Risk information generation method and device, electronic equipment and computer readable medium

Abstract

The embodiment of the invention discloses a risk information generation method and device, electronic equipment and a computer readable medium. A specific embodiment of the method comprises the steps of obtaining real-time order information and real-time order value conversion data; performing data processing on the real-time order value conversion data to obtain a data processing vector; in response to determining that a target user representation vector exists in a pre-stored user representation vector set, vector splicing is carried out on the target user representation vector and the data processing vector to obtain a spliced vector, a user corresponding to the target user representation vector and an order user corresponding to the real-time order information are the same user, and the spliced vector is obtained; the user representation vector set is generated offline based on a pre-trained user representation vector generation model; and inputting the splicing vector into a pre-trained risk information real-time generation model to output risk information for the real-time order information. The implementation mode is related to artificial intelligence, and the risk information for the real-time order information can be efficiently generated in real time.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of computer technology, and in particular to a risk information generation method, device, electronic device, and computer-readable medium. Background Art

[0002] At present, with the development of e-commerce platforms, various value conversion service products have attracted the attention of many young people. However, due to the low threshold and low cost of value conversion service products, there are also greater risks. For the generation of risk information, the usual method is to input historical data sets into deep learning neural networks to output risk information.

[0003] However, the inventors have found that when the above method is used to generate risk information, the following technical problems often occur:

[0004] Deep learning neural networks are often complex and involve a large amount of computation, which makes it impossible for the network to make real-time predictions of risk information in an online real-time risk control environment.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the inventive concept and therefore it may contain information that does not form the prior art that is already known in this country to a person of ordinary skill in the art. Summary of the invention

[0006] The content of this disclosure is used to introduce concepts in a brief form, which will be described in detail in the detailed implementation section below. The content of this disclosure is not intended to identify the key features or essential features of the technical solution claimed for protection, nor is it intended to limit the scope of the technical solution claimed for protection.

[0007] Some embodiments of the present disclosure propose risk information generation methods, devices, electronic devices, and computer-readable media to solve the technical problems mentioned in the above background technology section.

[0008] In a first aspect, some embodiments of the present disclosure provide a risk information generation method, including: obtaining real-time order information and corresponding real-time order value transformation data; performing data processing on the real-time order value transformation data to obtain a data processing vector; in response to determining that a target user representation vector exists in a pre-stored user representation vector set, performing vector splicing on the target user representation vector and the data processing vector to obtain a spliced ​​vector, wherein the user corresponding to the target user representation vector is the same user as the order user corresponding to the real-time order information, and the user representation vector set is generated offline based on a pre-trained user representation vector generation model; and inputting the spliced ​​vector into a pre-trained risk information real-time generation model to output risk information for the real-time order information.

[0009] Optionally, the method further includes: generating an order user representation vector for the order user using the user representation vector generation model; and updating a target user representation vector in the user representation vector set using the order user representation vector.

[0010] Optionally, the method further includes: in response to determining that the target user representation vector does not exist in the user representation vector set, generating an order user representation vector for the order user using the user representation vector generation model; and adding the order user representation vector to the user representation vector set.

[0011] Optionally, the user representation vector generation model is trained by the following steps: obtaining a first historical order value transformation data set for a historical order information set; constructing an object relationship structure network for the historical order information set based on a predetermined preset object set and the first historical order value transformation data set, wherein the object relationship structure network includes: an order user node; generating a user association graph based on the object relationship structure network, wherein the user association graph includes: a central node with the order user node as the center of the graph and at least one associated node adjacent to the central node; performing unsupervised training on the initial user representation vector generation model based on the user association graph to obtain the user representation vector generation model.

[0012] Optionally, the above-mentioned object relationship structure network for the above-mentioned historical order information set is constructed according to a predetermined preset object set and the above-mentioned first historical order value transformation data set, including: constructing an initial object relationship structure network according to the object association relationship between each preset object in the above-mentioned preset object set; extracting a historical order user feature data set from the above-mentioned first historical order value transformation data set; adding each feature data included in the above-mentioned historical order user feature data set as an associated node of the corresponding order user node to the initial object relationship structure network, and obtaining the object relationship structure network after adding the feature data as the above-mentioned object relationship structure network.

