Risk assessment method, device, equipment and computer-readable storage medium

Abstract

The present invention discloses a risk assessment method, device, equipment and computer-readable storage medium, the method comprising: obtaining the degree of association between entities of various entity types, and obtaining the risk label values ​​corresponding to the entities of the various entity types, wherein one entity type is a customer type; performing at least one round of iterative update calculation with the risk label value as the initial risk value, obtaining the target risk values ​​corresponding to the entities of the various entity types, and obtaining the risk value corresponding to the customer according to the target risk value corresponding to the entity of the customer type. Compared with the traditional scoring card model, the risk assessment scheme in the present invention does not need to extract the risk factor of the customer, and only needs to obtain the degree of association between the entities and the risk label value of each entity to achieve the risk assessment of the customer.

Description

Technical Field

[0001] The present invention relates to the field of risk control technology, and in particular to a risk assessment method, device, equipment and computer-readable storage medium. Background Art

[0002] In many application scenarios, it is necessary to conduct risk assessment on customers for risk control. The existing mainstream solution is based on the idea of ​​the traditional scoring card model: first obtain a sample of risky target Y, and extract the risk-related factor X corresponding to the sample, then perform variable derivation and other algorithm processing on X, and use the supervised algorithm to train the model F so that the result of F(X) is as consistent as possible with Y, and finally apply F to all customer groups to give a score. However, this traditional scoring card model requires knowing the risk factor X of the sample customer. When the risk factor X of the sample customer is missing, the trained model will have poor scoring effect or even fail to train the model. Summary of the invention

[0003] The main purpose of the present invention is to provide a risk assessment method, device, equipment and computer-readable storage medium, aiming to solve the technical problem that when the risk factor X of the sample customer is missing, the trained model scoring effect will be poor or even the model cannot be trained.

[0004] To achieve the above object, the present invention provides a risk assessment method, which comprises the following steps:

[0005] Obtaining the degree of association between entities of various entity types, and obtaining the risk label values ​​corresponding to the entities of the various entity types, wherein one entity type is a customer type;

[0006] Perform at least one round of iterative update calculation using the risk tag value as the initial risk value to obtain target risk values ​​corresponding to entities of various entity types, and obtain the risk value corresponding to the customer based on the target risk value corresponding to the customer type entity;

[0007] The steps of performing a round of iterative update calculation include:

[0008] For the entities to be updated among the entities of the various entity types, at least the degree of association between the entities of the various entity types and the entities to be updated is used as a weight, and the risk values ​​corresponding to the entities of the various entity types are weighted averaged to obtain a risk average result, and the risk value of the entity to be updated after this round of iterative update is obtained based on the risk average result.

[0009] Optionally, the step of taking at least the association degree between the entities of the various entity types and the entity to be updated as weight and performing weighted averaging on the risk values ​​corresponding to the entities of the various entity types to obtain the risk average result comprises:

[0010] For the entity type to be calculated among the various entity types, taking the association degree between the entity of the entity type to be calculated and the entity to be updated as the weight, weighted average is performed on the risk values ​​corresponding to the entities of the current entity type to be calculated, so as to obtain the calculation result corresponding to the entity type to be calculated;

[0011] Obtaining preset type weights corresponding to the entity types to which the entity to be updated belongs for each of the entity types;

[0012] According to the type weight, at least the calculation results corresponding to the various entity types are weighted averaged to obtain a risk average result.

[0013] Optionally, the step of performing weighted averaging of at least the calculation results corresponding to the various entity types according to the type weights to obtain a risk average result includes:

[0014] Obtaining a preset label weight corresponding to the risk label value of the entity to be updated;

[0015] According to the type weight and the label weight, the calculation results corresponding to the various entity types and the risk label values ​​of the to-be-updated entities are weighted averaged to obtain a risk average result.

[0016] Optionally, before the step of performing at least one round of iterative update calculation using the risk label value as the initial risk value to obtain the target risk values ​​corresponding to the entities of the various entity types, the step further includes:

[0017] Obtaining a risk impact value of a first entity type on a second entity type among the various entity types;

[0018] A type weight of the first entity type relative to the second entity type is set according to the risk impact value, wherein a larger risk impact value corresponds to a larger type weight.

[0019] Optionally, after the step of obtaining the degree of association between entities of various entity types, the method further includes:

[0020] Normalizing the association matrix formed by the association degree between entities of any two entity types among the various entity types to obtain the normalized association degree between entities of any two entity types;

[0021] The step of taking at least the association degree between the entities of the various entity types and the entity to be updated as weight and performing weighted averaging on the risk values ​​corresponding to the entities of the various entity types to obtain the risk average result comprises:

[0022] At least taking the normalized association degree between the entities of the various entity types and the entity to be updated as a weight, weighted averaging is performed on the risk values ​​corresponding to the entities of the various entity types to obtain a risk average result.

[0023] Optionally, the step of obtaining risk label values ​​corresponding to entities of various entity types respectively includes:

[0024] Obtain risk information corresponding to entities of various entity types respectively;

[0025] Setting risk label values ​​corresponding to entities of various entity types respectively according to the risk information;

[0026] When the risk information corresponding to an entity to be set among the entities of the various entity types is empty, the risk label value corresponding to the entity to be set is set to 0.

[0027] Optionally, after performing a round of iterative update calculation steps, the following steps are further included:

[0028] Detect whether the preset stop condition of the iterative update calculation is met;

[0029] If satisfied, the iterative update calculation is terminated, and the risk values ​​of the entities of the various entity types after the iterative update in this round are used as the target risk values;

[0030] If not satisfied, the next round of iterative update calculation is performed based on the risk values ​​of the entities of the various entity types after the current round of iterative update.

