CN114820164A - Credit card limit evaluation method, device, equipment and medium - Google Patents

Abstract

The disclosure provides a credit card limit evaluation method which can be applied to the technical fields of artificial intelligence and finance. The credit card limit evaluation method comprises the following steps: acquiring card using behavior characteristics of a user to be evaluated according to credit card transaction data of the user to be evaluated; inputting the card-using behavior characteristics of a user to be evaluated into a generator in a pre-trained generation countermeasure network, and outputting a node transaction data sequence, wherein the node transaction data sequence comprises a first node transaction data sequence of the user to be evaluated and a second node transaction data sequence of a constructed user; the pre-trained generated countermeasure network is obtained by training based on real sequence information, pseudo sequence information and hidden characteristics; and obtaining the credit card limit evaluation value of the user to be evaluated based on the node transaction data sequence. The disclosure also provides a credit card limit evaluation device, equipment, storage medium and program product.

Description

Credit card limit assessment method, device, equipment and medium

technical field

The present disclosure relates to the field of artificial intelligence and financial technology, and in particular, to a credit card limit evaluation method, apparatus, device, medium and program product.

Background technique

With the development and progress of the economy, the credit system is becoming more and more perfect, and credit cards have been accepted and used by more and more people, and a new credit-based consumption model has been formed. Its biggest feature is that after the user and the card issuer reach a valid agreement, the user is allowed to overdraft consumption in advance within a certain credit limit, and the overdraft amount can be repaid within a specified period. Although this model can effectively stimulate the consumption of users and promote the continuous improvement of the credit mechanism, it also brings certain financial risks to the card issuer.

SUMMARY OF THE INVENTION

In view of the above problems, the present disclosure provides a credit card limit evaluation method, apparatus, device, medium and program product.

According to a first aspect of the present disclosure, there is provided a credit card limit assessment method, including:

According to the credit card transaction data of the user to be evaluated, obtain the card-using behavior characteristics of the user to be evaluated;

Input the card-using behavior characteristics of the user to be evaluated into the generator in the pre-trained generative adversarial network, and output the node transaction data sequence, wherein the node transaction data sequence includes the first node transaction data sequence of the user to be evaluated and the constructed user's transaction data sequence. The second node transaction data sequence; the pre-trained generative adversarial network is trained based on real sequence information, pseudo-sequence information, and hidden features; the real sequence information includes the card-using behavior characteristics of real users in the transaction network and the relationship between adjacent real users. The pseudo-sequence information includes the card-using behavior characteristics of pseudo-users generated by the generator in the initial generative adversarial network and the relationship information between pseudo-adjacent users; the hidden features include the relationship information between non-adjacent real users in the transaction network ;as well as

Based on the node transaction data sequence, the credit card limit evaluation value of the user to be evaluated is obtained.

According to an embodiment of the present disclosure, based on the node transaction data sequence, obtaining the credit card limit evaluation value of the user to be evaluated includes:

Calculate the frequency of occurrence of the nodes of the constructed user adjacent to the node of the user to be evaluated in the node transaction data sequence;

The constructed user whose frequency exceeds the preset threshold is regarded as the target user;

Based on the credit card limit value corresponding to the target user, the credit card limit evaluation value of the user to be evaluated is obtained.

According to an embodiment of the present disclosure, the pre-trained generative adversarial network is trained based on pseudo-sequence information, real sequence information, and hidden features, including:

Based on the credit card transaction data samples of real users, construct a transaction network with service providers and real users as nodes;

Perform node sampling on the transaction network to obtain real sequence information;

Use the generator in the initial generative adversarial network to process the transaction network to generate pseudo-sequence information;

Use graph convolutional network to process transaction network to obtain hidden features;

Input the real sequence information, pseudo sequence information and hidden features into the discriminator in the initial generation adversarial network, and output the discriminant score;

Based on the discriminant scores, the parameters of the initial generative adversarial network are adjusted.

According to the embodiment of the present disclosure, performing node sampling on the transaction network to obtain real sequence information includes:

According to the transaction records between the service provider and the real user, assign weights to the relationship information between the service provider and the real user in the transaction network;

Combined with the weight of the relationship information between the service provider and the real user, the random walk method is used to filter the real user nodes in the transaction network to obtain the real sequence information.

According to an embodiment of the present disclosure, using a generator in an initial generative adversarial network to process a transaction network, generating pseudo-sequence information includes:

Input the preset initial value into the calculation unit of the generator in the initial generation confrontation network, and output the calculation vector;

Based on the calculation vector, pseudo-sequence information is obtained.

According to an embodiment of the present disclosure, using a graph convolutional network to process a transaction network, the hidden features obtained include:

Input the card-using behavior characteristics of real users and the relationship information between adjacent users into the graph convolutional network;

Perform aggregation operations on the hidden layer of the graph convolutional network to obtain initial hidden features;

Perform a pooling operation on the initial hidden features to obtain the pooled initial hidden features;

Perform preset rule operations on the pooled initial hidden features to output hidden features.

