Risk information assessment method, device and equipment
By acquiring and processing various types of customer churn risk information, and using preset processing and update strategies to train the evaluation model, the problems of low accuracy and high computing resource consumption in the existing technology are solved, and higher quality risk assessment and more stable model performance are achieved.
Patent Information
- Application Number
- CN202510149838.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-11
- Publication Date
- 2025-05-16
AI Technical Summary
The existing technology has problems such as low accuracy, poor data quality and high computing resource consumption in the prediction of customer churn risk, which is difficult to meet the practical application needs.
By obtaining multiple types of information to be evaluated by the object, using preset processing strategies and update strategies to process and update the sample feature information, and training the evaluation model to obtain risk assessment results.
It improves the convergence speed of the model, reduces the consumption of computing resources, enhances the robustness and adaptability of the model, and can more accurately reflect data characteristics.
Smart Images
Figure CN120013254A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of big data and financial technology, and specifically to a risk information assessment method, device and equipment. Background Art
[0002] In a complex market environment, customer churn is prone to occur due to the influence of multiple factors (such as subjective factors of customers, internal factors of the organization, and external environmental factors). In related technologies, the analysis methods for customer churn mainly include statistical analysis methods and traditional machine learning methods to predict the risks and causes of customer churn.
[0003] However, statistical analysis methods have strict requirements on assumptions, resulting in low pertinence and accuracy in prediction results. For traditional machine learning methods, basic data are difficult to collect and the data quality is low, which makes overfitting problems prone to occur. The model has poor generalization ability on new data and consumes a lot of computing resources, making it difficult to meet actual application needs. Summary of the invention
[0004] In view of the above problems, the present disclosure provides a risk information assessment method, apparatus, device, medium and program product.
[0005] According to a first aspect of the present disclosure, a risk information assessment method is provided, comprising: obtaining multiple types of information to be assessed corresponding to an object, wherein the information to be assessed includes object information and business information; inputting the object information and business information into an assessment model to obtain a risk assessment result corresponding to the object; wherein the assessment model is determined in the following manner: processing multiple types of sample information to be assessed based on a preset processing strategy to obtain sample feature information corresponding to the sample information to be assessed, wherein the sample feature information includes first feature information and second feature information, the second feature information is reference feature information corresponding to the first feature information, and the second feature information includes a reference feature and an initial weight value corresponding to the reference feature; updating the initial weight value based on a preset update strategy to obtain target sample feature information, wherein the preset update strategy includes a parameter update strategy and an information update strategy; and training an initial assessment model using the target sample feature information to obtain an assessment model.
[0006] According to an embodiment of the present disclosure, the business information includes multiple sub-business information, and the method also includes: comparing the risk assessment result with the risk threshold to obtain a comparison result; determining the risk influencing factor from the business information based on the comparison result and the correlation information, the correlation information characterizing the correlation between the multiple sub-business information and the risk assessment result; updating the business recommendation strategy corresponding to the object based on the risk influencing factor.
[0007] According to an embodiment of the present disclosure, the initial weight value is updated based on a preset update strategy to obtain target sample feature information, including: determining multiple initial information to be updated corresponding to the initial weight value, wherein the initial information to be updated includes an initial speed parameter, an initial position parameter and an initial parameter; based on the preset update strategy, the initial information to be updated and the objective function value, determining the target sample feature information.
[0008] According to an embodiment of the present disclosure, based on a preset update strategy, initial information to be updated and an objective function value, target sample characteristic information is determined, including: using a parameter update strategy to update initial parameters to obtain target parameters; based on the information update strategy and the target parameters, initial speed parameters and initial position parameters are updated to determine the target sample characteristic information.
[0009] According to an embodiment of the present disclosure, initial speed parameters and initial position parameters are updated based on an information update strategy and target parameters to determine target sample feature information, including: updating the initial speed parameters and initial position parameters based on the information update strategy and target parameters to obtain multiple intermediate update information, the multiple intermediate update information including intermediate speed parameters and intermediate position parameters; determining the difference between the intermediate update information and the target function value; determining the intermediate update information as the target update information when the difference meets a preset condition; and updating the initial weight value based on the target update information to obtain the target sample feature information.
[0010] According to an embodiment of the present disclosure, a preset processing strategy includes a feature processing strategy and an association update strategy; multiple types of sample information to be evaluated are processed based on the preset processing strategy to obtain sample feature information corresponding to the sample information to be evaluated, including: processing the sample information to be evaluated based on the feature processing strategy to obtain first feature information and second feature information; determining the sample feature information based on the association update strategy, the first feature information and the second feature information.
[0011] According to an embodiment of the present disclosure, the sample information to be evaluated is processed based on a feature processing strategy to obtain first feature information and second feature information, including: performing dimensionality reduction processing on the sample information to be evaluated based on the feature processing strategy to obtain the first feature information; and constructing the second feature information using the first feature information.
[0012] According to an embodiment of the present disclosure, sample feature information is determined based on an association update strategy, first feature information, and second feature information, including: determining an initial degree of association between the first feature information and the second feature information; and updating the initial degree of association based on the association update strategy to obtain sample feature information.
