Post-loan risk assessment method and device, electronic equipment and storage medium

Abstract

The invention provides a post-loan risk assessment method and device, electronic equipment and a storage medium. The post-loan risk assessment method comprises the following steps: firstly, obtaining supply chain information of a loan merchant; and then constructing a supply chain knowledge graph of the loan merchant based on the supply chain information. Performing feature extraction on the supply chain knowledge graph through a message passing neural network model to obtain a feature vector of a loan merchant and a feature vector of each supply chain merchant; and then, on the basis of the feature vector of the loan merchant and the feature vector of each supply chain merchant, a risk value of the loan merchant and a risk value of each supply chain merchant are calculated by using a pre-trained discriminator. And based on the risk value of the loan merchant and the risk value of each supply chain merchant, calculating the risk value of the supply chain where the loan merchant is located. And finally, carrying out weighted summation on the risk value of the loan merchant and the risk value of the supply chain where the loan merchant is located to obtain a final risk value of the loan merchant.

Description

Technical Field

[0001] The present application relates to the field of financial technology, and in particular to a post-loan risk assessment method, device, electronic device and storage medium. Background Art

[0002] In the field of supply chain financing financial services, regular testing of customer-related loan risks is an important part of post-loan management.

[0003] In the prior art, the assessment of merchant post-loan risk is usually conducted by modeling based on the information of a single merchant. However, in the supply chain scenario, a merchant usually has multiple upstream suppliers and downstream distributors. The current merchant may be affected by other merchants in the same supply chain, resulting in potential risks. Therefore, the accuracy of the assessment of merchant post-loan risk in the prior art is low. Summary of the invention

[0004] In view of this, the present application provides a post-loan risk assessment method, device, electronic device and storage medium to solve the problem of low accuracy in the assessment of merchant post-loan risk in the prior art.

[0005] To achieve the above objectives, this application provides the following technical solutions:

[0006] The first aspect of the present application discloses a post-loan risk assessment method, comprising:

[0007] Obtaining the supply chain information of the loan merchant; wherein the supply chain information includes data of each node in the supply chain where the loan merchant is located, and the nodes include core enterprises, suppliers, distributors, commodities, transactions, risk events, supply chain financing, monitoring and reminders, risk management strategies, credit ratings, supply chain depth, industry and technology risks;

[0008] Constructing a supply chain knowledge graph of the loan merchant based on the supply chain information;

[0009] Extracting features from the supply chain knowledge graph through a message passing neural network model to obtain a feature vector of the loan merchant and a feature vector of each supply chain merchant; wherein the supply chain merchant and the loan merchant are merchants in the same supply chain;

[0010] Based on the feature vector of the loan merchant and the feature vector of each supply chain merchant, a risk value of the loan merchant and a risk value of each supply chain merchant are calculated using a pre-trained discriminator; wherein the discriminator is trained based on a generative adversarial network;

[0011] Based on the risk value of the loan merchant and the risk value of each supply chain merchant, the risk value of the supply chain in which the loan merchant is located is calculated;

[0012] The risk value of the loan merchant and the risk value of the supply chain in which the loan merchant is located are weightedly summed to obtain a final risk value of the loan merchant.

[0013] Optionally, in the above method, constructing the supply chain knowledge graph of the loan merchant based on the supply chain information includes:

[0014] Defining node attributes of each of the nodes in the supply chain information;

[0015] Constructing node relationships between the nodes;

[0016] Each of the nodes, the node attributes and the node relationships are mapped into a graph database to construct a supply chain knowledge graph of the loan merchant.

[0017] Optionally, in the above method, the process of training the discriminator based on a generative adversarial network includes:

[0018] Get the historical feature vectors of historical loan merchants;

[0019] Performing dimensionality reduction processing on the historical feature vector to generate positive sample data;

[0020] Generate negative sample data from a priori Gaussian distribution through a generator in the generative adversarial network;

[0021] The discriminator is trained using the positive sample data and the negative sample data until the identification performance of the discriminator meets a preset condition.

[0022] Optionally, the above method, after performing weighted summation on the risk value of the loan merchant and the risk value of the supply chain in which the loan merchant is located to obtain the final risk value of the loan merchant, further comprises:

[0023] If the final risk value of the loan merchant is greater than a preset threshold, risk warning information of the loan merchant is generated, and a risk assessment is performed on the supply chain merchant.

