Risk control methods, devices, equipment, and storage media that support hierarchical feature backtracking

Abstract

This paper provides a risk control method, apparatus, device, and storage medium supporting hierarchical feature backtracking. The method includes: calculating standard features based on the user's original data; calculating derived features based on the standard features, generating a feature hierarchy relationship, whereby the feature hierarchy relationship represents the standard features and derived features, as well as the operational logic between derived features, with the top-level derived feature serving as the target feature; establishing a hierarchy relationship table based on the feature hierarchy relationship, whereby the hierarchy relationship table records the parent feature of each derived feature, which is the feature directly used when the derived feature is generated; and recursively calculating the feature result of the target feature based on the hierarchy relationship table. This paper supports the simultaneous backtracking of multiple features with dependencies, which can improve the efficiency and accuracy of feature result calculation.

Description

Technical Field

[0001] This invention relates to the financial field, and in particular to a risk control method, apparatus, device, and storage medium that supports hierarchical feature backtracking. Background Technology

[0002] In the financial sector, as markets and customer groups constantly evolve, user risks also change in real time. Existing technologies aim to support online risk control strategies by mining more effective features. Before features are deployed, feature results often need to be calculated. However, features are not independent; there are dependencies between different features. Current feature result calculation methods only support backtracking for single features, not for multiple dependent features simultaneously, resulting in low efficiency and accuracy in feature result calculation.

[0003] Therefore, there is an urgent need for a risk control method that supports hierarchical feature backtracking, which can support the simultaneous backtracking of multiple dependent features, thereby improving the efficiency and accuracy of feature result calculation. Summary of the Invention

[0004] The purpose of this embodiment is to provide a risk control method, apparatus, device, and storage medium that supports hierarchical feature backtracking, so as to improve the efficiency and accuracy of feature result calculation.

[0005] To achieve the above objectives, this paper provides a risk control method that supports hierarchical feature backtracking, including:

[0006] Standard features are calculated based on the user's original data;

[0007] Based on the standard features, derived features are calculated, and a feature hierarchy relationship is generated. The feature hierarchy relationship is used to characterize the standard features and derived features, as well as the operational logic between derived features. The derived feature at the top level is taken as the target feature.

[0008] Based on the feature hierarchy, a hierarchy table is established. The hierarchy table records the parent feature of each derived feature. The parent feature is the feature that the derived feature is directly based on when it is generated.

[0009] Based on the hierarchical relationship table, the feature results of the target feature are obtained through hierarchical recursive calculation.

[0010] Preferably, the step of obtaining the feature result of the target feature through hierarchical recursive calculation based on the hierarchical relationship table further includes:

[0011] Use the target feature as the current feature;

[0012] Based on the hierarchical relationship table, determine whether the current feature has a parent feature;

[0013] If so, determine the feature result of the parent feature of the current feature; calculate the feature result of the current feature based on the feature result of the parent feature of the current feature;

[0014] If not, then the feature result of the current feature is calculated based on the original data.

[0015] Preferably, the feature result of determining the parent feature of the current feature further includes:

[0016] The parent feature of the current feature is used as the current feature, and the judgment step is executed.

[0017] Preferably, the step of calculating the feature result of the current feature based on the feature result of the parent feature of the current feature further includes:

[0018] The feature result of the parent feature of the current feature is input into the trained tree model to obtain the feature result of the current feature.

[0019] Preferred options also include:

[0020] The user's historical risk control results are used to make decisions on the feature results of the target feature, and the effectiveness of the target feature obtained based on the feature hierarchy relationship is verified.

[0021] Preferably, the step of using the user's historical risk control results to make decisions on the feature results of the target feature, and verifying the effectiveness of the target feature obtained based on the feature hierarchy relationship, further includes:

[0022] Calculate the proportion of users whose historical risk control results are consistent with the feature results of the target feature;

[0023] If the user ratio is greater than or equal to the set ratio, then the target feature obtained based on the feature hierarchy relationship is valid;

[0024] If the user ratio is less than the set ratio, the target feature obtained based on the feature hierarchy relationship is invalid.