[0013] The second aspect of the present disclosure provides a risk information assessment device, including: an information to be assessed acquisition module, used to acquire multiple types of information to be assessed corresponding to an object, wherein the information to be assessed includes object information and business information; an information input module, used to input the object information and business information into an assessment model to obtain a risk assessment result corresponding to the object; wherein the assessment model is determined in the following manner: based on a preset processing strategy, multiple types of sample information to be assessed are processed to obtain sample feature information corresponding to the sample information to be assessed, wherein the sample feature information includes first feature information and second feature information, the second feature information is reference feature information corresponding to the first feature information, and the second feature information includes a reference feature and an initial weight value corresponding to the reference feature; based on a preset update strategy, the initial weight value is updated to obtain target sample feature information, wherein the preset update strategy includes a parameter update strategy and an information update strategy; and the target sample feature information is used to train an initial assessment model to obtain an assessment model.
[0014] A third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.
[0015] The fourth aspect of the present disclosure further provides a computer-readable storage medium on which a computer program or instruction is stored, and the steps of the above method are implemented when the above computer program or instruction is executed by a processor.
[0016] The fifth aspect of the present disclosure further provides a computer program product, including a computer program or instructions, which implement the steps of the above method when the above computer program or instructions are executed by a processor.
[0017] According to the risk information assessment method, device, equipment, medium and program product provided by the present invention, the first feature information and the second feature information of different dimensions corresponding to the sample information to be evaluated can be obtained through a preset processing strategy, so that the second feature information is updated using the preset update strategy to obtain the target sample feature information. Since the reference features and initial weight values in the second feature information are updated using the parameter update strategy and the information update strategy, the target sample feature information can more comprehensively reflect the characteristics of the data. Higher-quality data and reasonable features can accelerate the convergence speed of the model and reduce the consumption of computing resources. At the same time, the combination of different types of features makes the model more stable when facing new data, thereby enhancing the robustness and adaptability of the model. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] The above contents and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:
[0019] Figure 1 The application scenario diagram of the risk information assessment method, apparatus, device, medium and program product according to the embodiment of the present disclosure is schematically shown;
[0020] Figure 2 A flowchart of a risk information assessment method according to an embodiment of the present disclosure is schematically shown;
[0021] Figure 3 The following schematic diagrams show an example of a risk information assessment process according to an embodiment of the present disclosure, wherein (a) is an example of a schematic diagram of a training process of an assessment model, and (b) is an example of a schematic diagram of a risk assessment result determination process;
[0022] Figure 4 The structure block diagram of the risk information assessment device according to the embodiment of the present disclosure is schematically shown;
[0023] Figure 5 A block diagram of an electronic device suitable for implementing a risk information assessment method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION
[0024] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.
[0025] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.
[0026] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.
[0027] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).
[0028] In the technical solution of the present disclosure, the user information or object information (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.
[0029] In the scenario of using personal information for automated decision-making, the methods, devices, and systems provided by the embodiments of the present disclosure provide users with corresponding operation portals for users to choose to agree or reject the automated decision-making results; if the user chooses to reject, the expert decision-making process will be entered. The expression "automated decision-making" here refers to the activity of automatically analyzing and evaluating a person's behavioral habits, interests and hobbies, or economic, health, credit status, etc. through computer programs, and making decisions. The expression "expert decision-making" here refers to the activity of making decisions by people who specialize in a certain field, have specialized experience, knowledge and skills, and have reached a certain level of professionalism.
[0030] In the process of conceiving the present disclosure, the inventors found that in the related technologies, the statistical-based analysis methods have strict requirements on the assumptions, resulting in low pertinence and accuracy of the prediction results; for traditional machine learning methods, basic data are difficult to collect and the data quality is low, which easily leads to overfitting problems, resulting in poor generalization ability of the model on new data and high consumption of computing resources, making it difficult to meet actual application needs.
[0031] In view of this, the present invention can obtain the first feature information and the second feature information of different dimensions corresponding to the sample information to be evaluated through a preset processing strategy, and then use the preset update strategy to update the second feature information to obtain the target sample feature information. Since the reference features and initial weight values in the second feature information are updated by using the parameter update strategy and the information update strategy, the target sample feature information can more comprehensively reflect the characteristics of the data. Higher-quality data and reasonable features can accelerate the convergence speed of the model and reduce the consumption of computing resources. At the same time, the combination of different types of features makes the model more stable when facing new data, and enhances the robustness and adaptability of the model.
[0032] Embodiments of the present disclosure provide a risk information assessment method, apparatus, device, medium and program product, the assessment method comprising: obtaining multiple types of information to be assessed corresponding to an object, wherein the information to be assessed includes object information and business information; inputting the object information and business information into an assessment model to obtain a risk assessment result corresponding to the object; wherein the assessment model is determined in the following manner: processing multiple types of sample information to be assessed based on a preset processing strategy to obtain sample feature information corresponding to the sample information to be assessed, wherein the sample feature information includes first feature information and second feature information, the second feature information is reference feature information corresponding to the first feature information, and the second feature information includes a reference feature and an initial weight value corresponding to the reference feature; updating the initial weight value based on a preset update strategy to obtain target sample feature information, wherein the preset update strategy includes a parameter update strategy and an information update strategy; training an initial assessment model using the target sample feature information to obtain an assessment model.
[0033] Figure 1 The application scenario diagram of the risk information assessment method, apparatus, device, medium and program product according to the embodiments of the present disclosure is schematically shown.