A second aspect of the present disclosure provides a credit card limit assessment device, comprising:

The acquiring module is used to acquire the card-using behavior characteristics of the user to be evaluated according to the credit card transaction data of the user to be evaluated;

The generation module is used to input the card-using behavior characteristics of the user to be evaluated into the generator in the pre-trained generative adversarial network, and output the node transaction data sequence, wherein the node transaction data sequence includes the first node transaction data sequence of the user to be evaluated and the second node transaction data sequence of the constructed user; the pre-trained generative adversarial network is trained based on real sequence information, pseudo-sequence information, and hidden features; the real sequence information includes the card-using behavior characteristics of real users in the transaction network and The relationship information between adjacent real users; the pseudo-sequence information includes the card-using behavior characteristics of the pseudo-users generated by the generator in the initial generative adversarial network and the relationship information between the pseudo-adjacent users; the hidden features include the non-adjacent real users in the transaction network. relationship information between users; and

The evaluation module is used to obtain the evaluation value of the credit card limit of the user to be evaluated based on the node transaction data sequence.

A third aspect of the present disclosure provides an electronic device, comprising: one or more processors; a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more programs When executed by the processor, one or more processors are caused to execute the above method for evaluating credit card limit.

A fourth aspect of the present disclosure also provides a computer-readable storage medium having executable instructions stored thereon, the instructions, when executed by a processor, cause the processor to execute the above-mentioned method for evaluating a credit card limit.

A fifth aspect of the present disclosure also provides a computer program product, including a computer program, which implements the above credit card limit evaluation method when the computer program is executed by a processor.

According to an embodiment of the present disclosure, by inputting the card-using behavior characteristics of the user to be evaluated into a generator in a pre-trained generative adversarial network, a node transaction data sequence is generated, and the credit card limit evaluation of the user to be evaluated is obtained according to the node transaction data sequence. It can timely evaluate the card usage limit suitable for the user to be evaluated, which not only improves user satisfaction, but also reduces the financial risk of the card issuer. Using the generator in the pre-trained generative adversarial network can avoid the adverse effect of some abnormal transaction data on the quota evaluation. At the same time, it also reduces the investment in manual evaluation by business personnel based on expert knowledge, and improves the work efficiency of the entire business process.

Description of drawings

The foregoing and other objects, features and advantages of the present disclosure will become more apparent from the following description of embodiments of the present disclosure with reference to the accompanying drawings, in which:

FIG. 1 schematically shows an application scenario diagram of a credit card limit evaluation method, apparatus, device, medium and program product according to an embodiment of the present disclosure;

FIG. 2 schematically shows a flowchart of a credit card limit evaluation method according to an embodiment of the present disclosure;

3 schematically shows a flow chart of a method for obtaining the credit card limit evaluation value of a user to be evaluated based on a node transaction data sequence according to an embodiment of the present disclosure;

4 schematically shows a flow chart of a method obtained by training a pre-trained generative adversarial network based on pseudo-sequence information, real sequence information, and hidden features according to an embodiment of the present disclosure;

FIG. 5 schematically shows a schematic diagram of generative adversarial network training according to an embodiment of the present disclosure;

FIG. 6 schematically shows a structural block diagram of a credit card limit assessment device according to an embodiment of the present disclosure; and

FIG. 7 schematically shows a block diagram of an electronic device suitable for implementing a credit card limit assessment method according to an embodiment of the present disclosure.

Detailed ways

Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood, however, that these descriptions are exemplary only, and are not intended to limit the scope of the present disclosure. In the following detailed description, for convenience of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the present disclosure. It will be apparent, however, that one or more embodiments may be practiced without these specific details. Also, in the following description, descriptions of well-known structures and techniques are omitted to avoid unnecessarily obscuring the concepts of the present disclosure.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the present disclosure. The terms "comprising", "comprising" and the like as used herein indicate the presence of stated features, steps, operations and/or components, but do not preclude the presence or addition of one or more other features, steps, operations or components.

All terms (including technical and scientific terms) used herein have the meaning as commonly understood by one of ordinary skill in the art, unless otherwise defined. It should be noted that terms used herein should be construed to have meanings consistent with the context of the present specification and should not be construed in an idealized or overly rigid manner.

Where expressions like "at least one of A, B, and C, etc.," are used, they should generally be interpreted in accordance with the meaning of the expression as commonly understood by those skilled in the art (eg, "has A, B, and C") At least one of the "systems" shall include, but not be limited to, systems with A alone, B alone, C alone, A and B, A and C, B and C, and/or A, B, C, etc. ).

In the technical solution of the present disclosure, the acquisition, storage and application of the user's personal information involved all comply with the relevant laws and regulations, take necessary confidentiality measures, and do not violate public order and good customs.

In the technical solutions of the embodiments of the present disclosure, the authorization or consent of the user is obtained before the user's personal information is obtained or collected.

In the process of implementing the embodiments of the present disclosure, it is found that the credit card limit assessment method based on expert knowledge mainly uses a manual method, requires a large amount of labor and relies on complex business experience, and has huge labor and time costs, which is not only disadvantageous Subsequent expansion of credit card business will also reduce customer satisfaction and experience with cards. Based on the user's behavioral characteristics and the corresponding credit card limit, the card usage pattern is learned through the machine learning algorithm, and then the credit card can be analyzed and predicted based on these characteristics and the learned pattern. This method assumes that users are independent of each other, which is inconsistent with the phenomenon that each user is related to each other and affects each other in real life. This will lead to a certain deviation in the algorithm model after training, leaving certain financial hidden dangers to the credit card limit evaluation. Incorporating the relationships between customers into the learning of the model, we hope to increase the prediction accuracy of the model. However, in the actual transaction network, there are often some transaction relationships generated by abnormal situations (such as cash out, stealing, fraud, etc.). The introduction of these transaction relationships will not only not improve the learning effect of the algorithm, but will damage the model to a certain extent. accuracy, resulting in inaccurate quota assessments.