[0024] The second aspect of the present application discloses a post-loan risk assessment device, comprising:

[0025] An acquisition unit is used to acquire supply chain information of the loan merchant; wherein the supply chain information includes data of various nodes in the supply chain where the loan merchant is located, and the nodes include core enterprises, suppliers, distributors, commodities, transactions, risk events, supply chain financing, monitoring and reminders, risk management strategies, credit ratings, supply chain depth, industry and technical risks;

[0026] A construction unit, used to construct a supply chain knowledge graph of the loan merchant based on the supply chain information;

[0027] An extraction unit is used to extract features from the supply chain knowledge graph through a message passing neural network model to obtain a feature vector of the loan merchant and a feature vector of each supply chain merchant; wherein the supply chain merchant and the loan merchant are merchants in the same supply chain;

[0028] A first calculation unit is used to calculate the risk value of the loan merchant and the risk value of each supply chain merchant based on the feature vector of the loan merchant and the feature vector of each supply chain merchant using a pre-trained discriminator; wherein the discriminator is trained based on a generative adversarial network;

[0029] A second calculation unit is used to calculate the risk value of the supply chain where the loan merchant is located based on the risk value of the loan merchant and the risk value of each supply chain merchant;

[0030] The third calculation unit is used to perform weighted summation on the risk value of the loan merchant and the risk value of the supply chain where the loan merchant is located, so as to obtain a final risk value of the loan merchant.

[0031] Optionally, the construction unit of the above-mentioned device includes:

[0032] A definition subunit, used to define node attributes of each of the nodes in the supply chain information;

[0033] A first construction subunit is used to construct a node relationship between each of the nodes;

[0034] The second construction subunit is used to map each of the nodes, the node attributes and the node relationships into a graph database to construct a supply chain knowledge graph of the loan merchant.

[0035] Optionally, in the above device, the first computing unit includes:

[0036] An acquisition subunit is used to acquire historical feature vectors of historical loan merchants;

[0037] A processing subunit, used for performing dimensionality reduction processing on the historical feature vector to generate positive sample data;

[0038] A generating subunit, used for generating negative sample data from a priori Gaussian distribution through a generator in the generative adversarial network;

[0039] The training subunit is used to train the discriminator using the positive sample data and the negative sample data until the discrimination performance of the discriminator meets a preset condition.

[0040] Optionally, the above device further includes:

[0041] The early warning unit is used to generate risk early warning information of the loan merchant if the final risk value of the loan merchant is greater than a preset threshold, and to conduct risk assessment on the supply chain merchant.

[0042] The third aspect of the present application discloses an electronic device, comprising:

[0043] one or more processors;

[0044] a storage device having one or more programs stored thereon;

[0045] When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of the first aspects of the present invention.

[0046] A fourth aspect of the present application discloses a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the method described in any one of the first aspects of the present invention is implemented.

[0047] It can be seen from the above technical scheme that in a post-loan risk assessment method provided by the present application, the supply chain information of the loan merchant is first obtained. Then, the supply chain knowledge graph of the loan merchant is constructed based on the supply chain information. Then, the supply chain knowledge graph is feature extracted through the message passing neural network model to obtain the feature vector of the loan merchant and the feature vector of each supply chain merchant; wherein, the supply chain merchant and the loan merchant are merchants in the same supply chain. Then, based on the feature vector of the loan merchant and the feature vector of each supply chain merchant, the risk value of the loan merchant and the risk value of each supply chain merchant are calculated using a pre-trained discriminator. Based on the risk value of the loan merchant and the risk value of each supply chain merchant, the risk value of the supply chain in which the loan merchant is located is calculated. Finally, the risk value of the loan merchant and the risk value of the supply chain in which the loan merchant is located are weighted and summed to obtain the final risk value of the loan merchant. It can be seen that by using the method of the present application, the knowledge graph is used to construct data relationships. When evaluating the post-loan risk of the loan merchant, the supply chain data is combined, and the influence of other merchants in the supply chain on the loan merchant is fully considered, so that the evaluation result is more accurate. The problem of low accuracy in the evaluation of the post-loan risk of merchants in the prior art is solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0049] Figure 1 A flowchart of a post-loan risk assessment method disclosed in an embodiment of the present application;

[0050] Figure 2 This is an example diagram of the supply chain knowledge graph disclosed in the embodiments of this application;

[0051] Figure 3 A flowchart of an implementation of the training process of the discriminator in step S104 disclosed in an embodiment of the present application;

[0052] Figure 4 A schematic diagram of a post-loan risk assessment device disclosed in an embodiment of the present application;

[0053] Figure 5 A schematic diagram of an electronic device disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0055] In this application, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element.

