based on transaction flow data, and to construct a transaction knowledge graph with the transaction relationships as edges. In one embodiment, the construction module 520 can be used to execute step S220 described above, which will not be repeated here.

[0118] The segmentation module 530 is used to divide the target enterprise customer and related enterprise customers into multiple clusters based on the transaction relationships between nodes in the transaction knowledge graph. Different clusters correspond to different degrees of transaction relationship closeness. The target enterprise customer belongs to the first cluster, and the transaction relationship closeness between the nodes in the first cluster and the target enterprise customer is higher than that between the nodes in other clusters and the target enterprise customer. In one embodiment, the segmentation module 530 can be used to execute step S230 described above, which will not be repeated here.

[0119] The second acquisition module 540 is used to acquire risk event information of each node. In one embodiment, the second acquisition module 540 can be used to execute step S240 described above, which will not be repeated here.

[0120] The calculation module 550 is used to calculate the risk score of the target enterprise customer based on the risk event information of the nodes in each cluster and the corresponding weight of each cluster. The risk score represents the degree of impact of the associated risks suffered by the target enterprise customer due to the risk events of related enterprise customers. In one embodiment, the calculation module 550 can be used to perform step S250 described above, which will not be repeated here.

[0121] According to embodiments of this application, any multiple modules among the first acquisition module 510, construction module 520, partitioning module 530, second acquisition module 540, and calculation module 550 can be merged into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module. According to embodiments of this application, at least one of the first acquisition module 510, construction module 520, partitioning module 530, second acquisition module 540, and calculation module 550 can be at least partially implemented as hardware circuits, such as field-programmable gate arrays, programmable logic arrays, systems-on-a-chip, systems-on-a-substrate, systems-on-package, application-specific integrated circuits, or implemented by any other reasonable means of integrating or packaging circuits, or implemented by any one of software, hardware, and firmware, or by any appropriate combination of any of these three implementation methods. Alternatively, at least one of the first acquisition module 510, the construction module 520, the partitioning module 530, the second acquisition module 540, and the calculation module 550 may be implemented at least partially as a computer program module, which can perform corresponding functions when the computer program module is run.

[0122] In some embodiments, the partitioning module includes: a first determining submodule, used to extract transaction pairs whose bidirectional transaction frequencies both exceed a preset threshold based on the transaction relationships between nodes in the transaction knowledge graph, and to determine multiple node sets, wherein the nodes in each node set are interconnected through transaction pairs; a second determining submodule, used to determine the target node set where the target enterprise customer is located from the multiple node sets; and a partitioning submodule, used to partition the target enterprise customer and related enterprise customers into multiple clusters based on the closeness of their transaction relationships with the target node set.

[0123] In some embodiments, the partitioning submodule includes: a first partitioning unit, configured to partition nodes in the target node set whose transaction relationship closeness is greater than or equal to a first threshold into a first cluster, the first cluster including target enterprise customers; a second partitioning unit, configured to partition nodes in the transaction knowledge graph that have a direct transaction relationship with nodes in the first cluster and whose transaction relationship closeness with nodes in the first cluster is less than the first threshold but greater than or equal to a second threshold into a second cluster, the first threshold being greater than the second threshold; and a third partitioning unit, configured to partition nodes in the transaction knowledge graph that have a direct transaction relationship with nodes in the second cluster and are only indirectly associated with the first cluster through nodes in the second cluster into a third cluster.

[0124] In some embodiments, the second acquisition module includes: a first acquisition submodule, used to acquire the original text of risk events related to each node; a parsing submodule, used to perform structured parsing of the original text of risk events using a large language model to obtain structured risk event information; and an association submodule, used to associate the structured risk event information with the corresponding node.

[0125] In some embodiments, the structured risk event information includes the subject identifier of the risk event; the association submodule includes: an association unit, used to match the subject identifier with nodes in the transaction knowledge graph, and associate the structured risk event information with the successfully matched nodes.

[0126] In some embodiments, the calculation module includes: a second acquisition submodule, used to acquire risk event information of nodes in each cluster within a preset time range; and a third determination submodule, used to determine the risk score of the target enterprise customer based on the risk event information of nodes in each cluster within the preset time range and the weight corresponding to each cluster.