The embodiments of the present disclosure provide a credit card limit evaluation method, which includes: acquiring the card usage behavior characteristics of the user to be evaluated according to the credit card transaction data of the user to be evaluated; inputting the card usage behavior characteristics of the user to be evaluated into a pre-trained generation The generator in the adversarial network outputs the node transaction data sequence, wherein the node transaction data sequence includes the first node transaction data sequence of the user to be evaluated and the second node transaction data sequence of the constructed user; the pre-trained generative adversarial network is based on The real sequence information, pseudo-sequence information, and hidden feature training are obtained; and based on the node transaction data sequence, the credit card limit evaluation value of the user to be evaluated is obtained.

FIG. 1 schematically shows an application scenario diagram of a credit card limit evaluation method, apparatus, device, medium and program product according to an embodiment of the present disclosure.

As shown in FIG. 1 , an application scenario 100 according to this embodiment may include terminal devices 101 , 102 , 103 , a network 104 , and a server 105 . The network 104 is a medium used to provide a communication link between the terminal devices 101 , 102 , 103 and the server 105 . The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.

The user can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages and the like. Various communication client applications may be installed on the terminal devices 101 , 102 and 103 , such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only examples).

The terminal devices 101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

The server 105 may be a server that provides various services, such as a background management server (just an example) that provides support for websites browsed by users using the terminal devices 101 , 102 , and 103 . The background management server can analyze and process the received user requests and other data, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

It should be noted that the credit card limit evaluation method provided by the embodiment of the present disclosure may generally be executed by the server 105 . Correspondingly, the credit card limit evaluation apparatus provided by the embodiment of the present disclosure may generally be set in the server 105 . The credit card limit evaluation method provided by the embodiment of the present disclosure may also be executed by a server or server cluster different from the server 105 and capable of communicating with the terminal devices 101 , 102 , 103 and/or the server 105 . Correspondingly, the credit card limit evaluation apparatus provided by the embodiment of the present disclosure may also be provided in a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101 , 102 , 103 and/or the server 105 .

The credit card limit evaluation method provided by the embodiment of the present disclosure may also be executed by the terminal devices 101 , 102 , and 103 . Correspondingly, the credit card limit evaluation apparatus provided by the embodiment of the present disclosure may also be generally set in the terminal devices 101 , 102 , and 103 . The credit card limit evaluation method provided by the embodiment of the present disclosure may also be executed by other terminals different from the terminal devices 101 , 102 , and 103 . Correspondingly, the credit card limit evaluation apparatus provided by the embodiment of the present disclosure may also be set in other terminals different from the terminal devices 101 , 102 , and 103 .

It should be understood that the numbers of terminal devices, networks and servers in FIG. 1 are merely illustrative. There can be any number of terminal devices, networks and servers according to implementation needs.

Based on the scenario described in FIG. 1 , the method for evaluating the credit card limit of the disclosed embodiment will be described in detail below with reference to FIGS. 2 to 5 .

FIG. 2 schematically shows a flowchart of a credit card limit evaluation method according to an embodiment of the present disclosure.

As shown in FIG. 2 , the credit card limit evaluation method 200 of this embodiment includes operations S201 to S203.

In operation S201, according to the credit card transaction data of the user to be evaluated, the card-using behavior characteristics of the user to be evaluated are acquired.

According to the embodiments of the present disclosure, the credit card transaction data of the user to be evaluated can be first obtained through data obtaining software or other data obtaining methods. The characteristics of the user's card usage behavior may include: consumption amount, card usage times, maximum consumption amount, minimum consumption amount, and the like.

It should be noted that, before obtaining the credit card transaction data of the user to be evaluated, the authorization of the user to be evaluated for the credit card transaction data is obtained, and after the authorization is obtained, the credit card transaction data of the user to be evaluated is obtained.

In operation S202, the card-using behavior characteristics of the user to be evaluated are input into the generator in the pre-trained generative adversarial network, and the node transaction data sequence is output.

The node transaction data sequence includes the first node transaction data sequence of the user to be evaluated and the second node transaction data sequence of the constructed user; the pre-trained generative adversarial network is trained based on real sequence information, pseudo-sequence information, and hidden features; The real sequence information includes the card-using behavior characteristics of real users in the transaction network and the relationship information between adjacent real users; the pseudo-sequence information includes the card-using behavior characteristics and pseudo-phases of the fake users generated by the generator in the initial generative adversarial network. Relationship information between neighboring users; hidden features include relationship information between non-adjacent real users in the transaction network.

According to an embodiment of the present disclosure, the generator in the pre-trained generative adversarial network can input the card-using behavior characteristics of the user to be evaluated, and then correspondingly generate the first node transaction data sequence of the user to be evaluated and the second node transaction data sequence of the constructed user. Node transaction data sequence. The constructed user is the user associated with the user to be evaluated generated by the generator in the pre-trained generative adversarial network.