[0031] To achieve the above object, the present invention further provides a risk assessment device, comprising:

[0032] An acquisition module, used to acquire the degree of association between entities of various entity types, and to acquire the risk label values ​​corresponding to the entities of various entity types, wherein one entity type is a customer type;

[0033] An iterative update module, configured to perform at least one round of iterative update calculation using the risk tag value as the initial risk value to obtain target risk values ​​corresponding to the entities of the various entity types, and obtain the risk value corresponding to the customer according to the target risk value corresponding to the entity of the customer type;

[0034] Among them, performing a round of iterative update calculation specifically includes:

[0035] For the entities to be updated among the entities of the various entity types, at least the degree of association between the entities of the various entity types and the entities to be updated is used as a weight, and the risk values ​​corresponding to the entities of the various entity types are weighted averaged to obtain a risk average result, and the risk value of the entity to be updated after this round of iterative update is obtained based on the risk average result.

[0036] To achieve the above objectives, the present invention also provides a risk assessment device, which includes: a memory, a processor, and a risk assessment program stored in the memory and executable on the processor, wherein the risk assessment program implements the steps of the risk assessment method described above when executed by the processor.

[0037] In addition, to achieve the above-mentioned purpose, the present invention also proposes a computer-readable storage medium, on which a risk assessment program is stored, and when the risk assessment program is executed by a processor, the steps of the risk assessment method described above are implemented.

[0038] In the present invention, by obtaining the degree of association between entities of various entity types, and obtaining the risk label values ​​corresponding to the entities of various entity types respectively; performing at least one round of iterative update calculation with the risk label value as the initial risk value, the target risk values ​​corresponding to the entities of various entity types are obtained, and the risk value corresponding to the customer is obtained according to the target risk value corresponding to the entity of the customer type. Through at least one round of iterative update calculation, and by taking the degree of association between each entity and the entity to be updated as the weight in each round of iterative update calculation, the risk value of each entity is weighted averaged to obtain the risk value of the entity to be updated, so as to gradually transfer the risk value of each entity associated with the entity to be updated to the entity to be updated, thereby realizing the risk assessment of each entity to be updated, and then obtaining the risk assessment result of the customer. Compared with the traditional scoring card model, the above risk assessment method in the present invention does not need to extract the risk factor of the customer, but only needs to obtain the degree of association between the entities, and the entity type is not limited except for the customer type, and the number of entities under each entity type is not limited, so it will not be affected by insufficient collection of specific data. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A schematic diagram of the structure of the hardware operating environment involved in the embodiment of the present invention;

[0040] Figure 2 This is a schematic diagram of the process of the first embodiment of the risk assessment method of the present invention;

[0041] Figure 3 Schematic diagram of functional modules of a preferred embodiment of the risk assessment device of the present invention.

[0042] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0043] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0044] like Figure 1 As shown, Figure 1 It is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention.

[0045] It should be noted that the risk assessment device in the embodiment of the present invention can be a smart phone, a personal computer, a server and other devices, and no specific limitation is made here.

[0046] like Figure 1 As shown, the risk assessment device may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0047] Those skilled in the art will understand that Figure 1 The device structure shown in the figure does not constitute a limitation on the risk assessment device, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0048] like Figure 1 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a risk assessment program. The operating system is a program that manages and controls the hardware and software resources of the device and supports the operation of the risk assessment program and other software or programs. Figure 1 In the device shown, the user interface 1003 is mainly used for data communication with the client; the network interface 1004 is mainly used for establishing a communication connection with the server; and the processor 1001 can be used to call the risk assessment program stored in the memory 1005 and perform the following operations:

[0049] Obtaining the degree of association between entities of various entity types, and obtaining the risk label values ​​corresponding to the entities of the various entity types, wherein one entity type is a customer type;

[0050] Perform at least one round of iterative update calculation using the risk tag value as the initial risk value to obtain target risk values ​​corresponding to entities of various entity types, and obtain the risk value corresponding to the customer based on the target risk value corresponding to the customer type entity;

[0051] The operations for performing a round of iterative update calculation include:

[0052] For the entities to be updated among the entities of the various entity types, at least the degree of association between the entities of the various entity types and the entities to be updated is used as a weight, and the risk values ​​corresponding to the entities of the various entity types are weighted averaged to obtain a risk average result, and the risk value of the entity to be updated after this round of iterative update is obtained based on the risk average result.

[0053] Furthermore, the operation of taking at least the association degree between the entities of the various entity types and the entity to be updated as a weight and performing weighted averaging on the risk values ​​corresponding to the entities of the various entity types to obtain a risk average result includes:

[0054] For the entity type to be calculated among the various entity types, taking the association degree between the entity of the entity type to be calculated and the entity to be updated as the weight, weighted average is performed on the risk values ​​corresponding to the entities of the current entity type to be calculated, so as to obtain the calculation result corresponding to the entity type to be calculated;

[0055] Obtaining preset type weights corresponding to the entity types to which the entity to be updated belongs for each of the entity types;

[0056] According to the type weight, at least the calculation results corresponding to the various entity types are weighted averaged to obtain a risk average result.