[0034] like Figure 1 As shown, the application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.
[0035] The user can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).
[0036] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.
[0037] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.
[0038] It should be noted that the risk information assessment method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the risk information assessment device provided in the embodiment of the present disclosure can generally be set in the server 105. The risk information assessment method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the risk information assessment device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.
[0039] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.
[0040] Figure 2 The flowchart of the risk information assessment method according to the embodiment of the present disclosure is schematically shown.
[0041] like Figure 2 As shown, the risk information assessment method of this embodiment includes operations S210 to S220.
[0042] In operation S210, multiple types of information to be evaluated corresponding to an object are acquired, wherein the information to be evaluated includes object information and business information.
[0043] In the embodiments of the present disclosure, an object may represent a customer corresponding to an institution's related business or product. The information to be evaluated may be relevant information used to evaluate the risk of customer churn within a target time period, and may include object information and business information of different types or angles. Object information may represent basic information of the object itself, and business information may be product information or transaction information from a different angle than object information. The information to be evaluated may be obtained through application programming interfaces, customer relationship management software, and automated analysis tools, etc., which are not specifically limited here.
[0044] It should be noted that in the embodiments of the present disclosure, before obtaining the information to be evaluated, the user's consent or authorization may be obtained. For example, before operation S210, a request to obtain user information may be issued to the user (object). When the user agrees or authorizes that the user information may be obtained, operation S210 is performed.
[0045] In operation S220, the object information and the business information are input into the assessment model to obtain a risk assessment result corresponding to the object.
[0046] In the embodiments of the present disclosure, the evaluation model may be a model obtained by training and optimizing the initial evaluation model through target sample feature information. The risk evaluation result may represent the result that the subject to be evaluated has a loss risk.
[0047] In the embodiments of the present disclosure, a corresponding operation entry may be provided for the user to choose to agree or reject the automated decision result. That is, before the object information is input into the evaluation model, the user may obtain the instruction of agreeing or rejecting the processing / decision input through the corresponding operation entry. If the user agrees to process, the object information is input into the evaluation model, and step S220 is executed. If the user refuses to process / decision, the expert decision process is entered.
[0048] Preferably, the evaluation model is determined in the following manner: in operation S221, multiple types of sample information to be evaluated are processed based on a preset processing strategy to obtain sample feature information corresponding to the sample information to be evaluated, wherein the sample feature information includes first feature information and second feature information, the second feature information is reference feature information corresponding to the first feature information, and the second feature information includes a reference feature and an initial weight value corresponding to the reference feature; in operation S222, the initial weight value is updated based on a preset update strategy to obtain target sample feature information, wherein the preset update strategy includes a parameter update strategy and an information update strategy; in operation S223, the target sample feature information is used to train an initial evaluation model to obtain an evaluation model.
[0049] In an embodiment of the present disclosure, a preset processing strategy may characterize a method for correlation analysis between different types of data. The sample information to be evaluated may be various types of sample information corresponding to multiple sample objects acquired within a historical time period, including sample object information and sample business information. The sample feature information may characterize the correlation feature information between various types of sample information, and may include first feature information and second feature information. The first feature information may be feature information that is highly correlated with customer churn, and the second feature information may be a reference feature and an initial weight value corresponding to the first feature information. The initial weight value may characterize an initial trade-off parameter between the first feature information and the second feature information. The reference feature may provide domain-related prior knowledge for the evaluation model. Such prior knowledge enables the model to better understand the data distribution and provide the generalization capability of the model.
[0050] In an embodiment of the present disclosure, the preset update strategy may represent a method for updating the initial weight value of the reference feature, including a parameter update strategy and an information update strategy. The information update strategy may be an optimization algorithm based on swarm intelligence, and the parameter update strategy may be a method for optimizing and updating parameter information in the optimization algorithm.
[0051] For example, the first feature information and the second feature information can be determined by combining the kernel function on the basis of the correlation analysis method, including: calculating the kernel matrix through the radial basis function and mapping the sample information to the high-dimensional space, centralizing the kernel matrix to obtain the target kernel matrix after eliminating the influence of the mean; then performing singular value decomposition on the target kernel matrix to obtain the maximum singular value and the corresponding singular vector; and then determining the linear coefficient vector of the object information and the business information according to the result of the singular value decomposition. After obtaining the linear coefficient vector, it can be used to extract the feature information, thereby obtaining the first feature information and the second feature information.
[0052] According to the embodiments of the present disclosure, the first feature information and the second feature information of different dimensions corresponding to the sample information to be evaluated can be obtained through a preset processing strategy, so that the second feature information can be updated using a preset update strategy to obtain the target sample feature information. Since the reference features and initial weight values in the second feature information are updated using the parameter update strategy and the information update strategy, the target sample feature information can more comprehensively reflect the characteristics of the data. Higher-quality data and reasonable features can accelerate the convergence speed of the model and reduce the consumption of computing resources. At the same time, the combination of different types of features makes the model more stable when facing new data, thereby enhancing the robustness and adaptability of the model.
[0053] According to an embodiment of the present disclosure, the business information includes multiple sub-business information, and the method also includes: comparing the risk assessment result with the risk threshold to obtain a comparison result; determining the risk influencing factor from the business information based on the comparison result and the correlation information, the correlation information characterizing the correlation between the multiple sub-business information and the risk assessment result; updating the business recommendation strategy corresponding to the object based on the risk influencing factor.
