CN115797048A

CN115797048A - Account risk identification method and device, computer equipment and storage medium

Info

Publication number: CN115797048A
Application number: CN202211474634.1A
Authority: CN
Inventors: 梅宝泰; 李春霖; 王备; 彭诗笑
Original assignee: Industrial and Commercial Bank of China Ltd ICBC
Current assignee: Industrial and Commercial Bank of China Ltd ICBC
Priority date: 2022-11-23
Filing date: 2022-11-23
Publication date: 2023-03-14

Abstract

The application relates to an account risk identification method, an account risk identification device, computer equipment, a storage medium and a computer program product, which can be applied to the technical field of artificial intelligence and can improve the accuracy of an account risk identification result. The method comprises the following steps: responding to a service transaction request of a user account for a target service, and acquiring external account data corresponding to the user account related to the target service from an external data system; acquiring updated account characteristics of the user account based on the external account data; determining historical business related to the user account, and acquiring a pre-trained account risk identification model corresponding to the historical business; and inputting the updated account characteristics into the account risk identification model, and updating the historical risk identification result of the user account under the historical service based on the risk identification result output by the account risk identification model.

Description

Account risk identification method and device, computer equipment and storage medium

Technical Field

The present application relates to the field of artificial intelligence technologies, and in particular, to an account risk identification method, apparatus, computer device, storage medium, and computer program product.

Background

With the development of computer technology, when a user transacts related services, related mechanisms can acquire related data in time to identify credit risks existing in the user.

In the related art, risk identification is mainly performed in the process of handling a specific service by a user, a risk identification result is associated with the specific service, and the risk identification result can be referred to when performing related processing subsequently.

However, in practice, when the risk identification result is referred to again later, the risk identification result is different from the actual credit risk of the user, so that the risk identification result is inaccurate.

Disclosure of Invention

In view of the above, it is necessary to provide an account risk identification method, an account risk identification apparatus, a computer device, a computer readable storage medium, and a computer program product, which can improve accuracy of an account risk identification result.

In a first aspect, the present application provides an account risk identification method, including:

responding to a service transaction request of a user account for a target service, and acquiring external account data corresponding to the user account related to the target service from an external data system;

acquiring updated account characteristics of the user account based on the external account data;

determining historical business associated with the user account, and acquiring a pre-trained account risk identification model corresponding to the historical business;

inputting the updated account characteristics into the account risk identification model, and updating the historical risk identification result of the user account under the historical service based on the risk identification result output by the account risk identification model.

In one embodiment, the pre-trained account risk recognition model corresponding to the historical service is obtained by training through the following steps:

acquiring a plurality of training samples based on sample account characteristics corresponding to a plurality of sample user accounts handling the historical service;

under the preset constraint condition and the solution target, training a support vector machine to be trained by utilizing the plurality of training samples; the support vector machine classifies the training samples into high-risk samples and low-risk samples through corresponding hyperplanes, the constraint condition is that the distances between the low-risk samples and the hyperplanes and the distances between the high-risk samples and the hyperplanes meet a preset distance condition, and the solution target is that the sum of the distances between each training sample and the hyperplane is the largest;

and when the training end condition is met, obtaining a trained account risk identification model.

In one embodiment, the obtaining a plurality of training samples based on sample account characteristics corresponding to a plurality of sample user accounts handling the historical service includes:

acquiring initial account characteristics corresponding to a plurality of sample user accounts handling the historical service and a risk label of each sample user account;

mapping the initial account characteristics by adopting a preset kernel function to obtain sample account characteristics corresponding to each sample user account; the feature dimension of the sample account features is greater than the feature dimension of the primary account features;

and obtaining a plurality of training samples based on the sample account characteristics and the risk labels corresponding to each sample user account.

In one embodiment, the obtaining of the primary account characteristics corresponding to each of a plurality of sample user accounts transacting the historical service includes:

respectively acquiring external sample account data and internal sample account data of a plurality of sample user accounts handling the historical service from an external data system and an internal data system;

determining associated data items and non-associated data items in the external sample account data and the internal sample account data, and fusing data values of the external sample account data and the internal sample account data under the associated data items to obtain fused sample account data;

and determining the initial account characteristics corresponding to each sample user account based on the sample account data fused with each sample user account and the sample account data under the non-associated data item.

In one embodiment, the obtaining updated account characteristics of the user account based on the external account data includes:

acquiring original account data acquired in the user account service transaction process from an internal data system, and obtaining the internal account data of the user account based on the original account data;

and fusing the external account data and the internal account data, and acquiring the updated account characteristics of the user account based on the fused account data.