[0127] In some embodiments, risk event information includes the number of occurrences, types, and repetition of the same type of risk event on the same node; the third determining submodule includes: a first determining unit, used to determine the risk contribution value of each cluster based on the number of occurrences of risk events on nodes in each cluster within a preset time range, the weights corresponding to different types of risk events, and the enhancement factors corresponding to multiple occurrences of the same type of risk event on the same node; a second determining unit, used to determine the risk score of the target enterprise customer based on the risk contribution value of each cluster and the weight corresponding to each cluster; wherein, the weight corresponding to the first cluster is greater than the weight corresponding to other clusters.

[0128] Figure 6 A block diagram schematically illustrates an electronic device suitable for implementing a risk assessment method for enterprise customers according to an embodiment of this application.

[0129] like Figure 6 As shown, an electronic device 600 according to an embodiment of this application includes a processor 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory 602 or a program loaded from a storage portion 608 into a random access memory 603. The processor 601 may include, for example, a general-purpose microprocessor, an instruction set processor and / or an associated chipset and / or a dedicated microprocessor. The processor 601 may also include onboard memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for executing different steps of the method flow according to an embodiment of this application.

[0130] Random access memory 603 stores various programs and data required for the operation of electronic device 600. Processor 601, read-only memory 602, and random access memory 603 are interconnected via bus 604. Processor 601 executes various steps of the method flow according to embodiments of this application by executing programs in read-only memory 602 and / or random access memory 603. It should be noted that programs may also be stored in one or more memories other than read-only memory 602 and random access memory 603. Processor 601 may also execute various steps of the method flow according to embodiments of this application by executing programs stored in one or more memories.

[0131] According to embodiments of this application, the electronic device 600 may further include an input / output interface 605, which is also connected to a bus 604. The electronic device 600 may also include one or more of the following components connected to the input / output interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube, liquid crystal display, etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card, such as a local area network card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the input / output interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 610 as needed so that computer programs read from it can be installed into the storage section 608 as needed.

[0132] Embodiments of this application also provide a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of this application.

[0133] According to embodiments of this application, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including but not limited to: portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination thereof. In embodiments of this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this application, the computer-readable storage medium may include the read-only memory 602 described above, and / or random access memory 603, and / or one or more memories other than read-only memory 602 and random access memory 603.

[0134] Embodiments of this application also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to cause the computer system to implement the methods provided in the embodiments of this application.

[0135] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via the communication section 609, and / or installed from the removable medium 611. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0136] In embodiments of this application, the computer program can be downloaded and installed from a network via communication section 609, and / or installed from removable medium 611. When the computer program is executed by processor 601, it performs the functions defined in the system of embodiments of this application. According to embodiments of this application, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0137] According to embodiments of this application, program code for executing the computer programs provided in the embodiments of this application can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0138] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0139] Those skilled in the art will understand that the features described in the various embodiments of this application can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this application. In particular, the features described in the various embodiments of this application can be combined and / or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application.

Claims

1. A method for assessing enterprise customer risk, characterized in that, The method includes: Obtain transaction flow data of the target enterprise customer and its related enterprise customers; Using the target enterprise customer and the associated enterprise customer as nodes, the transaction relationship between the nodes is determined based on the transaction flow data, and a transaction knowledge graph is constructed using the transaction relationship as an edge. Based on the transaction relationships between nodes in the transaction knowledge graph, the target enterprise customer and the associated enterprise customer are divided into multiple clusters. Different clusters correspond to different degrees of transaction relationship tightness. The target enterprise customer belongs to the first cluster. The transaction relationship between the nodes in the first cluster and the target enterprise customer is tighter than that between the nodes in other clusters and the target enterprise customer. Obtain risk event information for each node; Based on the risk event information of nodes in each cluster and the corresponding weights of each cluster, the risk score of the target enterprise customer is calculated. The risk score represents the degree of impact of the associated risks suffered by the target enterprise customer due to the risk events of the associated enterprise customer.