[0054] In the embodiments of the present disclosure, the risk threshold can be determined based on experience or historical data, and the business information can include multiple sub-business information, such as business type, quota information, and usage information. In the case where the target object is a customer with a risk of churn, the factor with the greatest risk impact can be determined from the multiple sub-business information based on the multiple sub-information corresponding to the target object, and then the business recommendation strategy corresponding to the target object can be determined based on the magnitude of the risk impact.
[0055] For example, for customers with high risk of churn, refined management measures can be taken to provide personalized products, such as address-based discount plans or exclusive customer service. For customers with medium risk, customer satisfaction can be improved through regular follow-up visits, such as providing free product use or service upgrades. For customers with low risk, customer loyalty can be maintained by providing new products, such as regularly pushing new product information and providing priority experience opportunities.
[0056] According to the embodiments of the present disclosure, by updating the service recommendation strategy for customers according to the customer churn risk assessment results and the risk threshold, customer satisfaction and loyalty can be improved and customer churn can be reduced.
[0057] It can be understood that the above has explained how to update the business recommendation strategy based on the risk assessment results and risk influencing factors. The following will explain how to determine the target sample characteristics.
[0058] According to an embodiment of the present disclosure, a preset processing strategy includes a feature processing strategy and an association update strategy; multiple types of sample information to be evaluated are processed based on the preset processing strategy to obtain sample feature information corresponding to the sample information to be evaluated, including: processing the sample information to be evaluated based on the feature processing strategy to obtain first feature information and second feature information; determining the sample feature information based on the association update strategy, the first feature information and the second feature information.
[0059] In an embodiment of the present disclosure, a feature processing strategy may characterize a method for extracting features of different types of information of the same object to obtain correlations between different feature information. In multi-view learning, Universum data may be introduced to update the boundaries of multi-view learning, and different types of first feature information and second feature information may be more accurately divided. In the case where the sample information to be evaluated is sample object information and sample business information, data irrelevant to the target classification problem may be selected as Universum data. The association update strategy may be used to process the first feature information and the second feature information so that the correlation between the projections in the high-dimensional feature space is maximized.
[0060] For example, if the goal is to classify customers into high-risk customers and low-risk customers, other irrelevant customer data (such as customer data in non-target markets) can be used as Universum data. n ], sample business information set B=[b1,b2,…,b n ], you can u =[a 1u ,a 2u ,…,a nu ] and B u =[b 1u ,b 2 u ,…,b nu ] as Universum data.
[0061] For example, when generating sample reference information based on the sample information to be evaluated, when analyzing the correlation between a1 (male, state-owned enterprise, general, annual product usage for 3-5 years, highest degree of master's degree, age 25-30, etc.) and b1 (business type, bank product contract information, bank product usage information, etc.), a2 (female, private enterprise, general, product usage time for 1-3 years, highest degree of master's degree, age 25-30) and b2 (business type, product contract information, product usage information) can be used as Universum samples.
[0062] In the embodiments of the present disclosure, by introducing Universum data, domain-related prior knowledge can be provided to the model, which helps the model to better understand the data distribution and improve the generalization performance of the learning tool; at the same time, by introducing Universum data, the decision boundary of multi-view learning can be modified to make it more accurately divide different categories of data, thereby improving the performance and robustness of the classifier.
[0063] According to an embodiment of the present disclosure, the sample information to be evaluated is processed based on a feature processing strategy to obtain first feature information and second feature information, including: performing dimensionality reduction processing on the sample information to be evaluated based on the feature processing strategy to obtain the first feature information; and constructing the second feature information using the first feature information.
[0064] In an embodiment of the present disclosure, Universum data is introduced into the kernel canonical correlation analysis KCCA, and by modifying the objective function, the distance between the Universum data and the sample information to be evaluated can be made as large as possible. For example, a new metric matrix can be constructed so that the distance between points of the same type is as small as possible, the distance between points of different types is as large as possible, and the distance between the Universum data and the target type data (sample information to be evaluated) is as large as possible. The first feature information may be a feature corresponding to the sample information to be evaluated, and the second feature information may be a feature (reference feature) corresponding to the Universum data.
[0065] Understandably, i and b i can be the target observation value of the same object in two perspectives, and a iu and b iu is the universe observation value of the same object in two perspectives. UKCCA aims to search for the projection vector w a and w b The objective function can be maximized, and the objective function ρ can be expressed as follows:
[0066] (1);
[0067] in, , are respectively related to the projection vector w a and w b The corresponding typical components, a can represent sample object information, and b can represent sample business information. is the cross-covariance matrix of the sample information set to be evaluated, and is the autocovariance matrix.
[0068] According to an embodiment of the present disclosure, sample feature information is determined based on an association update strategy, first feature information, and second feature information, including: determining an initial degree of association between the first feature information and the second feature information; and updating the initial degree of association based on the association update strategy to obtain sample feature information.
[0069] In an embodiment of the present disclosure, the association update strategy can be a method for processing nonlinear correlation data, which is used to process the acquired first feature information and second feature information of different types so that the correlation between the projections in the high-dimensional feature space is maximized, such as kernel canonical correlation analysis (Kernel CCA, KCCA).