[0057] Furthermore, the operation of performing weighted averaging of at least the calculation results corresponding to the various entity types according to the type weights to obtain a risk average result includes:

[0058] Obtaining a preset label weight corresponding to the risk label value of the entity to be updated;

[0059] According to the type weight and the label weight, the calculation results corresponding to the various entity types and the risk label values ​​of the to-be-updated entities are weighted averaged to obtain a risk average result.

[0060] Furthermore, before performing at least one round of iterative update calculation using the risk label value as the initial risk value to obtain the target risk values ​​corresponding to the entities of the various entity types, the processor 1001 may also be used to call the risk assessment program stored in the memory 1005 to perform the following operations:

[0061] Obtaining a risk impact value of a first entity type on a second entity type among the various entity types;

[0062] A type weight of the first entity type relative to the second entity type is set according to the risk impact value, wherein a larger risk impact value corresponds to a larger type weight.

[0063] Furthermore, after the operation of obtaining the association degree between entities of various entity types, the processor 1001 may also be used to call the risk assessment program stored in the memory 1005 to perform the following operations:

[0064] Normalizing the association matrix formed by the association degree between entities of any two entity types among the various entity types to obtain the normalized association degree between entities of any two entity types;

[0065] The operation of taking at least the association degree between the entities of the various entity types and the entity to be updated as a weight and performing weighted averaging on the risk values ​​corresponding to the entities of the various entity types to obtain the risk average result comprises:

[0066] At least taking the normalized association degree between the entities of the various entity types and the entity to be updated as a weight, weighted averaging is performed on the risk values ​​corresponding to the entities of the various entity types to obtain a risk average result.

[0067] Furthermore, the operation of obtaining the risk label values ​​corresponding to the entities of the various entity types includes:

[0068] Obtain risk information corresponding to entities of various entity types respectively;

[0069] Setting risk label values ​​corresponding to entities of various entity types respectively according to the risk information;

[0070] When the risk information corresponding to an entity to be set among the entities of the various entity types is empty, the risk label value corresponding to the entity to be set is set to 0.

[0071] Furthermore, after performing a round of iterative update calculation operations, the processor 1001 may also be used to call the risk assessment program stored in the memory 1005 to perform the following operations:

[0072] Detect whether the preset stop condition of the iterative update calculation is met;

[0073] If satisfied, the iterative update calculation is terminated, and the risk values ​​of the entities of the various entity types after the iterative update in this round are used as the target risk values;

[0074] If not satisfied, the next round of iterative update calculation is performed based on the risk values ​​of the entities of the various entity types after the current round of iterative update.

[0075] Based on the above structure, various embodiments of the risk assessment method are proposed.

[0076] Reference Figure 2 , Figure 2 Schematic diagram of the first embodiment of the risk assessment method of the present invention.

[0077] The embodiments of the present invention provide embodiments of the risk assessment method. It should be noted that although the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than here. In this embodiment, the execution subject of the risk assessment method may be a smart phone, a personal computer, a server or other device. For ease of description, the execution subject is omitted below to explain each embodiment. The risk assessment method includes:

[0078] Step S10, obtaining the association degree between entities of various entity types, and obtaining the risk label values ​​corresponding to the entities of various entity types, wherein one entity type is a customer type;

[0079] Customers are individuals, companies or organizations in business scenarios. For example, they can be specific individuals who apply for and use loans, and can be distinguished and identified by some ID (such as ID number, mobile phone number, etc.). In business scenarios, customers are associated with an entity due to some common reasons or needs. This entity is called an aggregate entity. Aggregate entities can be physical or virtual concepts. For example, if several customers apply for loans from a bank, the bank is an aggregate entity; if multiple customers click on a loan promotion link at a certain point in time, the link is an aggregate entity. Aggregate entities can be divided into different entity types. For example, a bank can be regarded as an entity type, and "xx Bank" is an entity under the bank type.

[0080] There are many kinds of associations between customers, customers and aggregated entities, and aggregated entities and aggregated entities. If customers and different types of aggregated entities are regarded as nodes, the association relationship is regarded as an edge, and the strength of the association between customers and aggregated entities is set as the weight of the edge, a graph structure can be obtained. Since there are multiple types of nodes in the graph, the graph structure obtained is a heterogeneous graph. The risk label value of the risk-represented customer is marked in the heterogeneous graph, and the risk value of all customers can be inferred using a semi-supervised iterative algorithm.

[0081] Specifically, in this embodiment, the customer is also regarded as an entity type, that is, "xx customer" is an entity under the customer type.

[0082] When performing risk assessment, the degree of association between entities of various entity types can be obtained. Among them, one type of entity type is a customer type, that is, it is necessary to conduct risk assessment on the customer; other types of entities can be manually defined or extracted from business data. Each entity type includes at least one entity, and obtaining the degree of association between entities of various entity types refers to obtaining the degree of association between any two entities under all entity types. Optionally, the two entities can be the same, that is, an association degree can also be set for the entity and itself. The degree of association between two entities can be set manually or extracted from business data, which is not limited in this embodiment; for example, when the value of the degree of association is set to 0 / 1, 0 is used to indicate that there is no association between the two entities, and 1 is used to indicate that there is an association between the two entities. If a piece of business data is that customer k successfully applies for a loan from bank l, the degree of association between customer k and bank l can be set to 1, indicating that customer k and bank l are associated; if there is no association relationship between the two entities or the association relationship between the two entities cannot be determined, the degree of association between the two entities can be set to 0. As required, in one embodiment, the degree of association may also be other value ranges to indicate the degree of association between two entities. The value range of the degree of association is not limited in this embodiment. It is understandable that entities of various entity types can be regarded as nodes in a heterogeneous graph, and the degree of association indicates the edge weight between two nodes. Two nodes with a degree of association of 0 can be regarded as having no edge connection.