[0056] Furthermore, in this document, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0057] As can be seen from the background technology, in the prior art, the assessment of merchant post-loan risk is usually conducted by modeling based on the information of a single merchant. However, in the supply chain scenario, a merchant usually has multiple upstream suppliers and downstream distributors. The current merchant may be affected by other merchants in the same supply chain, resulting in potential risks. Therefore, the accuracy of the assessment of merchant post-loan risk in the prior art is low.

[0058] In view of this, the present application provides a post-loan risk assessment method, device, electronic device and storage medium to solve the problem of low accuracy in the assessment of merchant post-loan risk in the prior art.

[0059] The first aspect of the present application provides a method for assessing post-loan risk, such as Figure 1 As shown, specifically including:

[0060] S101. Obtain the supply chain information of the loan merchant; wherein the supply chain information includes data of each node in the supply chain where the loan merchant is located, and the nodes include core enterprises, suppliers, distributors, commodities, transactions, risk events, supply chain financing, monitoring and reminders, risk management strategies, credit ratings, supply chain depth, industry and technology risks.

[0061] It should be noted that the supply chain post-loan system has been built by the account manager with high-quality, reliable data organization, and the merchant's historical transaction data has been saved. Therefore, it has a natural advantage in building a knowledge graph for the supply chain to obtain high-quality, real graph data, and has a wider range of applications. First, obtain the supply chain information of the loan merchant; among them, the supply chain information includes the relevant data of each node in the supply chain where the loan merchant is located, such as suppliers, customers, products, transactions, historical data and other data. Nodes can include core enterprises, suppliers, distributors, commodities, transactions, risk events, supply chain financing, monitoring and reminders, risk management strategies, credit ratings, supply chain depth, industry and technology risks, etc.

[0062] S102. Construct a supply chain knowledge graph of loan merchants based on supply chain information.

[0063] It should be noted that after obtaining the supply chain information of the loan merchant, the scope of the supply chain can be clarified to ensure that the scope of the customer's supply chain is clearly defined. Then, a specialized graph database or graph data modeling tool is used to build a supply chain knowledge graph of the loan merchant based on the supply chain information.

[0064] Optionally, in another embodiment of the present application, an implementation of the above step S102 may include:

[0065] Define the node attributes of each node in the supply chain information.

[0066] Build node relationships between nodes.

[0067] Map each node, node attribute and node relationship into the graph database to build the supply chain knowledge graph of loan merchants.

[0068] It should be noted that first, the relevant attributes are defined for each node in the supply chain information. Each node has relevant attributes, such as company name, transaction date, risk type, etc. Then, based on the actual supply chain financing data, the relationship between the nodes is established. For example, the relationship between the core enterprise and the supplier is established to represent the supply chain cooperation relationship; the relationship between the commodity and the supplier is established to represent the supply situation of the product. These relationships will help to understand the interaction between the various participants in the supply chain. Each node, node attribute and node relationship is mapped to the graph database to construct the supply chain knowledge graph of the loan merchant. The constructed knowledge graph can be found in Figure 2 .

[0069] S103. Extract features from the supply chain knowledge graph through a message passing neural network model to obtain a feature vector of the loan merchant and a feature vector of each supply chain merchant; wherein the supply chain merchant and the loan merchant are merchants in the same supply chain.

[0070] It should be noted that in order to improve the accuracy of post-loan risk detection in the supply chain, this embodiment adopts a message passing neural network model (MPNN model). This model can combine the attributes of a node and its neighboring nodes to capture abnormal features and network representations. The message passing neural network model is used to extract features from the supply chain knowledge graph to obtain feature vectors of loan merchants and feature vectors of each supply chain merchant; among them, supply chain merchants and loan merchants are merchants in the same supply chain.