For example, the consumption amount, the number of times of card use, the maximum consumption amount, and the minimum consumption amount can be input into the generator in the pre-trained generative adversarial network, and the first node transaction data of the user to be evaluated can be output. sequence and the second node transaction data sequence of the constructed user. The transaction data may include transaction amount, transaction party and service party information, and the like. The second node transaction data sequence of the constructed user can be generated by the generator in the pre-trained generative adversarial network according to the card-using behavior characteristics of the user to be evaluated.

According to the embodiments of the present disclosure, considering the influence of abnormal transaction data on the quota evaluation, the pseudo-sequence information generated by the generator and the real sequence information and hidden features in the transaction network are input into the discriminator to obtain a discriminant score; based on the discriminant score, adjust the The model parameters of the generative adversarial network are repeatedly trained to obtain a pre-trained generative adversarial network.

In operation S203, the credit card limit evaluation value of the user to be evaluated is obtained based on the node transaction data sequence.

According to the embodiment of the present disclosure, the second node transaction data sequence of the constructed user adjacent to the user to be evaluated can be selected from the node transaction data sequence; and then the credit card limit value corresponding to the constructed user's second node transaction data sequence can be selected. to obtain the credit card limit evaluation value of the user to be evaluated. It should be noted that the generator output in the pre-trained generative adversarial network is constructed with the current corresponding credit card limit value when the user is constructed.

For example, N constructed users adjacent to the user to be evaluated can be selected, where N is a positive integer; by calculating the average value of the credit card limit values corresponding to the N constructed users, the obtained average value is taken as the user to be evaluated. Credit card limit assessment value.

According to the embodiment of the present disclosure, by inputting the card-using behavior characteristics of the user to be evaluated into the generator in the pre-trained generative adversarial network, a node transaction data sequence is generated, and the credit card limit evaluation value of the user to be evaluated is obtained according to the node transaction data sequence. , which can timely evaluate the card usage limit suitable for the user to be evaluated, which can not only improve user satisfaction, but also reduce the financial risk of the card issuer. Using the generator in the pre-trained generative adversarial network can avoid the adverse effect of some abnormal transaction data on the quota evaluation. At the same time, it also reduces the investment in manual evaluation by business personnel based on expert knowledge, and improves the work efficiency of the entire business process.

FIG. 3 schematically shows a flowchart of a method for obtaining an evaluation value of a credit card limit of a user to be evaluated based on a node transaction data sequence according to an embodiment of the present disclosure.

As shown in FIG. 3 , the method 300 for obtaining the credit card limit evaluation value of the user to be evaluated based on the node transaction data sequence in this embodiment includes operations S301 to S303 .

In operation S301, the frequency of occurrence of the nodes of the constructed user adjacent to the node of the user to be evaluated in the node transaction data sequence is calculated.

In operation S302, a constructed user whose frequency exceeds a preset threshold is used as a target user.

According to an embodiment of the present disclosure, the preset threshold value may be a frequency value determined according to the accuracy of the amount that actually needs to be evaluated.

In operation S303, the credit card limit evaluation value of the user to be evaluated is obtained based on the credit card limit value corresponding to the target user.

According to the embodiment of the present disclosure, the average value of the credit card limit values corresponding to the target user can be calculated, and the average value can be used as the credit card limit evaluation value of the user to be evaluated.

For example, according to the calculated frequency of occurrence of the nodes of the constructed user adjacent to the node of the user to be evaluated, the constructed users can be sorted in descending order, and then the constructed users corresponding to the frequencies exceeding the preset threshold can be used as Target users. Calculate the average value of the credit card limit of these target users as the credit card limit evaluation value of the user to be evaluated.

According to the embodiment of the present disclosure, based on the pre-trained generative adversarial network, the back-inference idea is used to obtain the credit card limit evaluation value of the user to be evaluated according to the credit card limit value of the user associated with the user to be evaluated. It can timely evaluate a credit card limit suitable for credit card users, which not only improves customer satisfaction, but also reduces the financial risk of card issuers.

FIG. 4 schematically shows a flowchart of a method for training a pre-trained generative adversarial network based on pseudo-sequence information, real sequence information, and hidden features according to an embodiment of the present disclosure; FIG. 5 schematically shows an implementation according to the present disclosure. Schematic diagram of Generative Adversarial Network training for an example.

As shown in FIG. 4 , the method 400 of the pre-trained generative adversarial network training based on pseudo sequence information, real sequence information, and hidden features of this embodiment includes operations S401 to S406 .

In operation S401, based on the real user's credit card transaction data samples, a transaction network with the server and the real user as nodes is constructed.

According to the embodiment of the present disclosure, the entire transaction network can be regarded as a heterogeneous two-part network G={V,U,E,A} by taking the service party and the real user as nodes, and the set V={v _i |1 ≤i≤N} represents the node representing the real user; the set U={u _j |1≤j≤M} represents the node representing the server. If there is a card transaction between the real user vi and the service party u _j , there is an edge e _{ij ∈} E between these two nodes. The service party may be, for example, a transaction party when the user uses a credit card for transaction, that is, a merchant. As shown in Figure 5, a transaction network can be constructed with merchants and users as nodes.

According to the embodiment of the present disclosure, the transaction network can not only represent the relationship between the real user and the service party, but also can represent the feature vector a _i ∈ A for each real user node v _i ; for example, the basic feature of each real user can be Including age, occupation, place of residence, and spending power, etc., each element in the vector _ai can represent the quantified value of one of the basic features of each real user.