[0083] For entities of various entity types, the risk label values ​​corresponding to each entity can be pre-labeled to indicate the current risk level of each entity. The labeling method is not limited in this embodiment, and can be manual labeling or automatic labeling. The value range of the risk label value is also not limited in this embodiment. For example, 0 can be used to indicate no risk and unknown risk, and 1 can be used to indicate risk.

[0084] Step S20, performing at least one round of iterative update calculation with the risk tag value as the initial risk value, obtaining target risk values ​​corresponding to the entities of the various entity types, and obtaining the risk value corresponding to the customer according to the target risk value corresponding to the entity of the customer type;

[0085] After obtaining the degree of association between entities of various entity types and the risk label values ​​corresponding to each entity, the risk label value of each entity can be used as the initial risk value, and at least one round of iterative update calculation can be performed on the risk value of each entity. Through the iterative update calculation process, the risk value of each entity can be propagated among entities along the network in the heterogeneous graph, thereby inferring the final risk level of each entity.

[0086] The risk values ​​corresponding to the entities of various entity types obtained after the last round of update calculations are used as target risk values. The target risk values ​​can be used for business guidance in business scenarios, such as whether to approve a customer's loan application. When it is necessary to obtain the risk value of a customer, the target risk value corresponding to the entity of the customer type can be selected from the entities of various entity types. Similarly, when it is necessary to obtain the risk value of other types of entities, the target risk value corresponding to the entity of that type can also be selected accordingly.

[0087] The steps of performing a round of iterative update calculation in step S20 include:

[0088] Step S201, for the entities to be updated among the entities of the various entity types, at least the degree of association between the entities of the various entity types and the entities to be updated is used as a weight, and the risk values ​​corresponding to the entities of the various entity types are weighted averaged to obtain a risk average result, and the risk value of the entity to be updated after the current iterative update is obtained based on the risk average result.

[0089] The process of each round of iterative update calculation is similar, and one round of iterative update calculation is used as an example for explanation. When performing one round of iterative update calculation, the risk value of the to-be-updated entity among the entities of various entity types is updated. Among them, the to-be-updated entity is the entity whose risk value needs to be updated in this round of iterative update calculation. Each entity among the entities of various entity types can be used as the to-be-updated entity, or only some entities can be used as the to-be-updated entity. The specific setting can be made as needed, for example, only the customer type entity is used as the to-be-updated entity.

[0090] For the entity to be updated, the association degree between the entities of various entity types and the entity to be updated can be used as the weight, and the risk values ​​corresponding to the current entities of various entity types can be weighted averaged to obtain the risk average result, and the risk value of the entity to be updated after the current round of iteration update can be obtained according to the risk average result. Among them, "at least" contains two meanings; one meaning is that at least the association degree between the entities of various entity types and the entity to be updated is used as the weight, that is, in some embodiments, the association degree and other data can be combined as the weight, or the association degree can be converted as the weight; the other meaning is that at least the risk values ​​corresponding to the current entities of various entity types are weighted averaged, that is, in some embodiments, the risk values ​​of the current entities and other data can be weighted averaged together. However, it should be noted that no matter which meaning, it contains the meaning that the greater the association degree, the greater the weight of the corresponding risk value, that is, the more the entity is related to the entity to be updated, the more its risk value is transferred to the entity to be updated.

[0091] The risk values ​​currently corresponding to entities of various entity types refer to the risk values ​​corresponding to entities of various entity types before the start of this round of iterative update calculation. It can be understood that if it is the first round of iterative update calculation, the risk values ​​currently corresponding to entities of various entity types are the risk label values ​​corresponding to entities of various entity types. If it is the iterative update calculation after the first round, the risk values ​​currently corresponding to entities of various entity types are the risk values ​​corresponding to entities of various entity types after the previous round of iterative update calculation.

[0092] In one embodiment, only the degree of association between entities of various entity types and the entity to be updated can be used as the weight, and the risk values ​​corresponding to the current entities of various entity types can be weighted averaged to obtain the risk average result. That is, for each entity of various entity types, the degree of association between the entity and the entity to be updated is multiplied by the current risk value of the entity to obtain the weighted item of the entity, and then the weighted items of each entity are added and divided by the sum of the degree of association between each entity and the entity to be updated, and the result obtained is the risk average result.

[0093] The risk value of the entity to be updated after the iterative update calculation in this round is obtained according to the risk average result. Specifically, the risk average result can be directly used as the risk value of the entity to be updated after the iterative update calculation in this round, or the risk average result can be processed and used as the risk value of the entity to be updated after the iterative update calculation in this round. For example, the risk average result and the current risk value of the entity to be updated are weighted averaged as the risk value of the entity to be updated after the iterative update calculation in this round, and the weight can be set as needed.

[0094] The number of rounds of iterative update calculation is not limited in this embodiment. For example, a fixed number of rounds can be set, or the calculation can be stopped when a certain condition is met.