[0025] Preferably, the proportion of users with different historical risk control results is the same.

[0026] Preferred options also include:

[0027] When the target feature obtained based on the feature hierarchy is invalid, the root cause of the invalidity of the target feature is determined based on the feature results of the standard feature and the derived feature, according to the hierarchy table.

[0028] On the other hand, this embodiment provides a risk control device that supports hierarchical feature backtracking, the device comprising:

[0029] The standard feature determination module uses the user's raw data to calculate standard features.

[0030] The derived feature determination module is used to calculate derived features based on the standard features, generate a feature hierarchy relationship, and the feature hierarchy relationship is used to characterize the standard features and derived features, as well as the operation logic between derived features, and the derived feature at the top level is used as the target feature.

[0031] The relationship table creation module is used to create a hierarchical relationship table based on the feature hierarchy relationship. The hierarchical relationship table records the parent feature of each derived feature. The parent feature is the feature that the derived feature is directly based on when it is generated.

[0032] The feature result calculation module is used to calculate the feature result of the target feature by recursively calculating it in layers based on the hierarchical relationship table.

[0033] In another aspect, embodiments of this document also provide a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the computer program, when executed by the processor, performs instructions of any of the methods described above.

[0034] In another aspect, the embodiments herein also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor of a computer device, performs instructions for any of the methods described above.

[0035] As can be seen from the technical solutions provided in the embodiments above, the embodiments of this paper, through the method of these embodiments, can obtain standard features based on the user's original data, further calculate derived features based on the standard features, thereby generating a feature hierarchy relationship, and establishing a hierarchy relationship table based on the feature hierarchy relationship. The feature results of the target features are obtained through hierarchical recursion. These embodiments represent the dependencies between features through feature hierarchy relationships, and utilize hierarchical recursion to backtrack multiple features simultaneously, thereby obtaining feature results with higher computational efficiency and accuracy.

[0036] To make the above and other objects, features and advantages of this document more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments or prior art described herein, the accompanying drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this article. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This document illustrates a flowchart of a risk control method supporting hierarchical feature backtracking, as provided in an embodiment of this paper.

[0039] Figure 2 This document illustrates a flowchart illustrating the process of obtaining the feature results of the target feature through hierarchical recursive calculation provided in the embodiments of this paper.

[0040] Figure 3 This document illustrates a flowchart of a process for verifying the effectiveness of target features obtained based on feature hierarchy relationships, as provided in the embodiments of this paper.

[0041] Figure 4 This document illustrates a schematic diagram of the module structure of a risk control device supporting hierarchical feature backtracking, as provided in an embodiment of this paper.

[0042] Figure 5 A schematic diagram of the structure of the computer device provided in the embodiments of this article is shown.

[0043] Explanation of symbols in the attached drawings:

[0044] 100. Standard Feature Determination Module;

[0045] 200. Derivative Feature Determination Module;

[0046] 300. Relationship table creation module;

[0047] 400. Feature Result Calculation Module;

[0048] 502. Computer equipment;

[0049] 504, Processor;

[0050] 506. Memory;

[0051] 508. Drive mechanism;

[0052] 510. Input / output module;

[0053] 512. Input devices;

[0054] 514. Output devices;

[0055] 516. Presentation equipment;

[0056] 518. Graphical User Interface;

[0057] 520. Network interface;

[0058] 522. Communication link;

[0059] 524. Communication bus. Detailed Implementation

[0060] The technical solutions in the embodiments described below will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments described herein, and not all of the embodiments. Based on the embodiments described herein, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this document.

[0061] In the financial sector, as markets and customer groups constantly evolve, user risks also change in real time. Existing technologies aim to support online risk control strategies by mining more effective features. Before features are deployed, feature results often need to be calculated. However, features are not independent; there are dependencies between different features. Current feature result calculation methods only support backtracking for single features, not for multiple dependent features simultaneously, resulting in low efficiency and accuracy in feature result calculation.