In operation S402, node sampling is performed on the transaction network to obtain real sequence information.

According to the embodiment of the present disclosure, sampling the nodes of the transaction network to obtain the real sequence information may include: according to the transaction records between the service party and the real user, assigning a weight to the relationship information between the service party and the real user in the transaction network; The weight of the relationship information with the real user uses the random walk method to filter the real user nodes in the transaction network to obtain the real sequence information. As shown in Figure 5, the real sequence can be obtained according to the transaction network, and the real sequence carries real sequence information.

For example, a weight w _ij can be assigned to the edge e _ij , and its calculation method can be shown in the following formula (1):

Among them, s _ij is equal to the accumulated card transaction amount between the real user vi and the service party _{u j .} That is to say, for a real user, the greater the accumulated transaction amount with a certain service party, the greater the weight of the edge e _ij .

It should be noted that since each real user often has transaction behaviors with multiple service parties, and there will be multiple repeated transactions, in order to better show the relationship between real users and service parties in each transaction, Here, a weight w _{ij is assigned to the edge e ij .}

The random walk method can start from a real user node and alternately jump to the server and the next real user node. The jump probability is proportional to the weight of the edge between the real user and the server node. After defining the sampling length T, filter out the server node, only keep the initial real user node and the remaining T-1 real user nodes, which together form a real sequence sample seq _i used by the training model, and repeat the above random walk steps for many times , and generate a real sequence set TSEQ containing several samples, that is, real sequence information. where T is a positive integer greater than 1. T can be reasonably set according to actual needs.

In operation S403, the transaction network is processed using the generator in the initial generative adversarial network to generate pseudo-sequence information.

According to an embodiment of the present disclosure, the generator in the initial generative adversarial network may be a long short-term memory network for generating pseudo-sequence information. A long short-term memory network includes multiple computational units. As shown in Figure 5, the pseudo-sequence can be obtained according to the generator in the initial generative adversarial network, and the pseudo-sequence carries pseudo-sequence information.

For example, a standard normal distribution and an all-zero vector in the transaction network can be used as initial values, respectively, and input into the hidden layer and input layer of the computing unit of the generator at the first moment, and the calculation at the current moment can be obtained. The output layer of the unit outputs vector values. The vector will be converted into a one-hot encoding when it is output, and the encoding value is the element number corresponding to 1, which can be regarded as the first generated pseudo node. Then, the one-hot encoding corresponding to the node is used as the input of the input layer of the computing unit at the next moment, and the pseudo node generated at the next moment can be obtained in the same way. By repeating the above steps T times, a pseudo-node sequence seq _t ∈ FSEQ of length T can be obtained.

Wherein, each computing unit can obtain the result of the output layer of the computing unit through the following steps (2)-(6).

f _t =σ(W _f ·CONCAT(o _t-1 ,h _t )+b _f ) (2)

i _t =σ(W _i ·CONCAT(o _t-1 ,h _t )+b _i ) (3)

o _t =σ(W _o ·CONCAT(o _t-1 ,h _t )+b _o )*tanh(C _t ) (6)

Among them, W _f ,W _i ,W _C ,W _o and b _f ,b _i ,b _C ,b _o are the mapping matrix and the bias vector respectively, which are used for linear change. h _t and o _t represent the input and output of the computing unit at time t, respectively. Both σ(·) and tanh(·) are activation functions, which increase the learning ability of the model. C _t represents the state value of the computing unit connecting two adjacent moments, and the long-term and short-term memory network realizes the selective forgetting and memory of the previous data through this state value.

It should be noted that the activation function can generally be an exponential function, which is used to perform a nonlinear transformation on the result.

In operation S404, a graph convolutional network is used to process the transaction network to obtain hidden features.

According to the embodiment of the present disclosure, three-step operations can be performed on the hidden layer of the graph convolutional network as shown in FIG. 5 , and the first aggregation operation can be performed, for example, according to the following formula (7):

代表当前隐藏层前一层中节点u的隐藏特征向量。特别地，当k＝1(第一个隐藏层)时，

等于节点u对应客户的用卡行为特征组成的向量。N(v)代表节点v在构造的客户-商家户交易网络中，客户v的二阶邻居客户节点(即和客户v有某个相同商家发生过交易行为的其它客户)。AVG表示对向量中各元素求平均值的操作。in,

Represents the hidden feature vector of node u in the previous layer of the current hidden layer. In particular, when k=1 (the first hidden layer),

It is equal to the vector composed of the card-using behavior characteristics of the customer corresponding to the node u. N(v) represents the second-order neighbor customer node of customer v in the customer-merchant merchant transaction network constructed by node v (ie, other customers who have transaction behavior with the same merchant as customer v). AVG represents the operation of averaging the elements in a vector.

The second pooling operation, for example, can be performed according to the following formula (8):

向量；b表示

向量。Among them, σ( ) represents the nonlinear activation function, which is used to increase the learning ability of the graph convolution network. Generally, the ReLu function can be selected. W ^k represents a linear mapping matrix for dimensionality reduction. CONCAT(a,b) means to concatenate the two vectors a and b. Among them, a means

vector; b means

vector.

Thirdly, for each node, the pooled initial hidden feature vector calculated in the current layer is subjected to the regularization operation of the following formula (9) as the final output to ensure the robustness of the model:

where ||·|| represents the two-norm of the vector.