[0095] In this embodiment, by obtaining the degree of association between entities of various entity types, and obtaining the risk label values ​​corresponding to the entities of various entity types respectively; performing at least one round of iterative update calculation with the risk label value as the initial risk value, the target risk values ​​corresponding to the entities of various entity types are obtained, and the risk value corresponding to the customer is obtained according to the target risk value corresponding to the entity of the customer type. Through at least one round of iterative update calculation, and by taking the degree of association between each entity and the entity to be updated as the weight in each round of iterative update calculation, the risk value of each entity is weighted averaged to obtain the risk value of the entity to be updated, so as to gradually transfer the risk value of each entity associated with the entity to be updated to the entity to be updated, thereby realizing the risk assessment of each entity to be updated, and then obtaining the risk assessment result of the customer. Compared with the traditional scoring card model, the above risk assessment method in this embodiment does not need to extract the risk factor of the customer, but only needs to obtain the degree of association between the entities, and the entity type is not limited except for the customer type, and the number of entities under each entity type is not limited, so it will not be affected by insufficient collection of specific data.

[0096] Furthermore, in one embodiment, the step of obtaining the risk label values ​​corresponding to the entities of the various entity types in step S10 includes:

[0097] Step S101, obtaining risk information corresponding to entities of various entity types respectively;

[0098] Step S102, setting risk label values ​​corresponding to entities of the various entity types respectively according to the risk information; wherein, when the risk information corresponding to an entity to be set among the entities of the various entity types is empty, setting the risk label value corresponding to the entity to be set to 0.

[0099] The risk label value can be automatically set according to the risk information of each entity. Among them, the risk information can be information used to indicate whether the entity is risky, or information indicating the risk level of the entity. There are many ways to set the risk label value according to the risk information, which is not limited in the present embodiment. For example, when the risk information is information indicating whether the entity is risky, if the risk information of an entity indicates that the entity is risky, the risk label value of the entity can be set to 1, and if the risk information of an entity indicates that the entity is risk-free, the risk label value of the entity can be set to 0. For another example, when the risk information is information indicating the risk level of the entity, the value indicating the risk level in the risk information of the entity can be directly used as the risk label value of the entity.

[0100] The entity for which the risk tag value needs to be set is regarded as the entity to be set. When the risk information corresponding to an entity to be set is empty, that is, when the risk information of the entity to be set cannot be obtained, the risk tag value of the entity to be set can be set to 0. That is, the risk assessment method of this embodiment can be applied to the application scenario where the risk information of some entities is clear and the risk level of another entity with unknown risk needs to be assessed.

[0101] Furthermore, in one embodiment, after performing a round of iterative update calculation in step S20, the following steps are further included:

[0102] Step S202, detecting whether a preset stop condition of iterative update calculation is met;

[0103] The stop condition of the iterative update calculation can be set in advance, that is, the preset stop condition. The preset stop condition can be that the changes in the risk values ​​of the last two rounds are within a certain threshold, or that the risk value's effect of distinguishing risks no longer increases, or that the number of iterations reaches a certain upper limit.

[0104] After completing a round of iterative update calculations, it can be detected whether the preset stop condition is met.

[0105] Step S203: if the conditions are met, the iterative update calculation is terminated, and the risk values ​​of the entities of the various entity types after the iterative update in this round are used as the target risk values;

[0106] Step S204: If not satisfied, the next round of iterative update calculation is performed based on the risk values ​​of the entities of the various entity types after the iterative update in the current round.

[0107] Further, based on the above first embodiment, a second embodiment of the risk assessment method of the present invention is proposed. In this embodiment, the step of taking at least the degree of association between the entities of the various entity types and the entity to be updated as a weight, and performing weighted averaging on the risk values ​​corresponding to the entities of the various entity types to obtain the risk average result in step S201 includes:

[0108] Step S2011, for the entity type to be calculated among the various entity types, taking the association degree between the entity of the entity type to be calculated and the entity to be updated as the weight, weighted average is performed on the risk values ​​corresponding to the entities of the current entity type to be calculated, so as to obtain the calculation result corresponding to the entity type to be calculated;

[0109] In this embodiment, the risk values ​​corresponding to the entities of each entity type may be weighted averaged based on the association degree between the entities of each entity type and the entity to be updated and the type weights corresponding to the entities of each entity type.

[0110] Specifically, for the entity type to be calculated, the association degree between the entity of the entity type to be calculated and the entity to be updated is used as the weight, and the risk values ​​corresponding to the entities of the current entity type to be calculated are weighted averaged to obtain the calculation result corresponding to the entity type to be calculated. Each entity type can be used as the entity type to be calculated for the above calculation.

[0111] Step S2012, obtaining preset type weights corresponding to the entity type to which the entity to be updated belongs for each of the entity types;

[0112] Step S2013: performing weighted averaging on at least the calculation results corresponding to the various entity types according to the type weights to obtain a risk average result.

[0113] The weights (hereinafter referred to as type weights) corresponding to the entity types to which the entity to be updated belongs can be set or calculated in advance as needed. After the calculation results corresponding to the various entity types are calculated, the weighted average of each calculation result can be at least obtained according to the type weight to obtain the risk average result, wherein at least means that in some embodiments, only the weighted average of each calculation result can be obtained to obtain the risk average result, and in other embodiments, each calculation result and other data can also be weighted averaged together to obtain the risk average result. Here, other data are not restricted. The size of the type weight determines the weight of the risk value of the entity of the corresponding entity type when it is passed to the entity to be updated. When setting the type weight, the degree of risk impact of one entity type on another entity type can be considered. The greater the risk impact, the larger the type weight of the entity type for the other entity type can be set, and vice versa.

[0114] Further, in one embodiment, the step S2013 includes:

[0115] Step a, obtaining a preset label weight corresponding to the risk label value of the entity to be updated;

[0116] In this embodiment, the calculation results corresponding to various entity types and the risk label values ​​of the entities to be updated can be weighted averaged together, that is, in each round of iterative update, the component of the initial risk value of the entity to be updated must be considered. The weight corresponding to the risk label value of the entity to be updated (hereinafter referred to as the label weight) can be set in advance as needed. The label weights of various entities of the same entity type can be set to be the same.