[0062] To address the aforementioned issues, this paper provides a risk control method that supports hierarchical feature backtracking. Figure 1 This is a flowchart illustrating a risk control method supporting hierarchical feature backtracking, as provided in the embodiments of this document. This specification provides the operational steps of the method described in the embodiments or flowcharts, but based on conventional or non-inventive labor, more or fewer operational steps may be included. The order of steps listed in the embodiments is merely one possible execution order among many and does not represent the only possible execution order. In actual system or device products, the methods shown in the embodiments or accompanying drawings can be executed sequentially or in parallel.

[0063] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings herein are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0064] Reference Figure 1 This paper presents a risk control method that supports hierarchical feature backtracking, including:

[0065] S101: Calculate the standard features based on the user's original data;

[0066] S102: Based on the standard features, calculate the derived features and generate a feature hierarchy relationship. The feature hierarchy relationship is used to characterize the standard features and derived features, as well as the operational logic between the derived features. The derived feature at the top level is taken as the target feature.

[0067] S103: Based on the feature hierarchy relationship, establish a hierarchy relationship table. The hierarchy relationship table records the parent feature of each derived feature. The parent feature is the feature that the derived feature is directly based on when it is generated.

[0068] S104: Based on the hierarchical relationship table, the feature results of the target feature are obtained by hierarchical recursive calculation.

[0069] In this embodiment, the user is the user who needs to be subject to risk control. Risk control refers to risk management. In a broad sense, it can include various types of risk management. In the financial field, risk control can refer to whether or not to grant a loan to a user, the specific amount of the loan granted, etc. This article takes whether or not to grant a loan to a user as an example to illustrate the method and logic of this article.

[0070] User raw data can include basic user information, loan information, and browsing history of web pages or software within a recently defined time period. Basic user information may include ID card details, educational background, work experience, etc. Loan information may include loan amounts, repayment status, and outstanding balances. Browsing history of web pages or software within a defined time period can be the frequency of browsing web pages or software in a list, which may include lending apps or websites. It should be noted that raw data can be categorized into proprietary data and third-party data, depending on its source. Proprietary data can be data held by the risk control center or the bank itself, while third-party data can be data obtained through third-party software such as Tongdun, Bairong, or Pudao.

[0071] Standard features can be obtained from the raw data. Standard features refer to the basic features for feature backtracking. These standard features have specific physical meanings. For example, the basic features needed to determine whether to grant a loan include the user's ID information, work experience, the number of loans the user has taken out, and the amount of debt the user owes. These standard features need to be extracted from a large amount of raw user data.

[0072] Based on standard features, derived features can be calculated, generating a feature hierarchy. Derived features can be derived from standard features or derived features themselves; they do not have a specific physical meaning. For example, using a user's ID information and work experience, we obtain user derived feature A; using the user's loan amount and debt amount, we obtain user derived feature B; and using derived features A and B, we obtain derived feature C. This forms a feature hierarchy. In this hierarchy, if the standard feature is at the bottom level (0), and derived features A and B are at level 1, then derived feature C is at level 2. The feature hierarchy represents the standard features and derived features, as well as the operational logic between derived features.

[0073] The method described in this embodiment allows for the generation of standard features from the user's original data, followed by the calculation of derived features based on these standard features. This generates a feature hierarchy, and a hierarchy table is established based on this hierarchy. The target feature is then obtained through hierarchical recursion. This embodiment represents the dependencies between features using a feature hierarchy, and utilizes hierarchical recursion to backtrack on multiple features simultaneously, resulting in feature results with higher computational efficiency and accuracy.

[0074] Based on the feature hierarchy, a hierarchy table can be established. This table records the parent features of each derived feature. For example, the parent features of derived feature C are derived features A and B. The parent features of derived feature A are the standard features: the user's ID information and work experience. The parent features of derived feature B are the standard features: the user's loan amount and debt amount. For standard features, the next level up can be the corresponding original data. In addition to recording the parent features of each derived feature, the hierarchy table can also record the original data corresponding to the standard features.