[0013] Optionally, the above-mentioned unsupervised training of the initial user representation vector generation model based on the above-mentioned user association graph set to obtain the above-mentioned user representation vector generation model includes: dividing the above-mentioned user association graph set to obtain a first user association graph subset and a second user association graph subset; constructing an unsupervised comparative learning loss function as the loss function of the initial user representation vector generation model; determining the above-mentioned first user association graph subset as a training sample set, and determining the above-mentioned second user association graph subset as a verification sample set; according to the above-mentioned training sample set and the above-mentioned verification sample set, according to the above-mentioned unsupervised comparative learning loss function, performing model unsupervised training on the initial user representation vector generation model to obtain the above-mentioned user representation vector generation model.

[0014] Optionally, the above-mentioned real-time generation model of risk information is trained through the following steps: obtaining a pre-stored historical user representation vector set and a corresponding second historical order value transformation data set; performing data processing on each second historical order value transformation data in the above-mentioned second historical order value transformation data set to generate a historical data processing vector, and obtain a historical data processing vector set; performing corresponding vector splicing on the above-mentioned historical user representation vector set and the above-mentioned historical data processing vector set to obtain a spliced ​​vector set; determining the risk label set corresponding to the above-mentioned historical user representation vector set; inputting each spliced ​​vector in the above-mentioned spliced ​​vector set into the initial risk information real-time generation model to generate predicted risk information, and obtain a predicted risk information set; based on the above-mentioned predicted risk information set and the above-mentioned risk label set, the above-mentioned initial risk information real-time generation model is supervised trained to obtain a risk information real-time generation model.

[0015] Optionally, the method further includes: performing risk control operations on the order user based on the risk information.

[0016] In a second aspect, some embodiments of the present disclosure provide a risk information generating device, including: an acquisition unit, configured to acquire real-time order information and corresponding real-time order value transformation data; a data processing unit, configured to perform data processing on the above-mentioned real-time order value transformation data to obtain a data processing vector; a vector splicing unit, configured to, in response to determining that a target user representation vector exists in a pre-stored user representation vector set, perform vector splicing on the above-mentioned target user representation vector and the above-mentioned data processing vector to obtain a splicing vector, wherein the user corresponding to the above-mentioned target user representation vector is the same user as the order user corresponding to the above-mentioned real-time order information, and the above-mentioned user representation vector set is generated offline based on a pre-trained user representation vector generation model; an input unit, configured to input the above-mentioned splicing vector into a pre-trained risk information real-time generation model to output risk information for the above-mentioned real-time order information.

[0017] Optionally, the apparatus further comprises: utilizing the user representation vector generation model to generate an order user representation vector for the order user; and updating a target user representation vector in the user representation vector set through the order user representation vector.

[0018] Optionally, the device further includes: in response to determining that the target user representation vector does not exist in the user representation vector set, generating an order user representation vector for the order user using the user representation vector generation model; and adding the order user representation vector to the user representation vector set.

[0019] Optionally, the above-mentioned device also includes: executing risk control operations for the above-mentioned order user based on the above-mentioned risk information.

[0020] In a third aspect, some embodiments of the present disclosure provide an electronic device comprising: one or more processors; a storage device on which one or more programs are stored, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation manner in the first aspect.

[0021] In a fourth aspect, some embodiments of the present disclosure provide a computer-readable medium having a computer program stored thereon, wherein when the program is executed by a processor, the method described in any implementation manner in the first aspect is implemented.

[0022] In a fifth aspect, some embodiments of the present disclosure provide a computer program product, including a computer program, which implements the method described in any implementation manner in the above-mentioned first aspect when executed by a processor.