[0117] Step b: according to the type weight and the label weight, weighted average is performed on the calculation results corresponding to the various entity types and the risk label values ​​of the entities to be updated to obtain a risk average result.

[0118] After calculating the calculation results corresponding to various entity types, the risk average result can be obtained by weighted averaging the various calculation results and the risk label value of the entity to be updated according to the type weight and label weight. That is, the risk label value of the entity to be updated is multiplied by the label weight to obtain the weighted item corresponding to the risk label value, and each calculation result is multiplied by the corresponding type weight to obtain the weighted item corresponding to each calculation result. The sum of each weighted item is divided by the sum of the type weight and the label weight, and the result is the risk average result.

[0119] Furthermore, in one embodiment, before step S20, the method further includes:

[0120] Step S30, obtaining a risk impact value of a first entity type on a second entity type among the various entity types;

[0121] Step S40: setting a type weight of the first entity type with respect to the second entity type according to the risk impact value, wherein a larger risk impact value corresponds to a larger type weight.

[0122] In this embodiment, for one entity type among the entity types (hereinafter referred to as the first entity type for distinction), a risk impact value of the first entity type on another entity type among the entity types (hereinafter referred to as the second entity type for distinction) can be preset, and a type weight of the first entity type for the second entity type can be set according to the risk impact value. The first entity type and the second entity type can be the same entity type, that is, a type weight of an entity type can be set for itself. The risk impact value of the first entity type on the second entity type can be manually set as needed. For example, when it is believed that the risk value of the bank has a great impact on the risk value of the customer, the risk impact value of the bank entity type on the customer type can be set to be larger. Setting the type weight according to the risk impact value can be such that the larger the risk impact value, the larger the corresponding type weight, that is, the larger the risk impact value of the first entity type on the second entity type, the larger the type weight of the first entity type for the second entity type.

[0123] Further, based on the above-mentioned first and / or second embodiments, a third embodiment of the risk assessment method of the present invention is proposed. In this embodiment, after the step of obtaining the degree of association between entities of various entity types, step S10 further includes:

[0124] Step S50, normalizing the association matrix formed by the association degree between entities of any two entity types among the various entity types, to obtain the normalized association degree between entities of any two entity types;

[0125] In this embodiment, when the value range of the degree of association between entities is not limited to 0-1, in order to avoid numerical problems caused by the large differences in the degrees of association being not in the same order of magnitude, the degrees of association can be normalized first, and the normalized degrees of association can be used as weights to calculate the risk value.

[0126] Specifically, the association matrix formed by the association degree between entities of each two entity types in various entity types may be normalized to obtain the normalized association degree between the two entities.

[0127] That is, for an entity type i and an entity type j, the association matrix between the entities of the two entity types is formed by R ij Indicates that the element in the nth row and mth column of the matrix represents the degree of association between the nth entity of type i and the mth entity of type j. ij After normalization, we get the normalized matrix S ij , the element in the nth row and mth column of the matrix represents the normalized degree of association between the nth entity of type i and the mth entity of type j.

[0128] The method for normalizing the correlation matrix is ​​not limited in this embodiment. For example, a Laplace matrix normalization method may be used.

[0129] For example, in one embodiment, the correlation matrix R can be calculated as follows: ij Normalize:

[0130] R ij Denote by W, calculate the two degree matrices Dnn and Dmm of W;

[0131] in, N is the number of entities of entity type i, M is the number of entities of entity type j;

[0132] S ij =D nn -1 / 2 WD mm -1 / 2 .

[0133] The step of taking at least the degree of association between the entities of the various entity types and the entity to be updated as weights and performing weighted averaging on the risk values ​​corresponding to the entities of the various entity types to obtain the risk average result in step S201 includes:

[0134] Step S2014: taking at least the normalized association degree between the entities of the various entity types and the entity to be updated as a weight, weighted averaging the risk values ​​corresponding to the entities of the various entity types is performed to obtain a risk average result.

[0135] When performing iterative update calculations on the risk value of the entity to be updated, the normalized degree of association between the entities of various entity types and the entity to be updated can be used as a weight to perform a weighted average on the risk values ​​corresponding to the entities of various entity types to obtain a risk average result. The specific implementation details can be referred to step S201, which will not be described in detail here.

[0136] The third embodiment can be implemented in combination with the second embodiment. Specifically, in step S201, the step of taking at least the degree of association between the entities of the various entity types and the entity to be updated as a weight, and performing weighted averaging on the risk values ​​corresponding to the entities of the various entity types to obtain the risk average result comprises:

[0137] For the entity type to be calculated among various entity types, taking the normalized association degree between the entity of the entity type to be calculated and the entity to be updated as the weight, weighted average is performed on the risk values ​​corresponding to the entities of the current entity type to be calculated, so as to obtain the calculation result corresponding to the entity type to be calculated;

[0138] Obtaining preset type weights corresponding to the entity types to which the entity to be updated belongs for each of the entity types;

[0139] According to the type weight, at least the calculation results corresponding to the various entity types are weighted averaged to obtain a risk average result.