[0070] For example, suppose the sample information to be evaluated is the sample object information set A=[a1,a2,…,a n ], sample business information set B=[b1,b2,…,b n ], using the kernel function combined with the associated update strategy (such as kernel canonical correlation analysis) to obtain the first feature information A i and the second feature information B i ; Thus, the kernel function is used to construct the kernel matrix K of sample object information and sample business information A and K B To eliminate the influence of the mean value of the data, the kernel matrix can be centralized to obtain K' A and K' B ; Then, the kernel canonical correlation analysis method is used to find the kernel matrix projection directions θ and δ, so that the correlation between different projection directions in the high-dimensional feature space is maximized; after the projection directions θ and δ are obtained, the original data can be projected into the low-dimensional space to obtain the feature representation after projection. It can be understood that the above objective function ρ can be converted into the following formula (2):
[0071] (2);
[0072] Among them, st represents the constraint condition and max represents the maximization process.
[0073] In a feasible embodiment, sample reference information of Universum data and sample reference information of Reuters21578 can also be introduced on the basis of kernel canonical correlation analysis, and the objective function ρ can be expressed as the following optimal problem, as shown in the following formula (3):
[0074] (3);
[0075] in, , , η and λ can be used to balance the trade-off parameters between the sample reference information of Universum data and the sample reference information of Reuters21578, respectively. The optimal solution is found through dynamic adjustment. m can represent the number of sample reference information of Universum data, and n can represent the number of sample reference information of Reuters21578. a iu and b iu is the universe observation value of the same object from two perspectives, a ir and b ir are the observations of the Reuters21578 sample from two views of the same object. , and b iu and b ir Typical weights for each.
[0076] According to the embodiments of the present disclosure, kernel canonical correlation analysis can effectively reduce the dimensionality of data while retaining basic relationships, making it easier to visualize and explain complex interactions. Based on kernel canonical correlation analysis combined with kernel functions to map data to a high-dimensional feature space, it is possible to capture nonlinear relationships in the data, making kernel canonical correlation analysis more advantageous than traditional correlation analysis when processing complex data sets, and can discover complex correlations that may not be obvious in the original feature space, thereby improving the training efficiency of the model.
[0077] According to an embodiment of the present disclosure, the initial weight value is updated based on a preset update strategy to obtain target sample feature information, including: determining multiple initial information to be updated corresponding to the initial weight value, wherein the initial information to be updated includes an initial speed parameter, an initial position parameter and an initial parameter; based on the preset update strategy, the initial information to be updated and the objective function value, determining the target sample feature information.
[0078] In the embodiments of the present disclosure, the initial speed parameter, the initial position parameter, and the initial parameter can respectively represent the iteration speed, iteration position, and iteration search capability information corresponding to the initial weight value in the preset update strategy. The objective function can be used to evaluate the quality of the candidate solution (combination of initial weight values) in the preset update strategy.
[0079] For example, taking the preset update strategy as the particle swarm optimization (PSO) algorithm as an example, the method of optimizing and updating the initial information to be updated based on the particle swarm optimization algorithm may include: initializing the particle swarm corresponding to the initial information to be updated and evaluating the fitness value of the particles; thereby updating the individual extreme value and the global extreme value, updating the particle speed and position and processing the boundary of the particle movement, and stopping the iteration when the iteration only reaches the iteration termination condition to obtain the global extreme value as the optimized and updated target sample feature information.
[0080] According to an embodiment of the present disclosure, based on a preset update strategy, initial information to be updated and an objective function value, target sample characteristic information is determined, including: using a parameter update strategy to update initial parameters to obtain target parameters; based on the information update strategy and the target parameters, initial speed parameters and initial position parameters are updated to determine the target sample characteristic information.
[0081] In the embodiments of the present disclosure, the initial parameter may represent the iterative search capability of the candidate value in the swarm intelligence optimization algorithm, and may also be referred to as the inertia weight. When the second feature information is updated using the swarm intelligence optimization algorithm, the inertia weight may be optimized and updated using the parameter update strategy to obtain the target parameter.
[0082] For example, the method of updating and optimizing the trade-off parameters η and λ using the particle swarm optimization PSO algorithm can be shown in the following formula (4):
[0083] (4);
[0084] in, It can represent the speed of particle i at the tth iteration; ω' can represent the inertia weight (target parameter), which is used to balance the global search ability and local search ability of the particle; c1 and c2 can represent learning factors, which represent the learning ability of the particle from individual experience and group experience respectively; rand1 and rand2 can represent random numbers between [0,1], which are used to increase the randomness and diversity of the search; It can represent the individual extreme value of particle i; Can represent global extreme values; It can represent the position of particle i at the tth iteration.
[0085] According to the updated velocity, the new position of the particle in the next iteration is calculated, that is, the new value of the trade-off parameter, as shown in the following formula (5):
[0086] x i (t+1)=x i (t)+v i(t+1) (5);
[0087] The initial parameter in formula (4) can be recorded as ω. The method of updating ω using the parameter update strategy to obtain the target parameter ω' can be shown in the following formula (5):
[0088] (6);
[0089] Where t is the current iteration number, T max The maximum number of iterations can be represented by and It can be set according to actual conditions or experience, and is not limited here. The cosine function with a value of 0-π is used to dynamically update the value of the inertia weight. It should be noted that the cosine function is decreasing within the interval. The initial inertia weight value is large and the decreasing speed is slow, so as to search the whole better and effectively avoid the problem of falling into the local optimal value; the inertia weight value at the end is small, and it will decrease at a relatively gentle rate of change, so as to better find the local optimal value.