[0075] Reference Figure 2 Based on the hierarchical relationship table, the feature results of the target feature can be obtained by recursive calculation in a hierarchical manner, specifically as follows:

[0076] S201: Use the target feature as the current feature;

[0077] S202: Based on the hierarchical relationship table, determine whether the current feature has a parent feature;

[0078] S203: If so, determine the feature result of the parent feature of the current feature; calculate the feature result of the current feature based on the feature result of the parent feature of the current feature;

[0079] S204: If not, then calculate the feature result of the current feature based on the original data.

[0080] The feature result of determining the parent feature of the current feature further includes:

[0081] The parent feature of the current feature is used as the current feature, and the judgment step is executed.

[0082] Specifically, when the current feature is derived feature C, it is determined that current feature C has a parent feature. When calculating the feature result of current feature C, it is necessary to consider the feature results of parent features A and B. When calculating the feature result of parent feature A, parent feature A is used as the current feature. It is determined that current feature A has a parent feature. When calculating the feature result of current feature A, it is necessary to consider the parent standard features: the user's ID card information and work experience. When calculating the parent feature, the user's ID card information, the user's ID card information is used as the current feature. If the current feature does not have a parent feature, then the feature result of the current feature, i.e., the user's ID card information, is calculated based on the original data. Furthermore, by back-feeding back the feature results of the user's ID card information and work experience, the feature result of derived feature A can be calculated. Similarly, the feature result of derived feature B can also be calculated. By back-feeding back the feature results of derived features A and B, the feature result of derived feature C can be calculated.

[0083] The feature results of standard or derived features at each level can be synchronized to the distributed file system. It should be noted that the above steps involve calculating the feature result of the current feature based on the original data. Since this is a "what if" branch, the current feature has no parent feature; that is, the current feature can only be a standard feature, meaning the feature result of the standard feature is calculated based on the original data. The specific calculation method takes ID card information as an example of a standard feature: After extracting the ID card information from the original data, the ID card information can include name, age, and gender. However, this information has different dimensions. Therefore, one-hot encoding can be used to convert the ID card information into one-hot encoded data. For example, gender is represented by 0 and 1 for female and 10 for male, thus obtaining the feature result of the standard feature represented by one-hot encoded data. One-hot encoded data is generally in 0-1 encoding form.

[0084] The above steps also involve calculating the feature result of the current feature based on the feature results of its parent features. Specifically, this can be done by inputting the feature results of the parent features of the current feature into a trained tree model to obtain the feature result of the current feature. Different current features correspond to different tree models, such as decision tree models or regression tree models. Tree models need to be trained, and the specific training process will not be elaborated here. For example, derived feature A corresponds to tree model a. Inputting the one-hot encoded data corresponding to the standard features (ID card information and work experience) into tree model a will yield the feature result of derived feature A. Similarly, the feature result of derived feature B can also be obtained. For example, derived feature C corresponds to tree model c. Inputting the feature results of derived features A and B into tree model c will yield the feature result of derived feature C. However, it should be noted that since the feature results of derived features A, B, and C are generated based on one-hot encoded data, the final feature results of derived features A, B, and C are all represented in encoded form.

[0085] Since the derived feature at the top level in the feature hierarchy is used as the target feature, and the feature result of the target feature, i.e. the feature result of the derived feature C, is in encoded form, if it is necessary to verify the effectiveness of the feature result of the target feature in the future, it needs to be converted into a score. Specifically, the encoded feature result can be input into a trained machine learning model to obtain the corresponding score. The machine learning model can be a K-model, a regression model, etc. The specific training method will not be elaborated in this article.

[0086] Furthermore, the embodiments in this paper can also use the user's historical risk control results to make decisions on the feature results of the target feature, and verify the effectiveness of the target feature obtained based on the feature hierarchy relationship.

[0087] Generally, a user's historical risk control results are either loan approval or disapproval. After obtaining a score based on the target feature's characteristics, a backtracking result can be derived from the score. For example, the score is typically between 0 and 1. If the score is less than 0.5, the backtracking result is disapproval; if the score is greater than or equal to 0.5, the backtracking result is loan approval. Furthermore, the user's historical risk control results are used to make decisions regarding the target feature's characteristics, thus verifying the effectiveness of the target feature.