[0023] The above-mentioned embodiments of the present disclosure have the following beneficial effects: the risk information generation method of some embodiments of the present disclosure can generate risk information for real-time order information efficiently and in real time. Specifically, the reason why risk information cannot be predicted in real time is that deep learning neural networks are often complex and have a large amount of calculation, which makes the network unable to perform real-time prediction of risk information in an online real-time risk control environment. Based on this, the risk information generation method of some embodiments of the present disclosure first obtains real-time order information and corresponding real-time order value transformation data as input source data of the risk information real-time generation model for subsequent risk information generation. Then, the real-time order value transformation data is processed to obtain a data processing vector. The real-time order value transformation data is converted into a vector form to facilitate feature learning of the subsequent risk information real-time generation model. Then, in response to determining that the pre-stored user representation vector set has a target user representation vector, the target user representation vector and the data processing vector are vector spliced ​​to obtain a spliced ​​vector. Among them, the user corresponding to the target user representation vector is the same user as the order user corresponding to the real-time order information, and the user representation vector set is generated offline based on the pre-trained user representation vector generation model. Here, the user representation vector set is pre-generated offline through the user representation vector generation model, which reduces the amount of calculation for generating user representation vectors in the real-time risk control prediction environment, and avoids the situation where the network cannot perform real-time prediction of risk information in the online real-time risk control environment due to the complexity of the online deep learning neural network and the large amount of calculation. By determining the target user representation vector corresponding to the real-time order information from the pre-stored user representation vector set, the operation is simple and does not require more complex calculations in the real-time risk control environment, which improves the prediction speed of risk information. In addition, the target user representation vector and the above-mentioned data processing vector are vector spliced ​​to facilitate subsequent input into the real-time prediction model of risk information. Finally, the above-mentioned spliced ​​vector is input into the pre-trained real-time generation model of risk information to output risk information for the above-mentioned real-time order information. In summary, the user representation vector set generated offline by the user representation vector generation model reduces the amount of calculation for generating the user representation vector corresponding to the real-time order information in the real-time risk control environment, reduces the prediction review of the online real-time generation model of risk information, and can effectively realize the real-time prediction of risk information. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that components and elements are not necessarily drawn to scale.

[0025] Figure 1 is a schematic diagram of an application scenario of the risk information generation method according to some embodiments of the present disclosure;

[0026] Figure 2 is a flow chart of some embodiments of the risk information generation method according to the present disclosure;

[0027] Figure 3 is a schematic diagram of an object relationship structure network in some embodiments of the risk information generation method according to the present disclosure;

[0028] Figure 4 are flow charts of other embodiments of the risk information generating method according to the present disclosure;

[0029] Figure 5 is a schematic diagram of the structure of some embodiments of the risk information generating device according to the present disclosure;

[0030] Figure 6 It is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION

[0031] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0032] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present disclosure can be combined with each other.

[0033] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0034] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0035] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0036] With regard to the collection, storage, and use of order information (such as real-time order information, real-time order value transformation data, and historical order information sets) involved in this disclosure, before performing corresponding operations, the relevant organizations or individuals shall fulfill obligations including conducting information security impact assessments, fulfilling the obligation to inform the information subject, and obtaining the authorization and consent of the information subject in advance.

[0037] The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0038] Figure 1 It is a schematic diagram of an application scenario of the risk information generation method according to some embodiments of the present disclosure.

[0039] exist Figure 1 In the application scenario, first, the electronic device 101 can obtain the real-time order information 102 and the corresponding real-time order value transformation data 103. Then, the electronic device 101 can perform data processing on the real-time order value transformation data 103 to obtain a data processing vector 104. Then, in response to determining that the pre-stored user representation vector set 105 has a target user representation vector 106, the target user representation vector 106 and the data processing vector 104 are vector spliced ​​to obtain a spliced ​​vector 108. Among them, the user corresponding to the target user representation vector 105 is the same user as the order user corresponding to the real-time order information 102. The user representation vector set 105 is generated offline based on a pre-trained user representation vector generation model 107. In this application scenario, the user corresponding to the target user representation vector 105 can be "Li**". Then, the electronic device 101 can input the spliced ​​vector 108 into the pre-trained risk information real-time generation model 109 to output risk information 110 for the real-time order information 102. In this application scenario, the risk information 110 can be "89".

[0040] It should be noted that the electronic device 101 can be hardware or software. When the electronic device is hardware, it can be implemented as a distributed cluster consisting of multiple servers or terminal devices, or it can be implemented as a single server or a single terminal device. When the electronic device is embodied as software, it can be installed in the hardware devices listed above. It can be implemented as multiple software or software modules for providing distributed services, for example, or it can be implemented as a single software or software module. No specific limitation is made here.

[0041] It should be understood that Figure 1 The number of electronic devices in the embodiment is only for illustration. Any number of electronic devices may be provided according to implementation requirements.