[0140] Further, in one embodiment, the step of performing weighted averaging of at least the calculation results corresponding to the various entity types according to the type weights to obtain a risk average result may include:

[0141] Obtaining a preset label weight corresponding to the risk label value of the entity to be updated;

[0142] According to the type weight and the label weight, the calculation results corresponding to the various entity types and the risk label values ​​of the to-be-updated entities are weighted averaged to obtain a risk average result.

[0143] The above implementation is expressed by the following calculation formula:

[0144]

[0145] Among them, t represents the round of iterative update, F i(t+1) represents the vector of risk values ​​of each entity of the ith entity type in round t+1, L is the number of entity types, λ ij is the type weight of the i-th entity type for the j-th entity type, α i is the label weight corresponding to the entity of the i-th entity type, y i is a vector consisting of the risk label values ​​of each entity of the i-th entity type.

[0146] For all risk identification scenarios with heterogeneous graph structures and missing identifiers, you can try to use this embodiment to model the risks instead of the traditional scorecard model. In actual scenarios, many types of entities are usually involved. This embodiment can make full use of this information to identify risks that traditional methods cannot identify. And you can define association matrices, risk label values, and different weights according to specific problems to achieve different business goals. In addition, the use of iterative algorithms makes it feasible to calculate on ultra-large-scale data.

[0147] In addition, the embodiment of the present invention also provides a risk assessment device, referring to Figure 3 , the device comprises:

[0148] An acquisition module 10 is used to acquire the association degree between entities of various entity types, and to acquire the risk label values ​​corresponding to the entities of various entity types, wherein one entity type is a customer type;

[0149] Iterative update module 20, used to perform at least one round of iterative update calculation with the risk tag value as the initial risk value, to obtain the target risk values ​​corresponding to the entities of the various entity types, and to obtain the risk value corresponding to the customer according to the target risk value corresponding to the entity of the customer type;

[0150] Among them, performing a round of iterative update calculation specifically includes:

[0151] For the entities to be updated among the entities of the various entity types, at least the degree of association between the entities of the various entity types and the entities to be updated is used as a weight, and the risk values ​​corresponding to the entities of the various entity types are weighted averaged to obtain a risk average result, and the risk value of the entity to be updated after this round of iterative update is obtained based on the risk average result.

[0152] Furthermore, the iterative update module 20 is also used for:

[0153] For the entity type to be calculated among the various entity types, taking the association degree between the entity of the entity type to be calculated and the entity to be updated as the weight, weighted average is performed on the risk values ​​corresponding to the entities of the current entity type to be calculated, so as to obtain the calculation result corresponding to the entity type to be calculated;

[0154] Obtaining preset type weights corresponding to the entity types to which the entity to be updated belongs for each of the entity types;

[0155] According to the type weight, at least the calculation results corresponding to the various entity types are weighted averaged to obtain a risk average result.

[0156] Furthermore, the iterative update module 20 is also used for:

[0157] Obtaining a preset label weight corresponding to the risk label value of the entity to be updated;

[0158] According to the type weight and the label weight, the calculation results corresponding to the various entity types and the risk label values ​​of the to-be-updated entities are weighted averaged to obtain a risk average result.

[0159] Furthermore, the acquisition module 10 is also used for:

[0160] Obtaining a risk impact value of a first entity type on a second entity type among the various entity types;

[0161] The device also includes:

[0162] A setting module is configured to set a type weight of the first entity type with respect to the second entity type according to the risk impact value, wherein a larger risk impact value corresponds to a larger type weight.

[0163] Furthermore, the device also includes:

[0164] A normalization module, used to normalize the association matrix formed by the association degree between entities of any two entity types among the various entity types, to obtain the normalized association degree between entities of any two entity types;

[0165] The iterative update module 20 is also used for:

[0166] At least taking the normalized association degree between the entities of the various entity types and the entity to be updated as a weight, weighted averaging is performed on the risk values ​​corresponding to the entities of the various entity types to obtain a risk average result.

[0167] Furthermore, the acquisition module 10 is also used for:

[0168] Obtain risk information corresponding to entities of various entity types respectively;

[0169] Setting risk label values ​​corresponding to entities of various entity types respectively according to the risk information;

[0170] When the risk information corresponding to an entity to be set among the entities of the various entity types is empty, the risk label value corresponding to the entity to be set is set to 0.

[0171] Furthermore, the iterative update module 20 is also used for:

[0172] Detect whether the preset stop condition of the iterative update calculation is met;

[0173] If satisfied, the iterative update calculation is terminated, and the risk values ​​of the entities of the various entity types after the iterative update in this round are used as the target risk values;

[0174] If not satisfied, the next round of iterative update calculation is performed based on the risk values ​​of the entities of the various entity types after the current round of iterative update.

[0175] The expanded contents of the specific implementation manner of the risk assessment device of the present invention are basically the same as those of the above-mentioned risk assessment method embodiments, and will not be elaborated here.

[0176] In addition, an embodiment of the present invention further proposes a computer-readable storage medium, on which a risk assessment program is stored. When the risk assessment program is executed by a processor, the steps of the risk assessment method described below are implemented.

[0177] The various embodiments of the risk assessment device and computer-readable storage medium of the present invention may refer to the various embodiments of the risk assessment method of the present invention, and will not be described in detail here.