[0090] For the learning factors c1 and c2 in formula (4), a linear algorithm can be used so that c1 changes from large to small and c2 changes from small to large during the iteration process, as shown in the following formula (7):
[0091] (7);
[0092] Where t is the current iteration number, T max The maximum number of iterations can be characterized.
[0093] According to an embodiment of the present disclosure, initial speed parameters and initial position parameters are updated based on an information update strategy and target parameters to determine target sample feature information, including: updating the initial speed parameters and initial position parameters based on the information update strategy and target parameters to obtain multiple intermediate update information, the multiple intermediate update information including intermediate speed parameters and intermediate position parameters; determining the difference between the intermediate update information and the target function value; determining the intermediate update information as the target update information when the difference meets a preset condition; and updating the initial weight value based on the target update information to obtain the target sample feature information.
[0094] In an embodiment of the present disclosure, the process of updating the trade-off parameters using a particle swarm optimization (PSO) algorithm may include: initializing a particle swarm, determining the number and dimension of particles, initializing particle positions and speeds, and randomly generating an initial position of each particle (initial value of a model parameter) in a solution space. At the same time, each particle is randomly assigned an initial speed, and the magnitude and direction of the speed determine how the particle moves in the solution space; thereby, by defining a fitness function and calculating a fitness value, the current position (model parameter) of each particle may be substituted into the fitness function (objective function), and its corresponding fitness value may be calculated to evaluate the performance of the particle; and then, individual extreme values and global extreme values may be updated to obtain intermediate speed parameters and intermediate position parameters; and whether to stop iteration may be determined based on the fitness function or the number of iterations.
[0095] For example, in an iterative update, when the updated speed exceeds a preset maximum speed, it can be limited to a speed threshold (V max ) range to prevent the particle from moving too fast and missing the optimal solution. If the new position is outside the boundary of the solution space, it can be adjusted back to within the boundary to ensure that the particle is always within the valid search range.
[0096] For example, in iterative updates, a maximum number of iterations is pre-set. When the number of iterations reaches this value, the iteration can be stopped and the current global extreme value can be output as the optimized trade-off parameter. Alternatively, if the fitness value corresponding to the global extreme value changes very little or tends to be stable in multiple consecutive iterations, it means that the particle swarm has converged to a better solution, and the iteration can be terminated at this time.
[0097] Figure 3 An example schematic diagram of a risk information assessment process according to an embodiment of the present disclosure is schematically shown, wherein (a) is an example schematic diagram of a training process of an assessment model, and (b) is an example schematic diagram of a risk assessment result determination process.
[0098] like Figure 3 As shown in Figure (a), multiple types of sample information to be evaluated 32 can be processed according to a preset processing strategy 31 to obtain sample feature information 33 corresponding to the sample information to be evaluated 32, wherein the sample feature information 33 includes first feature information 331 and second feature information 332, and the second feature information 332 includes a reference feature 3321 and an initial weight value 3322 corresponding to the reference feature 3321; thereby, the initial weight value 3322 is updated based on a preset update strategy 34 to obtain target sample feature information 35; and further, the target sample feature information 35 can be used to train an initial evaluation model 36 to obtain an evaluation model 37.
[0099] like Figure 3As shown in Figure (b), after obtaining the trained evaluation model 37, multiple types of information to be evaluated 310 corresponding to the object can be obtained, and the information to be evaluated may include object information 311 and business information 312; thereby, the object information 311 and the business information 312 are input into the evaluation model 37 to obtain a risk assessment result 320 corresponding to the object.
[0100] Based on the above risk information assessment method, the present disclosure also provides a risk information assessment device. Figure 4 The device is described in detail.
[0101] Figure 4 The structure block diagram of the risk information assessment device according to an embodiment of the present disclosure is schematically shown.
[0102] like Figure 4 As shown, the risk information assessment device of this embodiment includes a to-be-assessed information acquisition module 410 and an information input module 420 .
[0103] The information to be evaluated acquisition module 410 is used to acquire multiple types of information to be evaluated corresponding to the object, wherein the information to be evaluated includes object information and business information. In one embodiment, the information to be evaluated acquisition module 410 can be used to perform the operation S210 described above, which will not be described in detail here.
[0104] The information input module 420 is used to input the object information and the business information into the assessment model to obtain the risk assessment result corresponding to the object. In one embodiment, the information input module 420 can be used to perform the operation S220 described above, which will not be described in detail here.
[0105] Preferably, the evaluation model is determined in the following manner: based on a preset processing strategy, multiple types of sample information to be evaluated are processed to obtain sample feature information corresponding to the sample information to be evaluated, wherein the sample feature information includes first feature information and second feature information, the second feature information is reference feature information corresponding to the first feature information, and the second feature information includes a reference feature and an initial weight value corresponding to the reference feature; based on a preset update strategy, the initial weight value is updated to obtain target sample feature information, wherein the preset update strategy includes a parameter update strategy and an information update strategy; and the target sample feature information is used to train the initial evaluation model to obtain the evaluation model.