It should be noted that the card-using behavior characteristics of each user and the relationship information between adjacent users can be input into the above graph convolutional network, and finally the hidden feature vector of each node is obtained.

In operation S405, the real sequence information, pseudo sequence information and hidden features are input into the discriminator in the initial generation adversarial network, and a discriminant score is output.

According to an embodiment of the present disclosure, the discriminator may include a long short-term memory network, and may output a discriminant score for the node sequence at the last computing unit.

In operation S406, based on the discriminant score, parameters of the initial generative adversarial network are adjusted.

According to the embodiment of the present disclosure, the discriminant score can be brought into the loss function to calculate the loss value, and by continuously adjusting the parameters of the initial generative adversarial network, the training is converged, and a pre-trained generative adversarial network is obtained.

For example, the loss function of the generative adversarial network model can be expressed as the following equation (10):

和

分别表示针对真实序列信息以及伪序列信息的生成对抗网络的参数。Among them, f _θ (seq _i ) represents the discrimination score of the discriminator output of the adversarial network generated for real sequence information; f _θ (seq _t ) represents the discrimination score of the discriminator output of the adversarial network generated for pseudo-sequence information;

and

represent the parameters of the generative adversarial network for real sequence information and pseudo-sequence information, respectively.

In order to train a generative adversarial network for credit card limit evaluation, the score should be as low as possible for pseudo-sequence information; conversely, it should be as high as possible for real-sequence information. The following optimization objective can be used as shown in equation (11), given the learning rate, and the Adam optimization method can be used to learn the optimal parameters.

min _G max _D V(D,G) (11)

Among them, D represents the discriminator and G represents the generator.

According to the embodiments of the present disclosure, in addition to considering the user's card-using behavior feature information, the learning of user correlation data is added to the generative adversarial network by acquiring hidden features and relationship information between adjacent users. When evaluating the credit card limit, a more accurate and reasonable evaluation result can be obtained. It can also be well applied to the complex credit card transaction network, avoiding the adverse effects of some abnormal transaction data on the limit evaluation, making the generative adversarial network used for credit card limit evaluation more robust and credible.

According to an embodiment of the present disclosure, using a generator in an initial generative adversarial network to process a transaction network, and generating pseudo-sequence information may include: inputting a preset initial value into a calculation unit of the generator in the initial generative adversarial network, and outputting a calculation vector; Calculate the vector to obtain pseudo-sequence information. The preset initial value may be a value that obeys a preset rule in the trading network; the preset rule may be, for example, obeying a standard normal distribution or an all-zero vector.

According to the embodiments of the present disclosure, considering the influence of abnormal transaction data on the limit evaluation, using the pseudo-sequence information generated by the generator to participate in the training of the generative adversarial network is conducive to the accurate evaluation of the credit card limit value of the trained generative adversarial network.

According to an embodiment of the present disclosure, using a graph convolutional network to process a transaction network to obtain hidden features may include: inputting card-using behavior characteristics of real users and relationship information between adjacent users into the graph convolutional network; The hidden layer performs the aggregation operation to obtain the initial hidden features; performs the pooling operation on the initial hidden features to obtain the pooled initial hidden features; performs the preset rule operation on the pooled initial hidden features to output the hidden features.

According to the embodiments of the present disclosure, the learning of user correlation data is added to the generative adversarial network by acquiring hidden features, and a more accurate and reasonable evaluation result can be obtained when evaluating the credit card limit.

Based on the above credit card limit evaluation method, the present disclosure also provides a credit card limit evaluation device. The device will be described in detail below with reference to FIG. 6 .

FIG. 6 schematically shows a structural block diagram of a credit card limit evaluation apparatus according to an embodiment of the present disclosure.

As shown in FIG. 6 , the apparatus 600 for evaluating credit card limit in this embodiment includes an acquiring module 610 , a generating module 620 and an evaluating module 630 .

The obtaining module 610 is configured to obtain the card-using behavior characteristics of the user to be evaluated according to the credit card transaction data of the user to be evaluated. In an embodiment, the obtaining module 610 may be configured to perform the operation S201 described above, which will not be repeated here.

The generating module 620 is configured to input the card-using behavior characteristics of the user to be evaluated into the generator in the pre-trained generative adversarial network, and output the node transaction data sequence, wherein the node transaction data sequence includes the first node transaction data sequence of the user to be evaluated. and the second node transaction data sequence of the constructed user; the pre-trained generative adversarial network is trained based on real sequence information, pseudo-sequence information, and hidden features; the real sequence information includes the card-using behavior characteristics of real users in the transaction network and The relationship information between adjacent real users; the pseudo-sequence information includes the card-using behavior characteristics of the pseudo-users generated by the generator in the initial generative adversarial network and the relationship information between the pseudo-adjacent users; the hidden features include the non-adjacent real users in the transaction network. Relationship information between users. In one embodiment, the generating module 620 may be configured to perform the operation S202 described above, which will not be repeated here.

The evaluation module 630 is configured to obtain the evaluation value of the credit card limit of the user to be evaluated based on the node transaction data sequence. In an embodiment, the evaluation module 630 may be configured to perform the operation S203 described above, which will not be repeated here.