[0071] First, define the knowledge graph (attribute network) as G = {V, A, X}. Among them, V = {v1, ..., vn} represents the node (entity), X represents the attribute of the node, the dimension is n*m, and A is the adjacency matrix of whether there are links between nodes. The purpose of anomaly detection is to detect nodes whose patterns are significantly different from most reference instances in attributes and structure. Typical attribute networks carry rich information, including node attributes and links that describe the relationship between nodes. Due to the sample information representation in the form of graph nodes and edges, anomaly detection in attribute networks is more challenging than sample-based tasks.

[0072] In order to obtain effective embedding representations (feature representations) of entities and relations in the knowledge graph, Xavier uniform distribution is used for feature initialization, and messages are passed on the graph through the message passing neural network MPNN structure to update the embedding representation of the entity. In this process, the message passing mechanism plays a vital role, which mainly involves how to collect the feature information of the entity from its neighboring entities and use it to update the state and representation of the entity. MPNN uses the graph attention mechanism to implement this process. Specifically, each entity sends its feature information to its neighboring entities in the form of attention weights, and the neighboring entities aggregate the feature information of the entity according to these attention weights and send the aggregated feature information back to the original entity as a response. In this way, MPNN can achieve deep feature extraction of graph data in a forward propagation process to enhance the feature representation of the current node by fusing the effective features of the neighboring entities, thereby improving the accuracy and performance of anomaly detection in the knowledge graph.

[0073] S104. Based on the feature vector of the loan merchant and the feature vector of each supply chain merchant, a pre-trained discriminator is used to calculate the risk value of the loan merchant and the risk value of each supply chain merchant; wherein the discriminator is trained based on a generative adversarial network.

[0074] It should be noted that after obtaining the feature vector of the loan merchant and the feature vector of each supply chain merchant, each feature vector is input into a pre-trained discriminator for calculation to obtain the risk value of the loan merchant and the risk value of each supply chain merchant. The discriminator is trained based on a generative adversarial network (GAN network).

[0075] Optionally, in another embodiment of the present application, an implementation of step S104 is as follows: Figure 3 As shown, this may include:

[0076] S301, obtaining historical feature vectors of historical loan merchants.

[0077] It should be noted that the historical feature vectors of historical loan merchants are first obtained in order to generate training data for training the discriminator.

[0078] S302: Perform dimension reduction processing on the historical feature vector to generate positive sample data.

[0079] It should be noted that the historical feature vector is subjected to dimensionality reduction processing through the encoder to generate positive sample data.

[0080] S303. Generate negative sample data from a priori Gaussian distribution through a generator in a generative adversarial network.

[0081] It should be noted that negative samples are generated by the generator in the Generative Adversarial Network (GAN), that is, the generator generates negative samples by sampling random signals from the prior distribution and transforming them through the neural network. These negative samples are used by the discriminator to judge their authenticity.

[0082] S304: Use the positive sample data and the negative sample data to train the discriminator until the identification performance of the discriminator meets the preset conditions.

[0083] It should be noted that the task of the discriminator is to continuously improve its ability to identify positive and negative samples based on the supervisory signal. Therefore, in the continuous adversarial training process of GAN, the discriminator is trained using positive sample data and negative sample data. At this time, the generator and the discriminator will confront each other, jointly improving the generator's ability to generate high-quality negative samples, while improving the discriminator's ability to distinguish between positive and negative samples. The discriminator inputs a feature vector and outputs a risk value between 0 and 1, the larger the risk, and optimizes the parameters through the binary cross entropy loss function until the discriminator's identification performance meets the preset conditions, such as the identification accuracy rate reaches 95%.

[0084] S105. Based on the risk value of the loan merchant and the risk value of each supply chain merchant, calculate the risk value of the supply chain where the loan merchant is located.

[0085] It should be noted that the risk value of the loan merchant and the risk value of each supply chain merchant are calculated, and the risk value of the loan merchant and the risk value of each supply chain merchant are added together to obtain the risk value of the entire supply chain in which the loan merchant is located.

[0086] S106. Perform a weighted summation of the risk value of the loan merchant and the risk value of the supply chain in which the loan merchant is located to obtain a final risk value of the loan merchant.

[0087] It should be noted that based on the preset calculation formula, the risk value of the loan merchant and the risk value of the supply chain in which the loan merchant is located are weighted and summed to obtain the final risk value of the loan merchant, and the formula is as follows:

[0088] risk=βcur+(1-β)chain

[0089] Among them, risk is the final risk value of the loan merchant, cur is the risk value of the loan merchant calculated by the discriminator, chain is the risk value of the supply chain, and β is the preset weight.