[0178] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including 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 "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0179] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0180] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0181] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A risk assessment method, characterized in that: The method comprises the following steps: Obtaining the degree of association between entities of various entity types, and obtaining the risk label values ​​corresponding to the entities of the various entity types, wherein one entity type is a customer type; Perform at least one round of iterative update calculation using the risk tag value as the initial risk value to obtain target risk values ​​corresponding to entities of various entity types, and obtain the risk value corresponding to the customer based on the target risk value corresponding to the customer type entity; The steps of performing a round of iterative update calculation include: For the entities to be updated among the entities of the various entity types, at least the degree of association between the entities of the various entity types and the entities to be updated is used as a weight, and the risk values ​​corresponding to the entities of the various entity types are respectively weighted averaged to obtain a risk average result, and the risk value of the entity to be updated after the current round of iterative update is obtained according to the risk average result; The step of obtaining the risk label values ​​corresponding to the entities of the various entity types includes: Obtain risk information corresponding to entities of various entity types respectively; Setting risk label values ​​corresponding to entities of various entity types respectively according to the risk information; Wherein, when the risk information corresponding to an entity to be set among the entities of the various entity types is empty, the risk tag value corresponding to the entity to be set is set to 0; The step of taking at least the association degree between the entities of the various entity types and the entity to be updated as weight and performing weighted averaging on the risk values ​​corresponding to the entities of the various entity types to obtain the risk average result comprises: For the entity type to be calculated among the various entity types, taking the association degree between the entity of the entity type to be calculated and the entity to be updated as the weight, weighted average is performed on the risk values ​​corresponding to the entities of the current entity type to be calculated, so as to obtain the calculation result corresponding to the entity type to be calculated; Obtaining preset type weights corresponding to the entity types to which the entity to be updated belongs for each of the entity types; According to the type weight, at least the calculation results corresponding to the various entity types are weighted averaged to obtain a risk average result.

2. The risk assessment method according to claim 1, characterized in that: The step of performing weighted averaging of at least the calculation results corresponding to the various entity types according to the type weights to obtain a risk average result comprises: Obtaining a preset label weight corresponding to the risk label value of the entity to be updated; According to the type weight and the label weight, the calculation results corresponding to the various entity types and the risk label values ​​of the to-be-updated entities are weighted averaged to obtain a risk average result.

3. The risk assessment method according to claim 1, characterized in that: Before the step of performing at least one round of iterative update calculation using the risk label value as the initial risk value to obtain the target risk values ​​corresponding to the entities of the various entity types, the step further includes: Obtaining a risk impact value of a first entity type on a second entity type among the various entity types; A type weight of the first entity type relative to the second entity type is set according to the risk impact value, wherein a larger risk impact value corresponds to a larger type weight.

4. The risk assessment method according to claim 1, characterized in that: After the step of obtaining the degree of association between entities of various entity types, the method further includes: Normalizing the association matrix formed by the association degree between entities of any two entity types among the various entity types to obtain the normalized association degree between entities of any two entity types; The step of taking at least the association degree between the entities of the various entity types and the entity to be updated as weight and performing weighted averaging on the risk values ​​corresponding to the entities of the various entity types to obtain the risk average result comprises: At least taking the normalized association degree between the entities of the various entity types and the entity to be updated as a weight, weighted averaging is performed on the risk values ​​corresponding to the entities of the various entity types to obtain a risk average result.

5. The risk assessment method according to any one of claims 1 to 4, characterized in that: After a round of iterative update calculation steps, it also includes: Detect whether the preset stop condition of the iterative update calculation is met; If satisfied, the iterative update calculation is terminated, and the risk values ​​of the entities of the various entity types after the iterative update in this round are used as the target risk values; If not satisfied, the next round of iterative update calculation is performed based on the risk values ​​of the entities of the various entity types after the current round of iterative update.

6. A risk assessment device, characterized in that: The device comprises: An acquisition module is used to acquire the degree of association between entities of various entity types, and to acquire the risk label values ​​corresponding to the entities of the various entity types, wherein one entity type is a customer type; the acquisition module is also used to acquire the risk information corresponding to the entities of the various entity types; and to set the risk label values ​​corresponding to the entities of the various entity types according to the risk information; wherein, when the risk information corresponding to an entity to be set among the entities of the various entity types is empty, the risk label value corresponding to the entity to be set is set to 0; An iterative update module is used to perform at least one round of iterative update calculation with the risk tag value as the initial risk value, to obtain the target risk values ​​corresponding to the entities of the various entity types, and to obtain the risk value corresponding to the customer according to the target risk value corresponding to the entity of the customer type; the iterative update module is also used to, for the entity types to be calculated among the various entity types, take the degree of association between the entities of the entity types to be calculated and the entities to be updated as the weight, to perform weighted averaging on the risk values ​​corresponding to the entities of the current entity types to be calculated, to obtain the calculation results corresponding to the entity types to be calculated; obtain the preset type weights corresponding to the entity types of the various entity types for the entity types to which the entity to be updated belongs; and perform weighted averaging on at least the calculation results corresponding to the various entity types according to the type weights to obtain the risk average result; Among them, performing a round of iterative update calculation specifically includes: For the entities to be updated among the entities of the various entity types, at least the degree of association between the entities of the various entity types and the entities to be updated is used as a weight, and the risk values ​​corresponding to the entities of the various entity types are weighted averaged to obtain a risk average result, and the risk value of the entity to be updated after this round of iterative update is obtained based on the risk average result.

7. A risk assessment device, characterized in that: The risk assessment device comprises: a memory, a processor, and a risk assessment program stored in the memory and executable on the processor, wherein the risk assessment program implements the steps of the risk assessment method according to any one of claims 1 to 5 when executed by the processor.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a risk assessment program, and when the risk assessment program is executed by a processor, the steps of the risk assessment method according to any one of claims 1 to 5 are implemented.

Images