According to an embodiment of the present disclosure, any number of modules in the acquisition module 610 , the generation module 620 and the evaluation module 630 may be combined into one module for implementation, or any one of the modules may be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules may be combined with at least part of the functionality of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the acquisition module 610, the generation module 620, and the evaluation module 630 may be implemented, at least in part, 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 any other reasonable means of integrating or packaging a circuit can be implemented in hardware or firmware, or in software, hardware, and firmware. Any one of these implementations or an appropriate combination of any of them is implemented. Alternatively, at least one of the acquisition module 610, the generation module 620, and the evaluation module 630 may be implemented, at least in part, as computer program modules that, when executed, may perform corresponding functions.

FIG. 7 schematically shows a block diagram of an electronic device suitable for implementing a credit card limit assessment method according to an embodiment of the present disclosure.

As shown in FIG. 7 , an electronic device 700 according to an embodiment of the present disclosure includes a processor 701 that can be loaded into a random access memory (RAM) 703 according to a program stored in a read only memory (ROM) 702 or from a storage part 708 program to perform various appropriate actions and processes. The processor 701 may include, for example, a general-purpose microprocessor (eg, a CPU), an instruction set processor and/or a related chipset, and/or a special-purpose microprocessor (eg, an application-specific integrated circuit (ASIC)), and the like. The processor 701 may also include on-board memory for caching purposes. The processor 701 may include a single processing unit or multiple processing units for performing different actions of the method flow according to the embodiments of the present disclosure.

In the RAM 703, various programs and data necessary for the operation of the electronic device 700 are stored. The processor 701 , the ROM 702 and the RAM 703 are connected to each other through a bus 704 . The processor 701 performs various operations of the method flow according to the embodiment of the present disclosure by executing the programs in the ROM 702 and/or the RAM 703 . Note that the program may also be stored in one or more memories other than the ROM 702 and the RAM 703 . The processor 701 may also perform various operations of the method flow according to the embodiments of the present disclosure by executing programs stored in the one or more memories.

According to an embodiment of the present disclosure, the electronic device 700 may also include an input/output (I/O) interface 705 that is also connected to the bus 704 . The electronic device 700 may also include one or more of the following components connected to the I/O interface 705: an input portion 706 including a keyboard, mouse, etc.; including components such as a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and An output section 707 for speakers and the like; a storage section 708 including a hard disk and the like; and a communication section 709 including a network interface card such as a LAN card, a modem, and the like. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I/O interface 705 as needed. A removable medium 711, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is mounted on the drive 710 as needed so that a computer program read therefrom is installed into the storage section 708 as needed.

The present disclosure also provides a computer-readable storage medium. The computer-readable storage medium may be included in the device/apparatus/system described in the above embodiments; it may also exist alone without being assembled into the device/system. device/system. The above-mentioned computer-readable storage medium carries one or more programs, and when the above-mentioned one or more programs are executed, implement the method according to the embodiment of the present disclosure.

According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, such as, but not limited to, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM) , erasable programmable read only memory (EPROM or flash memory), portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. In this disclosure, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of the present disclosure, a computer-readable storage medium may include one or more memories other than ROM 702 and/or RAM 703 and/or ROM 702 and RAM 703 described above.

Embodiments of the present disclosure also include a computer program product comprising a computer program containing program code for performing the method illustrated in the flowchart. When the computer program product runs in the computer system, the program code is used to make the computer system implement the methods provided by the embodiments of the present disclosure.

When the computer program is executed by the processor 701, the above-described functions defined in the system/apparatus of the embodiments of the present disclosure are performed. According to embodiments of the present disclosure, the systems, apparatuses, modules, units, etc. described above may be implemented by computer program modules.

In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device, a magnetic storage device, or the like. In another embodiment, the computer program may also be transmitted, distributed in the form of a signal over a network medium, and downloaded and installed through the communication section 709, and/or installed from a removable medium 711. The program code embodied by the computer program may be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the foregoing.

In such an embodiment, the computer program may be downloaded and installed from the network via the communication portion 709 and/or installed from the removable medium 711 . When the computer program is executed by the processor 701, the above-described functions defined in the system of the embodiment of the present disclosure are performed. According to embodiments of the present disclosure, the above-described systems, apparatuses, apparatuses, modules, units, etc. can be implemented by computer program modules.

According to the embodiments of the present disclosure, the program code for executing the computer program provided by the embodiments of the present disclosure may be written in any combination of one or more programming languages. programming language, and/or assembly/machine language to implement these computational programs. Programming languages include, but are not limited to, languages such as Java, C++, python, "C" or similar programming languages. The program code may execute entirely on the user computing device, partly on the user device, partly on a remote computing device, or entirely on the remote computing device or server. Where remote computing devices are involved, the remote computing devices may be connected to the user computing device over any kind of network, including a local area network (LAN) or wide area network (WAN), or may be connected to an external computing device (eg, using an Internet service provider business via an Internet connection).

The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code that contains one or more functions for implementing the specified logical function(s) executable instructions. It should also be noted that, in some alternative implementations, the functions noted in the blocks may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It is also noted that each block of the block diagrams or flowchart illustrations, and combinations of blocks in the block diagrams or flowchart illustrations, can be implemented in special purpose hardware-based systems that perform the specified functions or operations, or can be implemented using A combination of dedicated hardware and computer instructions is implemented.