In one embodiment, the fusing the external account data and the internal account data, and acquiring the updated account feature of the user account based on the fused account data includes:

determining associated data items and non-associated data items of the external account data and the internal account data based on respective data items of the external account data and the internal account data;

fusing the data values of the external account data and the internal account data under the associated data items to obtain account data updated by the associated data items;

and acquiring the updated account characteristics of the user account based on the updated account data of the associated data item and the account data under the non-associated data item.

In one embodiment, the obtaining external account data corresponding to the user account associated with the target service from an external data system includes:

determining account attributes associated with the target service;

and acquiring attribute information of the user account under the account attribute from an external data system, and taking the attribute information as external account data corresponding to the user account.

In a second aspect, the application further provides an account risk identification device. The device comprises:

the external data acquisition module is used for responding to a service transaction request of a user account for a target service, and acquiring external account data corresponding to the user account related to the target service from an external data system;

the account characteristic updating module is used for acquiring the updated account characteristics of the user account based on the external account data;

the model acquisition module is used for determining the historical business related to the user account and acquiring a pre-trained account risk identification model corresponding to the historical business;

and the risk identification result acquisition module is used for inputting the updated account characteristics into the account risk identification model and updating the historical risk identification result of the user account under the historical service based on the risk identification result output by the account risk identification model.

In a third aspect, the application also provides a computer device. The computer device comprises a memory storing a computer program and a processor implementing the following steps when executing the computer program:

In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of:

In a fifth aspect, the present application further provides a computer program product. The computer program product comprising a computer program which when executed by a processor performs the steps of:

and inputting the updated account characteristics into the account risk identification model, and updating the historical risk identification result of the user account under the historical service based on the risk identification result output by the account risk identification model.

According to the account risk identification method, the account risk identification device, the computer equipment, the storage medium and the computer program product, in response to a service transaction request of a user account for a target service, external account data corresponding to the user account associated with the target service can be acquired from an external data system, account characteristics after the user account is updated are acquired based on the external account data, historical services associated with the user account are determined, a pre-trained account risk identification model corresponding to the historical services is acquired, the updated account characteristics can be input into the account risk identification model, and historical risk identification results of the user account under the historical services are updated based on risk identification results output by the account risk identification model. According to the method and the device, when the situation that the user transacts the target service and obtains the updated account characteristics is detected, the risk identification result of the user under the historical service can be timely and actively triggered and updated based on the updated account characteristics, and the accuracy of the risk identification result of the historical service is ensured under the condition that the account risk identification model corresponding to the historical service is prevented from actively checking the number.

Drawings

FIG. 1 is a diagram of an application environment of a method for account risk identification in one embodiment;

FIG. 2 is a schematic flow chart diagram illustrating a method for account risk identification, according to an embodiment;

FIG. 3 is a flowchart illustrating steps for obtaining an account risk identification model, in one embodiment;

FIG. 4 is a flowchart illustrating the steps of obtaining training samples in one embodiment;

FIG. 5 is a schematic illustration of a feature space in one embodiment;

FIG. 6 is a schematic flow chart diagram illustrating another account risk identification methodology in one embodiment;

FIG. 7 is a block diagram of an account risk identification device in one embodiment;

FIG. 8 is a diagram of an internal structure of a computer device, in one embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.

It should be noted that the account risk identification method, apparatus, computer device, storage medium, and computer program product provided by the present application may be applied to the technical field of artificial intelligence, and may also be applied to other related fields.

The account risk identification method provided by the embodiment of the application can be applied to an application environment as shown in fig. 1, where the application environment may include a terminal and a server, and the terminal may communicate with the server through a network. The terminal can be but is not limited to various personal computers, notebook computers, smart phones, tablet computers, internet of things equipment and portable wearable equipment, and the internet of things equipment can be smart sound boxes, smart televisions, smart vehicle-mounted equipment and the like; the portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, and the like.

The server can be implemented by an independent server or a server cluster formed by a plurality of servers, the server can have a corresponding data storage system, the data storage system can store data to be processed by the server, for example, internal data can be stored, and the data storage system can be integrated at the server, or can be placed on a cloud or other network servers.

In the application, a user can trigger a service transaction request for a target service through a terminal, respond to the service transaction request for the target service by a user account, a server side can obtain external account data corresponding to the user account associated with the target service from an external data system, then obtain an updated account characteristic of the user account based on the external account data, determine a historical service associated with the user account, obtain a pre-trained account risk identification model corresponding to the historical service, further input the updated account characteristic into the account risk identification model, and update a historical risk identification result of the user account under the historical service based on a risk identification result output by the account risk identification model.

In one embodiment, as shown in fig. 2, an account risk identification method is provided, which is described by taking the application of the method to the server side in fig. 1 as an example, and includes the following steps:

s201, in response to a service transaction request of a user account for a target service, external account data corresponding to the user account associated with the target service is acquired from an external data system.

As an example, the target service may be a service for which the user account applies for transaction, such as a first transaction or a second transaction of the user account.