[0106] According to the embodiments of the present disclosure, based on the information to be evaluated acquisition module 410 and the information input module 420 in the risk information assessment device, the first feature information and the second feature information of different dimensions corresponding to the sample information to be evaluated can be obtained through a preset processing strategy, so that the second feature information is updated using the preset update strategy to obtain the target sample feature information. Since the reference features and initial weight values in the second feature information are updated using the parameter update strategy and the information update strategy, the target sample feature information can more comprehensively reflect the characteristics of the data. Higher-quality data and reasonable features can accelerate the convergence speed of the model and reduce the consumption of computing resources. At the same time, the combination of different types of features makes the model more stable when facing new data, thereby enhancing the robustness and adaptability of the model.
[0107] According to an embodiment of the present disclosure, the business information includes multiple sub-business information, and the device also includes: a result comparison module, a factor determination module and a strategy update module. The result comparison module is used to compare the risk assessment result with the risk threshold to obtain a comparison result; the factor determination module is used to determine the risk influencing factor from the business information based on the comparison result and the correlation information, and the correlation information represents the correlation between the multiple sub-business information and the risk assessment result; the strategy update module is used to update the business recommendation strategy corresponding to the object based on the risk influencing factor.
[0108] According to an embodiment of the present disclosure, the initial weight value is updated based on a preset update strategy to obtain target sample feature information, including: determining multiple initial information to be updated corresponding to the initial weight value, wherein the initial information to be updated includes an initial speed parameter, an initial position parameter and an initial parameter; based on the preset update strategy, the initial information to be updated and the objective function value, determining the target sample feature information.
[0109] According to an embodiment of the present disclosure, based on a preset update strategy, initial information to be updated and an objective function value, target sample characteristic information is determined, including: using a parameter update strategy to update initial parameters to obtain target parameters; based on the information update strategy and the target parameters, initial speed parameters and initial position parameters are updated to determine the target sample characteristic information.
[0110] According to an embodiment of the present disclosure, initial speed parameters and initial position parameters are updated based on an information update strategy and target parameters to determine target sample feature information, including: updating the initial speed parameters and initial position parameters based on the information update strategy and target parameters to obtain multiple intermediate update information, the multiple intermediate update information including intermediate speed parameters and intermediate position parameters; determining the difference between the intermediate update information and the target function value; determining the intermediate update information as the target update information when the difference meets a preset condition; and updating the initial weight value based on the target update information to obtain the target sample feature information.
[0111] According to an embodiment of the present disclosure, a preset processing strategy includes a feature processing strategy and an association update strategy; multiple types of sample information to be evaluated are processed based on the preset processing strategy to obtain sample feature information corresponding to the sample information to be evaluated, including: processing the sample information to be evaluated based on the feature processing strategy to obtain first feature information and second feature information; determining the sample feature information based on the association update strategy, the first feature information and the second feature information.
[0112] According to an embodiment of the present disclosure, the sample information to be evaluated is processed based on a feature processing strategy to obtain first feature information and second feature information, including: performing dimensionality reduction processing on the sample information to be evaluated based on the feature processing strategy to obtain the first feature information; and constructing the second feature information using the first feature information.
[0113] According to an embodiment of the present disclosure, sample feature information is determined based on an association update strategy, first feature information, and second feature information, including: determining an initial degree of association between the first feature information and the second feature information; and updating the initial degree of association based on the association update strategy to obtain sample feature information.
[0114] According to an embodiment of the present disclosure, any multiple modules in the information acquisition module 410 to be evaluated and the information input module 420 can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the information acquisition module 410 to be evaluated and the information input module 420 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware and firmware or in any appropriate combination of any of them. Alternatively, at least one of the information acquisition module 410 to be evaluated and the information input module 420 can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be executed.
[0115] Figure 5 A block diagram of an electronic device suitable for implementing a risk information assessment method according to an embodiment of the present disclosure is schematically shown.
[0116] like Figure 5As shown, the electronic device according to the embodiment of the present disclosure includes a processor 501, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 502 or the program loaded from the storage part 508 to the random access memory (RAM) 503. The processor 501 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 501 may also include an onboard memory for caching purposes. The processor 501 may include a single processing unit or multiple processing units for performing different actions of the method flow according to the embodiment of the present disclosure.
[0117] In RAM 503, various programs and data required for the operation of the electronic device are stored. The processor 501, ROM 502 and RAM 503 are connected to each other via a bus 504. The processor 501 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 502 and / or RAM 503. It should be noted that the program can also be stored in one or more memories other than ROM 502 and RAM 503. The processor 501 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.
[0118] According to an embodiment of the present disclosure, the electronic device may further include an input / output (I / O) interface 505, which is also connected to the bus 504. The electronic device may further include one or more of the following components connected to the input / output (I / O) interface 505: an input portion 506 including a keyboard, a mouse, etc.; an output portion 507 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 508 including a hard disk, etc.; and a communication portion 509 including a network interface card such as a LAN card, a modem, etc. The communication portion 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the input / output (I / O) interface 505 as needed. A removable medium 511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 510 as needed, so that a computer program read therefrom is installed into the storage portion 508 as needed.
[0119] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.