Those skilled in the art will appreciate that various combinations and/or combinations of features recited in various embodiments and/or claims of the present disclosure are possible, even if such combinations or combinations are not expressly recited in the present disclosure. In particular, various combinations and/or combinations of the features recited in the various embodiments of the present disclosure and/or in the claims may be made without departing from the spirit and teachings of the present disclosure. All such combinations and/or combinations fall within the scope of this disclosure.

Embodiments of the present disclosure have been described above. However, these examples are for illustrative purposes only, and are not intended to limit the scope of the present disclosure. Although the various embodiments are described above separately, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. The scope of the present disclosure is defined by the appended claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art can make various substitutions and modifications, and these substitutions and modifications should all fall within the scope of the present disclosure.

Claims (10)

1. A credit card limit evaluation method comprises the following steps:

acquiring card using behavior characteristics of a user to be evaluated according to credit card transaction data of the user to be evaluated;

inputting the card-using behavior characteristics of the user to be evaluated into a generator in a pre-trained generation countermeasure network, and outputting a node transaction data sequence, wherein the node transaction data sequence comprises a first node transaction data sequence of the user to be evaluated and a second node transaction data sequence of a constructed user; the pre-trained generated confrontation network is obtained based on real sequence information, pseudo sequence information and hidden feature training; the real sequence information comprises card using behavior characteristics of real users in the transaction network and relationship information between adjacent real users; the pseudo sequence information comprises card-using behavior characteristics of pseudo users generated by a generator in an initially generated countermeasure network and relationship information between pseudo adjacent users; the hidden features comprise relationship information between non-adjacent real users in the transaction network; and

and obtaining the credit card limit evaluation value of the user to be evaluated based on the node transaction data sequence.

2. The method as claimed in claim 1, wherein the obtaining an evaluation value of credit card amount of the user to be evaluated based on the node transaction data sequence comprises:

calculating the frequency of appearance of the nodes of the constructed user adjacent to the nodes of the user to be evaluated in the node transaction data sequence;

taking the constructed users with the frequency exceeding a preset threshold value as target users;

and obtaining the credit card limit evaluation value of the user to be evaluated based on the credit card limit value corresponding to the target user.

3. The method of claim 1, wherein the pre-trained generative confrontation network is trained based on pseudo sequence information, real sequence information, hidden features comprising:

constructing the transaction network with a service party and the real user as nodes based on credit card transaction data samples of the real user;

carrying out node sampling on the transaction network to obtain the real sequence information;

processing the transaction network with a generator in the initially generated counterpoise network, generating the pseudo sequence information;

processing the transaction network by using a graph convolution network to obtain the hidden features;

inputting the real sequence information, the pseudo sequence information and the hidden feature into a discriminator in the initially generated countermeasure network, and outputting a discrimination score;

adjusting parameters of the initially generated countermeasure network based on a discriminant score.

4. The method of claim 3, wherein the sampling nodes of the transaction network to obtain the true sequence information comprises:

according to the transaction records of the service party and the real user, assigning a weight value to the relationship information of the service party and the real user in the transaction network;

and screening real user nodes in the transaction network by using a random walk method in combination with the weight of the relation information between the service party and the real user to obtain the real sequence information.

5. The method of claim 3, wherein said processing the trading network with a generator in the initial generation countermeasure network, generating the pseudo sequence information comprises:

inputting a preset initial value into a calculation unit of a generator in the initial generation countermeasure network, and outputting a calculation vector;

and acquiring the pseudo sequence information based on the calculation vector.

6. The method of claim 3, wherein said processing said transaction network using a graph convolution network to obtain said hidden features comprises:

inputting the card using behavior characteristics of the real user and the relationship information between the adjacent users into the graph convolution network;

performing aggregation operation on a hidden layer of the graph convolution network to obtain an initial hidden feature;

performing pooling operation on the initial hidden features to obtain pooled initial hidden features;

and carrying out preset rule operation on the pooled initial hidden features, and outputting the hidden features.

7. A credit card limit evaluation device includes:

the acquisition module is used for acquiring card using behavior characteristics of the user to be evaluated according to credit card transaction data of the user to be evaluated;

the generation module is used for inputting the card-using behavior characteristics of the user to be evaluated into a generator in a pre-trained generation countermeasure network and outputting a node transaction data sequence, wherein the node transaction data sequence comprises a first node transaction data sequence of the user to be evaluated and a second node transaction data sequence of a constructed user; the pre-trained generated confrontation network is obtained based on real sequence information, pseudo sequence information and hidden feature training; the real sequence information comprises card using behavior characteristics of real users in the transaction network and relationship information between adjacent real users; the pseudo sequence information comprises card-using behavior characteristics of pseudo users generated by a generator in an initially generated countermeasure network and relationship information between pseudo adjacent users; the hidden features comprise relationship information between non-adjacent real users in the transaction network; and

and the evaluation module is used for obtaining the credit card limit evaluation value of the user to be evaluated based on the node transaction data sequence.

8. An electronic device, comprising:

one or more processors;

a storage device for storing one or more programs,

wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the method of any of claims 1-6.

9. A computer readable storage medium having stored thereon executable instructions which, when executed by a processor, cause the processor to perform the method of any one of claims 1 to 6.

10. A computer program product comprising a computer program which, when executed by a processor, implements a method according to any one of claims 1 to 6.

Images