[0088] Specifically, refer to Figure 3 The step of using the user's historical risk control results to make decisions on the feature results of the target features, and verifying the effectiveness of the target features obtained based on the feature hierarchy relationship, further includes:

[0089] S301: Calculate the proportion of users whose historical risk control results are consistent with the feature results of the target feature;

[0090] S302: If the user ratio is greater than or equal to the set ratio, then the target feature obtained based on the feature hierarchy relationship is valid;

[0091] S303: If the user ratio is less than the set ratio, the target feature obtained based on the feature hierarchy relationship is invalid.

[0092] In this embodiment, there are multiple users, for example, 100 users. Users whose historical risk control results match the target feature's feature results are defined as those whose historical risk control results match the score obtained from the target feature's feature results, and whose backtracking results are based on that score. For example, if a user's historical risk control result is a loan disbursement, and the backtracking result is also a loan disbursement, then they are considered consistent. The proportion of consistent users among the 100 users is calculated. If this proportion is greater than or equal to a set proportion, the target feature obtained based on the feature hierarchy is valid; otherwise, it is invalid. The set proportion can be determined according to actual working conditions, and this document does not impose any limitations on it.

[0093] Furthermore, the proportion of users with different historical risk control results is the same. For 100 users, the proportion of users with historical risk control results of loan approval and users with historical risk control results of loan rejection is 50% each. That is, there are 50 users with historical risk control results of loan approval and 50 users with historical risk control results of loan rejection. This setting ensures the accuracy and reliability of the validity judgment.

[0094] In this embodiment, the method further includes: when the target feature obtained based on the feature hierarchy is invalid, determining the root cause of the invalidity of the target feature based on the hierarchy table and the feature results of the standard feature and the derived feature.

[0095] When the target feature is invalid, it indicates a problem in the calculation of one or more levels of the feature hierarchy, requiring further verification of the root cause. Based on the preceding discussion, the feature results of the standard or derived features at each level can be synchronized to the distributed file system. Specifically, the parent feature of each derived feature can be determined based on the hierarchy table; the feature result of the parent feature is extracted from the distributed file system, and the feature result of each derived feature is calculated based on the parent feature's feature result. This feature result serves as the verification result for each derived feature. The verification result of each derived feature is compared with the feature result of each derived feature recorded in the distributed file system. If they match, it proves that the calculation of the corresponding derived feature is correct; if they do not match, it proves that the calculation of the corresponding derived feature is incorrect, which is the root cause of the invalid target feature.

[0096] The method described in this article can identify and verify the root causes of invalid target features, provide the specific root causes of the problems to the staff, and make corresponding modifications and adjustments. By specifically identifying the causes of the problems, the problems can be solved in a targeted manner, thereby improving work efficiency.

[0097] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the acquisition, storage, use, and processing of data in the technical solutions described in the embodiments of this application all comply with the relevant provisions of national laws and regulations.

[0098] Based on the risk control method supporting hierarchical feature backtracking described above, this embodiment also provides a risk control device supporting hierarchical feature backtracking. The device may include a system (including a distributed system), software (application), module, component, server, client, etc., using the method described in this embodiment, combined with necessary implementation hardware. Based on the same innovative concept, the devices in one or more embodiments provided in this embodiment are as described in the following embodiments. Since the implementation schemes and methods for solving the problem by the device are similar, the implementation of the specific device in this embodiment can refer to the implementation of the aforementioned method, and repeated details will not be repeated. As used below, the terms "unit" or "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0099] Specifically, Figure 4 This is a schematic diagram of the module structure of one embodiment of a risk control device supporting hierarchical feature backtracking provided in this article. (Refer to...) Figure 4 As shown in the embodiment of this paper, a risk control device supporting hierarchical feature backtracking includes: a standard feature determination module 100, a derived feature determination module 200, a relationship table establishment module 300, and a feature result calculation module 400.