The data system may be divided into an internal data system and an external data system, where the internal data system refers to a data system owned by a business party or having an access right, and the internal data system may be used to store business data generated when the business party handles business for a user account, for example, if the business party is a banking institution, the internal data system may store basic information (such as information for identifying a user identity) submitted by the user account when handling business, a transaction record of the user account, and the like, and data associated with the user account stored in the internal data system may also be referred to as internal account data.

The external data system may be a data system in which the business party does not have access right, data associated with the user account stored in the external data system may also be referred to as external account data, and the external account data may include data that is not directly obtained from the business processing process of the business party itself in the business processing process.

In this step, the user may trigger a service transaction request for the target service through the terminal, and in response to the service transaction request for the target service from the user account, the server may determine external account data to be acquired, which is associated with the target service, and send an external account data acquisition request for the current user account to the external data system, so as to obtain external account data of the user account returned by the external data system.

S202, based on the external account data, the updated account characteristics of the user account are obtained.

As an example, the account characteristic may be information characterizing an attribute characteristic or behavior characteristic of the user account, such as a representation of the user account, a frequency with which the user account transacts business, and a characteristic of the user account applying for transacting business.

When the user account triggers the transaction target service, new external account data of the user account can be obtained from the external data system, and after the external account data is obtained, the account characteristics pre-stored for the user account can be updated to obtain the updated account characteristics of the user account.

S203, determining historical business related to the user account, and acquiring a pre-trained account risk identification model corresponding to the historical business.

The historical service and the target service are different services, the historical service can be a service applied and transacted by a user account, and the current service transaction state of the historical service can include processing or ending.

An account risk identification model may be used to identify a credit risk (or breach risk) for a user account; the risk identification result of the user account under the historical business can be determined based on a pre-trained account risk identification model, and each business can have a corresponding account risk identification model, so that the credit risk of the user account under one business can be identified by combining the account characteristics of the user account in a targeted manner. Illustratively, the account risk recognition model can be trained through various model training modes.

Compared with the prior art, after the risk identification result of the user account under a service is obtained, whether the user account has the related credit risk or not is determined continuously depending on the risk identification result, according to the method and the device, updating of the risk identification result of the historical service related to the user account can be automatically triggered after the updated account feature of the user account is obtained based on the link relation between the user account and the service, and the risk identification model related to the historical service is prevented from actively carrying out related query. Specifically, historical business transacted by a user account can be determined, and an account risk identification model pre-trained for each historical business is obtained.

And S204, inputting the updated account characteristics into the account risk identification model, and updating the historical risk identification result of the user account under the historical service based on the risk identification result output by the account risk identification model.

After the account risk identification model corresponding to the historical service is obtained, the updated account characteristics of the user account can be input into the account risk identification model, the account risk identification model re-determines the risk identification result of the user account under the corresponding historical service based on the currently input updated account characteristics of the user account, and updates the pre-stored historical risk identification result of the user account under the historical service by using the currently output risk identification result.

In the account risk identification method, in response to a service transaction request of a user account for a target service, external account data corresponding to the user account associated with the target service can be acquired from an external data system, account characteristics updated by the user account are acquired based on the external account data, historical services associated with the user account are determined, a pre-trained account risk identification model corresponding to the historical services is acquired, the updated account characteristics can be input into the account risk identification model, and historical risk identification results of the user account under the historical services are updated based on risk identification results output by the account risk identification model. According to the method and the device, when the situation that the user transacts the target service and obtains the updated account characteristics is detected, the risk identification result of the user under the historical service can be timely and actively triggered and updated based on the updated account characteristics, and the accuracy of the risk identification result of the historical service is ensured under the condition that the account risk identification model corresponding to the historical service is prevented from actively checking the number.

In one embodiment, as shown in fig. 3, the pre-trained account risk recognition model corresponding to the historical service may be obtained by training through the following steps:

s301, obtaining a plurality of training samples based on sample account characteristics corresponding to a plurality of sample user accounts handling historical services.

As an example, the user accounts used to provide training data may be referred to as sample user accounts, and the account features corresponding to the sample user accounts may be referred to as sample account features.

In a specific implementation, a plurality of sample user accounts transacting historical services can be determined, sample account characteristics corresponding to each sample user account can be obtained, and a plurality of training samples can be obtained based on the sample account characteristics corresponding to each sample user account.

S302, under the preset constraint condition and the solution target, training a support vector machine to be trained by utilizing a plurality of training samples; the support vector machine classifies the training samples into high-risk samples and low-risk samples through corresponding hyperplanes, the constraint condition is that the distances between the low-risk samples and the hyperplanes and the distances between the high-risk samples and the hyperplanes meet a preset distance condition, and the solution target is that the sum of the distances between each training sample and the hyperplane is the largest.