[0090] In this embodiment, the merchant risk calculation process can effectively combine the risk of a single merchant with the risk of the entire supply chain financial system, thereby more comprehensively assessing the risk in the supply chain. The introduction of the hyperparameter β allows the weight to be adjusted to balance the impact of a single merchant's risk on the overall risk according to specific needs.

[0091] In a post-loan risk assessment method provided by an embodiment of the present application, first, the supply chain information of the loan merchant is obtained. Then, the supply chain knowledge graph of the loan merchant is constructed based on the supply chain information. Then, the supply chain knowledge graph is subjected to feature extraction through a message passing neural network model to obtain a feature vector of the loan merchant and a feature vector of each supply chain merchant; wherein the supply chain merchant and the loan merchant are merchants in the same supply chain. Then, based on the feature vector of the loan merchant and the feature vector of each supply chain merchant, the risk value of the loan merchant and the risk value of each supply chain merchant are calculated using a pre-trained discriminator. Based on the risk value of the loan merchant and the risk value of each supply chain merchant, the risk value of the supply chain in which the loan merchant is located is calculated. Finally, the risk value of the loan merchant and the risk value of the supply chain in which the loan merchant is located are weighted and summed to obtain the final risk value of the loan merchant. It can be seen from this that, by using the method of the present application, a knowledge graph is used to construct a data relationship, and when the loan merchant is subjected to post-loan risk assessment, the supply chain data is combined, and the influence of other merchants in the supply chain on the loan merchant is fully considered, so that the assessment result is more accurate. The problem of low accuracy in the assessment of the post-loan risk of merchants in the prior art is solved.

[0092] Optionally, in another embodiment of the present application, after executing step S106, the following may also be included:

[0093] If the final risk value of the loan merchant is greater than the preset threshold, risk warning information for the loan merchant is generated and a risk assessment is performed on the supply chain merchant.

[0094] It should be noted that if the calculated final risk of the loan merchant is greater than the preset threshold, the threshold can be set according to the actual situation, such as 0.85. This means that the risk of the loan merchant is relatively high at this time, and risk warning information for the loan merchant is generated. The above method is used to conduct risk assessment on merchants in the same supply chain as the loan merchant, so as to timely check and discover the risks of upstream and downstream merchants that may be associated, and issue early warnings to detect large-scale, multi-party, and potential defaults that may be caused by the credit default of important enterprises. At the same time, combined with other non-financial factors (such as industry, geographical location, etc.), the overall risk level of the business district where the core enterprise is located can also be evaluated.

[0095] Another embodiment of the present application also discloses a post-loan risk assessment device, such as Figure 4 As shown, specifically including:

[0096] The acquisition unit 401 is used to acquire the supply chain information of the loan merchant; wherein the supply chain information includes data of each node in the supply chain where the loan merchant is located, and the nodes include core enterprises, suppliers, distributors, commodities, transactions, risk events, supply chain financing, monitoring and reminders, risk management strategies, credit ratings, supply chain depth, industry and technical risks.

[0097] The construction unit 402 is used to construct a supply chain knowledge graph of the loan merchant based on the supply chain information.

[0098] The extraction unit 403 is used to extract features from the supply chain knowledge graph through a message passing neural network model to obtain a feature vector of the loan merchant and a feature vector of each supply chain merchant; wherein the supply chain merchant and the loan merchant are merchants in the same supply chain.

[0099] The first calculation unit 404 is used to calculate the risk value of the loan merchant and the risk value of each supply chain merchant based on the feature vector of the loan merchant and the feature vector of each supply chain merchant using a pre-trained discriminator; wherein the discriminator is trained based on a generative adversarial network.

[0100] The second calculation unit 405 is used to calculate the risk value of the supply chain where the loan merchant is located based on the risk value of the loan merchant and the risk value of each supply chain merchant.

[0101] The third calculation unit 406 is used to perform weighted summation on the risk value of the loan merchant and the risk value of the supply chain in which the loan merchant is located, so as to obtain a final risk value of the loan merchant.