2. The method according to claim 1, characterized in that, The step of dividing the target enterprise customer and the associated enterprise customer into multiple clusters based on the transaction relationships between nodes in the transaction knowledge graph includes: Based on the transaction relationships between nodes in the transaction knowledge graph, transaction pairs whose bidirectional transaction frequencies both exceed a preset threshold are extracted, and multiple node sets are determined. The nodes in each node set are interconnected through the transaction pairs. Determine the target node set where the target enterprise customer is located from the plurality of node sets; Based on the closeness of the transaction relationship with the target node set, the target enterprise customers and the associated enterprise customers are divided into multiple clusters.

3. The method according to claim 2, characterized in that, The step of dividing the target enterprise customers and the associated enterprise customers into multiple clusters based on the closeness of their transaction relationship with the target node set includes: The nodes in the target node set whose transaction relationship is closer than or equal to a first threshold are divided into a first cluster, and the first cluster includes the target enterprise customer. The nodes in the transaction knowledge graph that have a direct transaction relationship with the nodes in the first cluster and whose transaction relationship with the nodes in the first cluster is less than the first threshold but greater than or equal to the second threshold are classified into the second cluster, where the first threshold is greater than the second threshold. The nodes in the transaction knowledge graph that have direct transaction relationships with nodes in the second cluster and are only indirectly associated with the first cluster through nodes in the second cluster are classified into the third cluster.

4. The method according to claim 1, characterized in that, The acquisition of risk event information for each node includes: Obtain the raw text of risk events related to each node; The original text of the risk event is parsed using a large language model to obtain structured risk event information; The structured risk event information is associated with the corresponding node.

5. The method according to claim 4, characterized in that, The structured risk event information includes the subject identifier of the risk event; The step of associating the structured risk event information with the corresponding node includes: The entity identifier is matched with nodes in the transaction knowledge graph, and the structured risk event information is associated with the successfully matched nodes.

6. The method according to claim 1, characterized in that, The step of calculating the risk score of the target enterprise customer based on the risk event information of nodes in each cluster and the corresponding weight of each cluster includes: Obtain risk event information for nodes in each cluster within a preset time range; The risk score of the target enterprise customer is determined based on the risk event information of nodes in each cluster within the preset time range and the weight of each cluster.

7. The method according to claim 6, characterized in that, The risk event information includes the frequency and type of risk events, as well as the recurrence of similar risk events on the same node; the step of determining the risk score of the target enterprise customer based on the risk event information of nodes in each cluster within the preset time range and the weight corresponding to each cluster includes: The risk contribution value of each cluster is determined based on the number of risk events occurring in each node within the preset time range, the weights corresponding to different types of risk events, and the enhancement factor corresponding to the occurrence of the same type of risk event multiple times in the same node. The risk score of the target enterprise customer is determined based on the risk contribution value of each cluster and the corresponding weight of each cluster. The weight of the first cluster is greater than the weight of the other clusters.

8. A business customer risk assessment device, characterized in that, The device includes: The first acquisition module is used to acquire transaction flow data of the target enterprise customer and its related enterprise customers; The construction module is used to determine the transaction relationship between the target enterprise customer and the associated enterprise customer as nodes based on the transaction flow data, and to construct a transaction knowledge graph with the transaction relationship as the edge; The segmentation module is used to divide the target enterprise customer and the associated enterprise customer into multiple clusters based on the transaction relationships between nodes in the transaction knowledge graph. Different clusters correspond to different degrees of transaction relationship tightness. The target enterprise customer belongs to the first cluster. The transaction relationship between the nodes in the first cluster and the target enterprise customer is tighter than that between the nodes in other clusters and the target enterprise customer. The second acquisition module is used to acquire risk event information for each node; The calculation module is used to calculate the risk score of the target enterprise customer based on the risk event information of the nodes in each cluster and the weight corresponding to each cluster. The risk score represents the degree of impact of the associated risk on the target enterprise customer due to the risk events of the associated enterprise customer.

9. An electronic device, comprising: One or more processors; Memory, used to store one or more computer programs. The characteristic feature is that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 7.

11. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 7.

Images