Among them, the Support Vector Machines (SVMs) are a binary model, and the basic model of the SVM may be a classifier with the largest interval defined on the feature space. In other examples, an SVR (Support Vector Regression) Regression model (which may also be referred to as an SVC (Support Vector Classification) model) may also be used for training.

In this step, after a plurality of training samples are obtained, a support vector machine to be trained, constraint conditions for training the support vector machine, and a solution target may be obtained. Specifically, the solution objective (which may also be referred to as a script concept principle) of the support vector machine is to determine a hyperplane having a maximum interval from the divided training samples in the feature space corresponding to the training samples. In this embodiment, the two classification processes may be performed on the multiple training samples through the hyperplane of the support vector machine, the training samples are divided into high-risk samples and low-risk samples, and when the training samples are divided into the high-risk samples and the low-risk samples through the support vector machine, the distance between the high-risk samples and the low-risk samples and the hyperplane in the feature space should satisfy a preset distance condition, for example, the distance between each of the high-risk samples and the low-risk samples and the hyperplane is greater than a preset value.

In an alternative embodiment, in the feature space corresponding to the training sample, the hyperplane may be represented as:

ω ^T x+b＝0

in the above formula, ω = (w) ₁ ； ₁ ；…； _d ) Where d is a feature dimension (which may also be referred to as a feature number) of the feature space, the feature dimension is determined based on a feature dimension corresponding to the sample account feature. Where b is the offset, the hyperplane can be written as (ω, b). For any of a plurality of training examples, the distance between the training example and the hyperplane may be determined based on the example account features of the training example, e.g., for example, the distance r to the hyperplane (ω, b) may be expressed as:

from the above equation, it can be seen that the hyperplane that requires finding the "maximum separation" can be equivalent to finding the constraint parameters ω and b that maximize the sum of multiple r; accordingly, the basic model of the support vector machine can be as follows:

in the above formula, y _i The value of (1) is +1 or-1, and the risk label of the sample user account can indicate the risk level of the sample user account, in this example, the label of the sample user account can include high risk and low risk, the high risk label can be set to be +1, and the low risk label is set to be-1; and m is the number of samples of the training sample.

Under the preset constraint condition, the support vector machine to be trained can be used for dividing the training samples into high-risk samples and low-risk samples to obtain corresponding prediction results, and the prediction results can be the sum of the distances between the training samples and the hyperplane. And then according to the prediction result and the solution target, adjusting the model parameters corresponding to the support vector machine to obtain an updated hyperplane, continuously dividing a plurality of training samples by using the hyperplane, repeating the steps, and iterating for a plurality of times.

And S303, when the training end condition is met, taking the current support vector machine as a trained account risk identification model.

When it is detected that the training end condition is met, for example, when the iteration number reaches a preset number or the difference between the predicted value and the solution target is smaller than a preset threshold, the current support vector machine can be used as a trained account risk identification model. Specifically, for example, for a trained account risk identification model, account features corresponding to a user account may be input into the model, and a corresponding predicted value is obtained through calculation, and if the predicted value is greater than 0, it is determined that the account risk identification result is a low risk, and if the predicted value is less than 0, it is determined that the account risk identification result is a high risk.

In this embodiment, the support vector machine may be trained to obtain the account risk identification model, and compared with models such as a neural network model and deep learning, on one hand, the support vector machine is strong in interpretability and more reliable in model result, and on the other hand, the training mode and the operation speed are faster and faster, required training samples are far less than those of the neural network model and the deep model (the neural network model and the deep model are generally based on an ultra-large data set), the overfitting property is low, the generalization capability is strong, and therefore the account risk identification model capable of outputting a reliable risk identification result can be quickly obtained.

In one embodiment, as shown in fig. 4, S301, obtaining a plurality of training samples based on sample account characteristics corresponding to a plurality of sample user accounts handling historical services may include the following steps:

s401, acquiring initial account characteristics corresponding to a plurality of sample user accounts handling historical services and a risk label of each sample user account.

Specifically, the initial account characteristics corresponding to each of a plurality of sample user accounts transacted with historical services may be acquired, where the initial account characteristics may be data obtained by performing at least one of the following preprocessing on account data associated with the sample user accounts: screening, fusing and standardizing.

Moreover, a respective risk label may also be obtained for each sample user account, and the risk label may be used to indicate a risk level of the sample user account, which may include, for example, high risk and low risk.

S402, mapping the initial account characteristics by adopting a preset kernel function to obtain sample account characteristics corresponding to each sample user account; the feature dimension of the sample account features is greater than the feature dimension of the primary account features.

After sample account features corresponding to sample user accounts are obtained, a preset kernel function can be obtained, the preset kernel function is adopted to map the primary account features, the primary account features in the feature space with lower feature dimensions are mapped to account features in the feature space with higher feature dimensions, and the mapped account features are determined to be the sample account features.