[0102] In this embodiment, the specific execution process of the acquisition unit 401, the construction unit 402, the extraction unit 403, the first calculation unit 404, the second calculation unit 405 and the third calculation unit 406 can be referred to the corresponding Figure 1 The content of the method embodiment will not be repeated here.

[0103] In a post-loan risk assessment device provided by an embodiment of the present application, first, the acquisition unit 401 acquires the supply chain information of the loan merchant. Then the construction unit 402 constructs the supply chain knowledge graph of the loan merchant based on the supply chain information. The extraction unit 403 then extracts features from the supply chain knowledge graph through a message passing neural network model to obtain a feature vector of the loan merchant and a feature vector of each supply chain merchant; wherein the supply chain merchant and the loan merchant are merchants in the same supply chain. Then, the first calculation unit 404 calculates the risk value of the loan merchant and the risk value of each supply chain merchant based on the feature vector of the loan merchant and the feature vector of each supply chain merchant using a pre-trained discriminator. The second calculation unit 405 calculates the risk value of the supply chain where the loan merchant is located based on the risk value of the loan merchant and the risk value of each supply chain merchant. Finally, the third calculation unit 406 performs a weighted summation of the risk value of the loan merchant and the risk value of the supply chain where the loan merchant is located to obtain the final risk value of the loan merchant. It can be seen from this that by using the method of the present application, a knowledge graph is used to construct a data relationship. When the loan merchant is evaluated for post-loan risk, the supply chain data is combined, and the influence of other merchants in the supply chain on the loan merchant is fully considered, so that the evaluation result is more accurate. The problem of low accuracy in the assessment of merchant post-loan risk in the prior art is solved.

[0104] Optionally, in another embodiment of the present application, an implementation of the construction unit 402 includes:

[0105] The definition subunit is used to define the node attributes of each node in the supply chain information.

[0106] The first construction subunit is used to construct the node relationship between various nodes.

[0107] The second construction subunit is used to map each node, node attribute and node relationship into the graph database to construct the supply chain knowledge graph of the loan merchant.

[0108] In this embodiment, the specific execution process of the definition subunit, the first construction subunit and the second construction subunit can refer to the corresponding method embodiment described above, which will not be repeated here.

[0109] Optionally, in another embodiment of the present application, an implementation of the first calculation unit 404 includes:

[0110] The acquisition subunit is used to obtain the historical feature vectors of historical loan merchants.

[0111] The processing subunit is used to perform dimension reduction processing on the historical feature vector to generate positive sample data.

[0112] The generation subunit is used to generate negative sample data from the prior Gaussian distribution through the generator in the generative adversarial network.

[0113] The training subunit is used to train the discriminator using positive sample data and negative sample data until the identification performance of the discriminator meets the preset conditions.

[0114] In this embodiment, the specific execution process of the acquisition subunit, the processing subunit, the generation subunit and the training subunit can be found in the corresponding Figure 3 The content of the method embodiment will not be repeated here.

[0115] Optionally, in another embodiment of the present application, the above-mentioned post-loan risk assessment device may further include:

[0116] The early warning unit is used to generate risk warning information for the loan merchant if the final risk value of the loan merchant is greater than a preset threshold, and to conduct risk assessment on the supply chain merchant.

[0117] In this embodiment, the specific execution process of the early warning unit can refer to the corresponding method embodiment described above, which will not be repeated here.

[0118] Another embodiment of the present application further provides an electronic device, such as Figure 5 As shown, specifically including:

[0119] One or more processors 501 .

[0120] The storage device 502 stores one or more programs.

[0121] When one or more programs are executed by one or more processors 501 , the one or more processors 501 implement any one of the methods in the above embodiments.

[0122] Another embodiment of the present application further provides a computer storage medium on which a computer program is stored, wherein when the computer program is executed by a processor, any one of the methods in the above embodiments is implemented.

[0123] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can refer to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without creative work.