[0120] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, it may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present disclosure, the computer-readable storage medium may include the ROM 502 and / or RAM 503 described above and / or one or more memories other than ROM 502 and RAM 503.
[0121] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the risk information assessment method provided by the embodiment of the present disclosure.
[0122] The above functions defined in the system / device of the embodiment of the present disclosure are performed when the computer program is executed by the processor 501. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.
[0123] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 509, and / or installed from the removable medium 511. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.
[0124] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 509, and / or installed from the removable medium 511. When the computer program is executed by the processor 501, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, means, module, unit, etc. described above can be implemented by a computer program module.
[0125] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).
[0126] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.
[0127] It will be appreciated by those skilled in the art that the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations or combinations are not explicitly described in the present disclosure. In particular, without departing from the spirit and teachings of the present disclosure, the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.
[0128] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.
Claims
1. A risk information assessment method, characterized in that: The method comprises: Acquire multiple types of information to be evaluated corresponding to the object, wherein the information to be evaluated includes object information and business information; Inputting the object information and the business information into an assessment model to obtain a risk assessment result corresponding to the object; The evaluation model is determined in the following way: Processing multiple types of sample information to be evaluated based on a preset processing strategy to obtain sample feature information corresponding to the sample information to be evaluated, wherein the sample feature information includes first feature information and second feature information, the second feature information is reference feature information corresponding to the first feature information, and the second feature information includes a reference feature and an initial weight value corresponding to the reference feature; The initial weight value is updated based on a preset update strategy to obtain target sample feature information, wherein the preset update strategy includes a parameter update strategy and an information update strategy; The target sample feature information is used to train an initial evaluation model to obtain the evaluation model.
2. The evaluation method according to claim 1, characterized in that: The service information includes a plurality of sub-service information, and the method further includes: Comparing the risk assessment result with the risk threshold to obtain a comparison result; Determining risk influencing factors from the business information based on the comparison result and correlation information, wherein the correlation information represents the correlation between the plurality of sub-business information and the risk assessment result; A service recommendation strategy corresponding to the object is updated based on the risk influencing factors.
3. The evaluation method according to claim 1, characterized in that: The initial weight value is updated based on a preset update strategy to obtain target sample feature information, including: Determine a plurality of initial information to be updated corresponding to the initial weight value, wherein the initial information to be updated includes an initial speed parameter, an initial position parameter and an initial parameter; Based on the preset update strategy, the initial information to be updated and the objective function value, target sample feature information is determined.
4. The evaluation method according to claim 3, characterized in that: Determining target sample feature information based on the preset update strategy, the initial information to be updated, and the objective function value includes: Using the parameter updating strategy to update the initial parameters, to obtain target parameters; The initial speed parameter and the initial position parameter are updated based on the information update strategy and the target parameter to determine the target sample feature information.
5. The evaluation method according to claim 4, characterized in that: The initial speed parameter and the initial position parameter are updated based on the information update strategy and the target parameter to determine the target sample feature information, including: Based on the information update strategy and the target parameter, the initial speed parameter and the initial position parameter are updated to obtain a plurality of intermediate update information, wherein the plurality of intermediate update information includes an intermediate speed parameter and an intermediate position parameter; Determining a difference between the intermediate update information and an objective function value; In the case where the difference satisfies a preset condition, determining the intermediate update information as target update information; The initial weight value is updated based on the target update information to obtain the target sample feature information.
6. The evaluation method according to claim 1, characterized in that: The preset processing strategy includes a feature processing strategy and an association update strategy; Based on a preset processing strategy, multiple types of sample information to be evaluated are processed to obtain sample feature information corresponding to the sample information to be evaluated, including: Processing the sample information to be evaluated based on the feature processing strategy to obtain the first feature information and the second feature information; The sample feature information is determined based on the association update strategy, the first feature information, and the second feature information.
7. The evaluation method according to claim 6, characterized in that: Processing the sample information to be evaluated based on the feature processing strategy to obtain the first feature information and the second feature information includes: Performing dimensionality reduction processing on the sample information to be evaluated based on the feature processing strategy to obtain the first feature information; The second feature information is constructed using the first feature information.
8. The evaluation method according to claim 6, characterized in that: Determining the sample feature information based on the association update strategy, the first feature information, and the second feature information includes: determining an initial correlation between the first feature information and the second feature information; The initial correlation degree is updated based on the correlation update strategy to obtain the sample feature information.
9. A risk information assessment device, characterized in that: The device comprises: An information acquisition module for evaluating, used to acquire multiple types of information to be evaluated corresponding to the object, wherein the information to be evaluated includes object information and business information; An information input module, used to input the object information and the business information into the assessment model to obtain a risk assessment result corresponding to the object; The evaluation model is determined in the following way: Processing multiple types of sample information to be evaluated based on a preset processing strategy to obtain sample feature information corresponding to the sample information to be evaluated, wherein the sample feature information includes first feature information and second feature information, the second feature information is reference feature information corresponding to the first feature information, and the second feature information includes a reference feature and an initial weight value corresponding to the reference feature; The initial weight value is updated based on a preset update strategy to obtain target sample feature information, wherein the preset update strategy includes a parameter update strategy and an information update strategy; The target sample feature information is used to train an initial evaluation model to obtain the evaluation model.
10. An electronic device, comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 8.