The feature space may refer to each index feature based on a training sample, a set formed by a plurality of index features is referred to as a feature space, one feature is used as one dimension, for a data set containing two-dimensional features, a coordinate graph of the data set is as shown in fig. 5, a two-dimensional coordinate in the graph is a feature space of a sample, and a line (plane) can be found in the graph to divide two types of samples.

S403, obtaining a plurality of training samples based on the sample account features and the risk labels corresponding to each sample user account.

After the sample account characteristics corresponding to each sample user account are obtained, a plurality of training samples can be obtained by combining the risk labels, wherein the sample account characteristics of each sample user account and the risk labels can form one training sample.

Specifically, in reality, there are various sample user accounts, and there are cases where linear division by hyperplane is not possible (i.e., linear inseparable). In this regard, low-dimensional data may be mapped from the original feature space to a higher-dimensional feature space according to a mathematical law of relevance, such that the training samples are linearly separable within the higher-dimensional feature space.

The mapping method may specifically be a kernel function, where the kernel function implicitly defines a feature space called "regenerated kernel hilbert space", and the kernel function is a semi-positive kernel matrix, and for a certain semi-positive kernel matrix, there is a corresponding mapping space (i.e., feature space). Therefore, in practice, the kernel function can be used to map the training samples with low feature dimensions from the low-dimensional feature space to the high-dimensional feature space, so as to form linearly separable training samples.

Illustratively, the kernel function may be a gaussian kernel function (RBF) or a sigmoid kernel function. When using a kernel function, the model of the support vector machine can be expressed as:

wherein α = (α) ₁ ； ₂ ；…； _m ) Is a lagrange multiplier introduced to convert the problem into a dual problem, with κ (·) being the kernel function. For the model in the above equation, ω and b can be recalculated by solving the lagrange multiplier. Certainly, in some examples, in order to avoid an overfitting phenomenon of the sample user features obtained after mapping in the model training process, the model is optimized, a soft interval may be added to the hyperplane, and a training artifact that is in the soft interval but is misclassified is properly tolerated, then the solution objective may be adjusted to:

wherein, C is a constant and represents the size of the soft interval, when C is infinite, the full interval approaches to 0, and vice versa; l is a loss function, which may illustratively be any of the following: 0/1 loss function, hinge loss function, exponential loss function, logarithmic loss function.

Further, in the case of considering both the kernel function and the soft interval, the expression of the support vector machine can be as follows:

for a dual support vector machine model, the predicted value f (x) of the model may be:

after the predicted value and the solution target of the support vector machine are determined, the formula in the support vector machine can be solved based on the solution target

And the solution corresponding to the maximum value L (ω, b, α) is the training result of the model. Specifically, let:

by devitalizing α by L (ω, b, α), α can be solved _i If α is _i If the constraint in the above formula is not satisfied, alpha is determined _i And if the solution is not the optimal solution, sequentially substituting all training samples for calculation, taking the point of the obtained maximum result as an extreme value, determining the optimal alpha, calculating omega and b according to the optimal alpha, and obtaining the trained support vector machine.

In this embodiment, the initial account features of the sample user accounts are mapped to a higher-dimensional feature space through the kernel function to obtain the sample account features, so that the finally trained support vector machine can perform nonlinear division on the risk levels of different user accounts, and the accuracy of the risk identification result is effectively improved.

In one embodiment, the obtaining of the primary account characteristics corresponding to each of the plurality of sample user accounts transacting the historical service in S401 may include the following steps:

respectively acquiring external sample account data and internal sample account data of a plurality of sample user accounts handling historical services from an external data system and an internal data system; determining associated data items and non-associated data items in the external sample account data and the internal sample account data, and fusing data values of the external sample account data and the internal sample account data under the associated data items to obtain fused sample account data; and determining the initial account characteristics corresponding to each sample user account based on the sample account data fused with each sample user account and the sample account data under the non-associated data item.

In this embodiment, the external sample account data and the internal sample account data of each of a plurality of sample user accounts handling historical services may be acquired from the external data system and the internal data system, respectively, and then the external sample account data and the internal sample account data may be feature-fused.

As an example, the feature fusion may refer to feature fusion performed on the external sample account data and the internal sample account data by using a cross feature fusion model, and in the feature fusion process, the feature fusion may be performed on account data with a homogenization characteristic in the external sample account data and the internal sample account data, or, for a similar feature, a plurality of data items are introduced in the internal data system and the external data system, for example, for the feature X, different data values exist in the external sample account data and the internal sample account data.

In the step, when feature fusion is performed, associated data items and non-associated data items in the external sample account data and the internal sample account data may be determined, where the associated data items may refer to data items with associations between the external sample account data and the internal sample account data, and the associated data items may include the same data items or the same kind of data items; the unassociated data items may refer to data items that are independent of each other between the external sample account data and the internal sample account data.