[0100] The standard feature determination module 100 is used to calculate standard features from the user's original data.

[0101] The derived feature determination module 200 is used to calculate derived features based on the standard features, generate a feature hierarchy relationship, the feature hierarchy relationship is used to characterize the standard features and derived features, as well as the operation logic between derived features, and the derived feature located at the top level is used as the target feature.

[0102] The relationship table establishment module 300 is used to establish a hierarchical relationship table based on the feature hierarchy relationship. The hierarchical relationship table records the parent feature of each derived feature. The parent feature is the feature that the derived feature is directly based on when it is generated.

[0103] The feature result calculation module 400 is used to calculate the feature result of the target feature by recursively calculating it in layers based on the hierarchical relationship table.

[0104] Reference Figure 5 As shown, based on the risk control method supporting hierarchical feature backtracking described above, one embodiment of this paper also provides a computer device 502, wherein the above method runs on the computer device 502. The computer device 502 may include one or more processors 504, such as one or more central processing units (CPUs) or graphics processing units (GPUs), each processing unit may implement one or more hardware threads. The computer device 502 may also include any memory 506 for storing any kind of information such as code, settings, data, etc. In one specific embodiment, a computer program is stored on the memory 506 and can run on the processor 504. When the computer program is run by the processor 504, it can execute instructions according to the above method. Non-limitingly, for example, the memory 506 may include any type of RAM, any type of ROM, flash memory device, hard disk, optical disk, etc. More generally, any memory can use any technology to store information. Further, any memory can provide volatile or non-volatile retention of information. Further, any memory can represent a fixed or removable component of the computer device 502. In one scenario, when processor 504 executes associated instructions stored in any memory or combination of memories, computer device 502 can perform any operation of the associated instructions. Computer device 502 also includes one or more drive mechanisms 508 for interacting with any memory, such as hard disk drive mechanisms, optical disk drive mechanisms, etc.

[0105] Computer device 502 may also include an input / output module 510 (I / O) for receiving various inputs (via input device 512) and providing various outputs (via output device 514). A specific output mechanism may include a presentation device 516 and an associated graphical user interface 518 (GUI). In other embodiments, the input / output module 510 (I / O), input device 512, and output device 514 may be omitted, and the device may function solely as a computer device within a network. Computer device 502 may also include one or more network interfaces 520 for exchanging data with other devices via one or more communication links 522. One or more communication buses 524 couple the components described above together.

[0106] Communication link 522 can be implemented in any way, such as via a local area network, a wide area network (e.g., the Internet), a point-to-point connection, or any combination thereof. Communication link 522 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc., governed by any protocol or combination of protocols.

[0107] Corresponding to Figures 1-3 In addition to the methods described above, this embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the above-described methods.

[0108] This embodiment also provides a computer-readable instruction, wherein when a processor executes the instruction, the program therein causes the processor to perform the following: Figures 1 to 3 The method shown.

[0109] It should be understood that in the various embodiments of this document, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this document.

[0110] It should also be understood that, in the embodiments herein, the term "and / or" is merely a description of the relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following associated objects have an "or" relationship.

[0111] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this document.

[0112] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0113] In the embodiments provided herein, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the couplings or direct couplings or communication connections shown or discussed may be indirect couplings or communication connections through some interfaces, devices, or units, or they may be electrical, mechanical, or other forms of connection.

[0114] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments described herein, depending on actual needs.

[0115] Furthermore, the functional units in the various embodiments of this document can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0116] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this paper, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this paper. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0117] This document uses specific embodiments to illustrate the principles and implementation methods of this document. The descriptions of the embodiments above are only for the purpose of helping to understand the methods and core ideas of this document. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this document. Therefore, the content of this specification should not be construed as a limitation of this document.