[0124] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0125] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A post-loan risk assessment method, characterized in that: include: Obtaining the supply chain information of the loan merchant; wherein the supply chain information includes data of each node in the supply chain where the loan merchant is located, and the nodes include core enterprises, suppliers, distributors, commodities, transactions, risk events, supply chain financing, monitoring and reminders, risk management strategies, credit ratings, supply chain depth, industry and technology risks; Constructing a supply chain knowledge graph of the loan merchant based on the supply chain information; Extracting features from the supply chain knowledge graph through a message passing neural network model to obtain a feature vector of the loan merchant and a feature vector of each supply chain merchant; wherein the supply chain merchant and the loan merchant are merchants in the same supply chain; Based on the feature vector of the loan merchant and the feature vector of each supply chain merchant, a risk value of the loan merchant and a risk value of each supply chain merchant are calculated using a pre-trained discriminator; wherein the discriminator is trained based on a generative adversarial network; Based on the risk value of the loan merchant and the risk value of each supply chain merchant, the risk value of the supply chain in which the loan merchant is located is calculated; The risk value of the loan merchant and the risk value of the supply chain in which the loan merchant is located are weightedly summed to obtain a final risk value of the loan merchant.

2. The method according to claim 1, characterized in that The step of constructing the supply chain knowledge graph of the loan merchant based on the supply chain information includes: Defining node attributes of each of the nodes in the supply chain information; Constructing node relationships between the nodes; Each of the nodes, the node attributes and the node relationships are mapped into a graph database to construct a supply chain knowledge graph of the loan merchant.

3. The method according to claim 1, characterized in that The process of training the discriminator based on a generative adversarial network includes: Get the historical feature vectors of historical loan merchants; Performing dimensionality reduction processing on the historical feature vector to generate positive sample data; Generate negative sample data from a priori Gaussian distribution through a generator in the generative adversarial network; The discriminator is trained using the positive sample data and the negative sample data until the identification performance of the discriminator meets a preset condition.

4. The method according to claim 1, characterized in that: After performing weighted summation on the risk value of the loan merchant and the risk value of the supply chain where the loan merchant is located to obtain the final risk value of the loan merchant, the method further includes: If the final risk value of the loan merchant is greater than a preset threshold, risk warning information of the loan merchant is generated, and a risk assessment is performed on the supply chain merchant.

5. A post-loan risk assessment device, characterized in that: include: An acquisition unit is used to acquire supply chain information of the loan merchant; wherein the supply chain information includes data of various nodes in the supply chain where the loan merchant is located, and the nodes include core enterprises, suppliers, distributors, commodities, transactions, risk events, supply chain financing, monitoring and reminders, risk management strategies, credit ratings, supply chain depth, industry and technical risks; A construction unit, used to construct a supply chain knowledge graph of the loan merchant based on the supply chain information; An extraction unit is used to extract features from the supply chain knowledge graph through a message passing neural network model to obtain a feature vector of the loan merchant and a feature vector of each supply chain merchant; wherein the supply chain merchant and the loan merchant are merchants in the same supply chain; A first calculation unit is used to calculate the risk value of the loan merchant and the risk value of each supply chain merchant based on the feature vector of the loan merchant and the feature vector of each supply chain merchant using a pre-trained discriminator; wherein the discriminator is trained based on a generative adversarial network; A second calculation unit is used to calculate the risk value of the supply chain where the loan merchant is located based on the risk value of the loan merchant and the risk value of each supply chain merchant; The third calculation unit is used to perform weighted summation on the risk value of the loan merchant and the risk value of the supply chain where the loan merchant is located, so as to obtain a final risk value of the loan merchant.

6. The device according to claim 5, characterized in that The construction unit comprises: A definition subunit, used to define node attributes of each of the nodes in the supply chain information; A first construction subunit is used to construct a node relationship between each of the nodes; The second construction subunit is used to map each of the nodes, the node attributes and the node relationships into a graph database to construct a supply chain knowledge graph of the loan merchant.

7. The device according to claim 5, characterized in that The first computing unit comprises: An acquisition subunit is used to acquire historical feature vectors of historical loan merchants; A processing subunit, used for performing dimensionality reduction processing on the historical feature vector to generate positive sample data; A generating subunit, used for generating negative sample data from a priori Gaussian distribution through a generator in the generative adversarial network; The training subunit is used to train the discriminator using the positive sample data and the negative sample data until the discrimination performance of the discriminator meets a preset condition.

8. The device according to claim 5, characterized in that Also includes: The early warning unit is used to generate risk early warning information of the loan merchant if the final risk value of the loan merchant is greater than a preset threshold, and to conduct risk assessment on the supply chain merchant.

9. An electronic device, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the method according to any one of claims 1 to 4.

10. A computer storage medium, characterized in that: A computer program is stored thereon, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images