After the associated data items are determined, the data values of the external sample account data and the internal sample account data under the associated data items may be obtained, and the data values of the external sample account data under the associated data items and the data values of the internal sample account data under the associated data items may be fused, for example, fusion may be performed in a manner of summing and averaging, so that the fused sample account data may be obtained.

After the fused sample account data is obtained, the initial account characteristics corresponding to each sample user account can be determined based on the fused sample account data of each sample user account and the sample account data under the non-associated data items.

In an optional embodiment, the fused sample account data and the sample account data under the non-associated data items may be preprocessed, including but not limited to encoding special data therein, and performing amplitude limiting or filtering on abnormal data, and the preprocessed data may improve the effect of the model.

In addition, the preprocessed account data can be standardized, and the model training speed and the training effect can be improved. The data standardization is to convert data, so that data values are more standard and reasonable, for example, the difference between upper and lower boundaries of the data between features is large, but the influence weight between the features is irrelevant to the upper and lower boundaries, and the standardized data can be obtained by calculating the relative value of the data in an interval, so that the convergence of a model can be accelerated. In one example, a standard deviation normalization method may be employed, with a specific calculation formula of

Wherein the content of the first and second substances,

for data after normalization, x is data before normalization,

and S is the standard deviation of the data item, wherein S is the data mean value of each sample under the data item. The mean value and the standard deviation of the method are subject to the training sample, and in the subsequent model use, the mean value and the standard deviation do not change with the introduction of evaluation data (such as a user account for determining a risk identification result).

In addition, in some embodiments, the external sample account data and the internal sample account data have various data items and data value types with various formats, and in order to facilitate the input to the model for processing, a preset data item therein may be encoded, for example, a gender type data item may be encoded by 01.

In the embodiment, on one hand, in the training process of the support vector machine, the external system data and the internal system data can be simultaneously utilized, and the account risk existing in the sample user account can be accurately identified by combining the characteristics of multiple aspects of the user account, so that the data value is effectively exerted; on the other hand, by carrying out feature fusion on the data values under the associated data items in the external sample account data and the internal sample account data, repeated feature dimensions can be reduced, independence between features can be improved, meanwhile, the problem of interference caused by different feature values of repeated features is avoided, and the model training speed and the model effect are improved.

In one embodiment, the step S202 of obtaining updated account characteristics of the user account based on the external account data may include the following steps:

acquiring original account data acquired in the user account service transaction process from an internal data system, and obtaining the internal account data of a user account based on the original account data; and fusing the external account data and the internal account data, and acquiring the updated account characteristics of the user account based on the fused account data.

In this embodiment, the primary account data collected in the user account transaction process may be obtained from the internal data system, where the primary account data may be data obtained based on the internal transaction process, and the primary account data may include basic data used to determine a label or score corresponding to the user account.

Further, the internal account data of the user account may be obtained based on the primary account data, for example, the primary account data may be used as the internal account data, or the primary account data may be filtered and the filtered primary account data may be used as the internal account data.

After the internal account data is obtained, the external account data and the internal account data can be fused, and the updated account characteristics of the user account are obtained based on the fused account data.

In this embodiment, on one hand, the internal account data corresponding to the user account may be obtained based on the original account data, so that direct use of the internal data score and the tag is avoided, and influence of the internal data score and the tag error value on account risk identification is reduced; on the other hand, by fusing the internal account data and the external account data and then acquiring the updated account characteristics, the account risk existing in the user account can be accurately identified by combining a user account service processing process and a plurality of characteristics in scenes irrelevant to service processing, the data value is effectively exerted, and the accuracy of an account risk identification result is improved.

In an embodiment, the fusing the external account data and the internal account data, and acquiring the updated account characteristics of the user account based on the fused account data may include the following steps:

determining associated data items and non-associated data items of the external account data and the internal account data based on respective data items of the external account data and the internal account data; fusing data values of the external account data and the internal account data under the associated data items to obtain account data updated by the associated data items; and acquiring the updated account characteristics of the user account based on the updated account data of the associated data item and the account data under the non-associated data item.

In this embodiment, after the external account data and the internal account data are acquired, data items corresponding to the external account data and the internal account data may be determined, and associated data items and non-associated data items between the data items of the external account data and the data items of the internal account data may be identified.

The associated data item may refer to a data item in which an association exists between the external account data and the internal account data, for example, one data item of the external account data may be a permanent address of the user account, and one data item of the internal account data may be a residential address of the user account, and since both data items may obtain a current location of the user account, the two data items may be determined as the associated data items; the associated data items may include the same data items, or may include homogeneous data items.

The unassociated data items may refer to data items that are independent of each other between the external sample account data and the internal sample account data.