Claims

1. A risk control method supporting hierarchical feature backtracking, characterized in that, include: Standard features are calculated based on the user's original data; Based on the standard features, derived features are calculated to generate a feature hierarchy relationship. The feature hierarchy relationship is used to characterize the standard features and derived features, as well as the operational logic between derived features. The derived feature at the top level is taken as the target feature. The feature hierarchy relationship uses the standard features as the base level and the derived features as the subsequent levels. The derived features at a certain level are features without specific physical meaning composed of the standard features or the derived features of the previous level. Based on the feature hierarchy, a hierarchy table is established. The hierarchy table records the parent feature of each derived feature. The parent feature is the feature that the derived feature is directly based on when it is generated. Based on the hierarchical relationship table, the feature results of the target feature are obtained by hierarchical recursive calculation. The feature result obtained by hierarchically and recursively calculating the target feature based on the hierarchical relationship table further includes: Use the target feature as the current feature; Based on the hierarchical relationship table, determine whether the current feature has a parent feature; If so, determine the feature result of the parent feature of the current feature; calculate the feature result of the current feature based on the feature result of the parent feature of the current feature; If not, then the feature result of the current feature is calculated based on the original data.

2. The risk control method supporting hierarchical feature backtracking according to claim 1, characterized in that, The feature result for determining the parent feature of the current feature further includes: The parent feature of the current feature is used as the current feature, and the judgment step is executed.

3. The risk control method supporting hierarchical feature backtracking according to claim 1, characterized in that, The step of calculating the feature result of the current feature based on the feature result of the parent feature of the current feature further includes: The feature result of the parent feature of the current feature is input into the trained tree model to obtain the feature result of the current feature.

4. The risk control method supporting hierarchical feature backtracking according to claim 1, characterized in that, Also includes: The user's historical risk control results are used to make decisions on the feature results of the target feature, and the effectiveness of the target feature obtained based on the feature hierarchy relationship is verified.

5. The risk control method supporting hierarchical feature backtracking according to claim 4, characterized in that, The step of using the user's historical risk control results to make decisions on the feature results of the target features, and verifying the effectiveness of the target features obtained based on the feature hierarchy relationship, further includes: Calculate the proportion of users whose historical risk control results are consistent with the feature results of the target feature; If the user ratio is greater than or equal to the set ratio, then the target feature obtained based on the feature hierarchy relationship is valid; If the user ratio is less than the set ratio, the target feature obtained based on the feature hierarchy relationship is invalid.

6. The risk control method supporting hierarchical feature backtracking according to claim 5, characterized in that, The proportion of users with different historical risk control results is the same.

7. The risk control method supporting hierarchical feature backtracking according to claim 4, characterized in that, Also includes: When the target feature obtained based on the feature hierarchy is invalid, the root cause of the invalidity of the target feature is determined based on the feature results of the standard feature and the derived feature, according to the hierarchy table.

8. A risk control device supporting hierarchical feature backtracking, characterized in that, The device includes: The standard feature determination module uses the user's raw data to calculate standard features. The derived feature determination module is used to calculate derived features based on the standard features, generate a feature hierarchy relationship, and the feature hierarchy relationship is used to characterize the standard features and derived features, as well as the operational logic between derived features, with the top-level derived feature as the target feature; wherein the feature hierarchy relationship is based on the standard features as the base level and the derived features as the subsequent levels, and the derived features of a certain level are features without specific physical meaning composed of the standard features or the derived features of the previous level; The relationship table creation module is used to create a hierarchical relationship table based on the feature hierarchy relationship. The hierarchical relationship table records the parent feature of each derived feature. The parent feature is the feature that the derived feature is directly based on when it is generated. The feature result calculation module is used to recursively calculate the feature result of the target feature based on the hierarchical relationship table, wherein the target feature is taken as the current feature; based on the hierarchical relationship table, it is determined whether the current feature has a parent feature; if so, the feature result of the parent feature of the current feature is determined; the feature result of the current feature is calculated based on the feature result of the parent feature of the current feature; if not, the feature result of the current feature is calculated based on the original data.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, When the computer program is run by the processor, it executes the instructions of the method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is run by the processor of the computer device, it executes the instructions of the method according to any one of claims 1-7.

Images