After the associated data item is determined, the data values of the external account data and the internal account data under the associated data item may be obtained, and the data values of the external account data under the associated data item and the data values of the internal account data under the associated data item may be fused, for example, the fusion may be performed in a manner of summing and averaging, so that the account data after the associated data item is updated may be obtained.

After the account data updated by the associated data item is obtained, the updated account characteristics of the user account can be obtained based on the account data updated by the associated data item and the account data under the non-associated data item. Specifically, for example, each data item may be used as one feature, and a data value corresponding to the data item may be used as a feature value of the feature, so that account features updated in multiple dimensions of the user account may be obtained.

In an optional embodiment, the account data updated by the associated data item and the account data under the non-associated data item may be preprocessed, including but not limited to encoding special data therein, performing amplitude limiting or filtering on abnormal data, where the preprocessed data may improve the effect of the model, and may also be normalized, where the preprocessing and normalizing mode may be the same as the processing mode of the external sample account data or the internal sample account data, and reference may be made to the foregoing specifically, which is not described herein again.

In this embodiment, by performing feature fusion on data values under associated data items in the external account data and the internal account data, repeated feature dimensions can be reduced, and meanwhile, account risk identification can be performed by combining multiple account features inside and outside the user account, so that the reliability of an account risk identification result is improved.

In one embodiment, S201, acquiring external account data corresponding to a user account associated with a target service from an external data system, may include the following steps:

determining account attributes associated with the target service; and acquiring attribute information of the user account under the account attribute from an external data system as external account data corresponding to the user account.

In a specific implementation, account attributes associated with the target service may be determined, where the account attributes may include inherent attributes of the user account, and attribute information corresponding to the account attributes may be used to determine rating or basic data of a label of the user account. In some cases, the service party does not directly obtain attribute information of the user account under the preset account attribute in the process of processing the service.

In this step, the server may determine an account attribute associated with the target service, and request to acquire attribute information of the user account under the account attribute from the external data system, so that the attribute information may be used as external account data corresponding to the user account.

In this embodiment, the external account data may be obtained based on the attribute information of the user account under the preset account attribute acquired from the external data system, so that direct use of external data scores and tags is avoided, influence of the external data scores and the error values of the tags on account risk identification is reduced, and accuracy of the account risk identification result is improved.

In order to enable those skilled in the art to better understand the above steps, the following description is illustrative by an example, but it should be understood that the embodiments of the present application are not limited thereto.

In a specific implementation, the associated information between the user account and the service may be first established in the external data system, as shown in fig. 6, when the user triggers the transaction a through the corresponding user account, in response to the transaction request for the service a, the server may invoke the external data query service, and obtain the external account data from the external data system by using the data service provided by the external company, and at the same time, may obtain the internal account data through the data service of the server itself.

After the external account data and the internal account data are obtained, on one hand, the server side can continue to continue the business process corresponding to the business A based on the currently obtained account data, and on the other hand, the latest external account data and the latest internal account data which are currently obtained can be used for determining the account risk of the user account under the business A and simultaneously triggering the account risk update of the business B transacted with the user account. Specifically, an account risk identification model a and an account risk identification model B of the service a and the service B may be obtained, and external account data and internal account data may be input to the models to obtain risk identification results output by the account risk identification model a and the account risk identification model B.

When the risk identification result indicates that the risk exists, the service A and/or the service B can be informed to carry out risk control; (ii) a If the risk identification result is low risk, the current flow may be ended. Through the mode, although the service B is not aware when the service A inquires the external account data, the account risk identification can be timely carried out according to the number checking result of the service A, the effect that other services do not need to check numbers and the server side automatically carries out the tracking risk identification based on the latest data can be realized.

It should be understood that, although the steps in the flowcharts related to the embodiments as described above are sequentially displayed as indicated by arrows, the steps are not necessarily performed sequentially as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a part of the steps in the flowcharts related to the embodiments described above may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the execution order of the steps or stages is not necessarily sequential, but may be performed alternately or alternately with other steps or at least a part of the steps or stages in other steps.

Based on the same inventive concept, the embodiment of the application also provides an account risk identification device for realizing the account risk identification method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so specific limitations in one or more embodiments of the account risk identification device provided below can be referred to the limitations of the account risk identification method in the above, and details are not described here.

In one embodiment, as shown in fig. 7, there is provided an account risk identification apparatus including:

an external data obtaining module 701, configured to obtain, in response to a service transaction request for a target service from a user account, external account data corresponding to the user account associated with the target service from an external data system;

an account feature updating module 702, configured to obtain, based on the external account data, an account feature after the user account is updated;

a model obtaining module 703, configured to determine a historical service associated with the user account, and obtain a pre-trained account risk identification model corresponding to the historical service;

a risk identification result obtaining module 704, configured to input the updated account characteristics into the account risk identification model, and update a historical risk identification result of the user account under the historical service based on a risk identification result output by the account risk identification model.

In one embodiment, the apparatus further comprises:

the sample acquisition module is used for acquiring a plurality of training samples based on sample account characteristics corresponding to a plurality of sample user accounts handling the historical business;

the support vector machine training module is used for training a support vector machine to be trained by utilizing the plurality of training samples under a preset constraint condition and a solution target; the support vector machine classifies the training samples into high-risk samples and low-risk samples through corresponding hyperplanes, the constraint condition is that the distances between the low-risk samples and the hyperplanes and the distances between the high-risk samples and the hyperplanes meet a preset distance condition, and the solution target is that the sum of the distances between each training sample and the hyperplane is the largest;

and the risk identification model acquisition module is used for acquiring the trained account risk identification model when the training end condition is met.

In one embodiment, the sample acquisition module is configured to:

In one embodiment, the account feature update module 702 is configured to:

In one embodiment, the external data obtaining module 701 is configured to:

determining account attributes associated with the target service;

and acquiring attribute information of the user account under the account attribute from an external data system as external account data corresponding to the user account.

The modules in the account risk identification device can be wholly or partially implemented by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.

In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 8. The computer device includes a processor, a memory, an Input/Output interface (I/O for short), and a communication interface. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface is connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store account data. The input/output interface of the computer device is used for exchanging information between the processor and an external device. The communication interface of the computer device is used for connecting and communicating with an external terminal through a network. The computer program is executed by a processor to implement an account risk identification method.

Those skilled in the art will appreciate that the architecture shown in fig. 8 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is provided, comprising a memory having a computer program stored therein and a processor that when executing the computer program performs the steps of:

determining historical business related to the user account, and acquiring a pre-trained account risk identification model corresponding to the historical business;

In one embodiment, the steps in the other embodiments described above are also implemented when the computer program is executed by a processor.

In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:

In one embodiment, the computer program when executed by the processor also performs the steps in the other embodiments described above.

In one embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, performs the steps of:

In one embodiment, the computer program when executed by the processor also implements the steps of the other embodiments described above.

It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) referred to in the present application are information and data authorized by the user or sufficiently authorized by each party, and the collection, use and processing of the related data need to comply with the relevant laws and regulations and standards of the relevant country and region.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, database, or other medium used in the embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high-density embedded nonvolatile Memory, resistive Random Access Memory (ReRAM), magnetic Random Access Memory (MRAM), ferroelectric Random Access Memory (FRAM), phase Change Memory (PCM), graphene Memory, and the like. Volatile Memory can include Random Access Memory (RAM), external cache Memory, and the like. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others. The databases referred to in various embodiments provided herein may include at least one of relational and non-relational databases. The non-relational database may include, but is not limited to, a block chain based distributed database, and the like. The processors referred to in the various embodiments provided herein may be, without limitation, general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, or the like.

The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.

The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present application. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, and these are all within the scope of protection of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. An account risk identification method, the method comprising:

2. The method of claim 1, wherein the pre-trained account risk recognition model corresponding to the historical service is obtained by training through the following steps:

under the preset constraint condition and the solution target, training a support vector machine to be trained by utilizing the plurality of training samples; the support vector machine classifies the training samples into high-risk samples and low-risk samples through corresponding hyperplanes, the constraint condition is that the distances between the low-risk samples and the hyperplanes and between the high-risk samples and the hyperplanes meet a preset distance condition, and the solution target is that the sum of the distances between each training sample and the hyperplane is maximum;

and when the training end condition is met, taking the current support vector machine as a trained account risk identification model.

3. The method of claim 2, wherein obtaining a plurality of training samples based on sample account characteristics corresponding to a plurality of sample user accounts transacting the historical business comprises:

4. The method of claim 3, wherein the obtaining of the primary account characteristics corresponding to each of the plurality of sample user accounts transacting the historical business comprises:

respectively acquiring external sample account data and internal sample account data of a plurality of sample user accounts transacting the historical business from an external data system and an internal data system;

5. The method of claim 1, wherein obtaining updated account characteristics of the user account based on the external account data comprises:

6. The method according to claim 5, wherein the fusing the external account data and the internal account data and obtaining updated account features of the user account based on the fused account data comprises:

7. The method according to any one of claims 1-6, wherein the obtaining external account data corresponding to the user account associated with the target service from an external data system comprises:

determining account attributes associated with the target service;

8. An account risk identification apparatus, the apparatus comprising:

the external data acquisition module is used for responding to a service transaction request of a user account for a target service and acquiring external account data corresponding to the user account related to the target service from an external data system;

9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 7.

10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.

11. A computer program product comprising a computer program, characterized in that the computer program realizes the steps of the method of any one of claims 1 to 7 when executed by a processor.