[0042] Continue to refer Figure 2, shows a process 200 of some embodiments of the risk information generation method according to the present disclosure. The risk information generation method comprises the following steps:

[0043] Step 201, obtain real-time order information and corresponding real-time order value transformation data. In some embodiments, the execution subject of the above risk information generation method (for example Figure 1 The electronic device 101 shown) can obtain real-time order information and corresponding real-time order value transformation data through a wired connection or a wireless connection. Among them, the real-time order information can be the order information received at the current moment. In practice, the above-mentioned order information can include but is not limited to at least one of the following: order number information, order user information, order item information. The real-time order value transformation data can be data corresponding to the real-time order information. In practice, the above-mentioned real-time order value transformation data can be real-time loan transaction data for the order corresponding to the real-time order information. The above-mentioned real-time loan transaction data can be a loan transaction record. For example, the real-time loan transaction data can include: loan platform information, loan amount information, loan user information, loan time information, and loan interest rate information.

[0044] Step 202, performing data processing on the above-mentioned real-time order value transformation data to obtain a data processing vector.

[0045] In some embodiments, the execution entity may perform data processing on the real-time order value transformation data to obtain a data processing vector.

[0046] As an example, the execution entity may convert the real-time order value transformation data into a vector form to obtain a data processing vector.

[0047] Step 203 , in response to determining that the pre-stored user representation vector set has a target user representation vector, vector concatenation is performed on the target user representation vector and the data processing vector to obtain a concatenated vector.

[0048] In some embodiments, in response to determining that a target user representation vector exists in a pre-stored user representation vector set, the execution entity may perform vector splicing of the target user representation vector and the data processing vector to obtain a spliced ​​vector. The user corresponding to the target user representation vector is the same user as the order user corresponding to the real-time order information, and the user representation vector set is generated offline based on a pre-trained user representation vector generation model. There is unique corresponding user information for the user representation vector in the user representation vector set. The user representation vector may represent a vector of the user's user feature information. The user feature information may include: feature association information between the user and the device, association information between the user and the loan platform, user loan frequency information, and user credit information. The user representation vector generation model may be a model for generating a user representation vector. In practice, the user representation vector generation model may be, but is not limited to, at least one of the following: a heterogeneous graph neural network model, a temporal neural network model.

[0049] As an example, the execution entity may perform horizontal splicing of the target user representation vector and the data processing vector to obtain a splicing vector.

[0050] Here, the target user representation vector and the data processing vector are concatenated to facilitate the subsequent input into the risk information to generate a real-time model.

[0051] In some optional implementations of some embodiments, the user representation vector generation model is trained by the following steps:

[0052] The first step is to obtain a first historical order value transformation data set for the historical order information set.

[0053] Among them, there is a one-to-one correspondence between the historical order information in the historical order information set and the first historical order value transformation data in the first historical order value transformation data set. The historical order information may be order information processed in historical time. The order generation time corresponding to the historical order information is later than the order generation time corresponding to the above-mentioned real-time order information. The first historical order value transformation data may be data corresponding to the historical order information. In practice, the above-mentioned first historical order value transformation data may be historical loan transaction data for the order corresponding to the historical order information. The above-mentioned historical loan transaction data may be a historical loan transaction record. For example, the historical loan transaction data may include: loan platform information, loan amount information, loan user information, loan time information, and loan interest rate information.

[0054] The second step is to construct an object relationship structure network for the above historical order information set based on a predetermined preset object set and the above first historical order value transformation data set.

[0055] The object relationship structure network includes: an order user node. The object relationship structure network can represent the object association relationship between each preset object in the preset object set. The preset object can be but is not limited to at least one of the following: a user object, a device object, a network environment object, and a merchant object. For example, the object association relationship may include: a usage relationship between a user object and a device object, a configuration relationship between a device object and a network environment object, and a transaction relationship between a user object and a merchant object.

[0056] As an example, see Figure 3 , showing a schematic diagram of the object relationship structure network.

[0057] The order user node corresponding to "User: User A" is connected to the device node corresponding to "Device: Model A", the merchant node corresponding to "Connected user: User B", the platform node corresponding to "Platform used: Platform B", the order number node corresponding to "Order number: 1***34" and the value node corresponding to "Amount involved: 10,000". The order user node corresponding to "User: A User" is connected to the device node corresponding to "Device: A Model" and the platform node corresponding to "Using Platform: A Platform". The merchant node corresponding to "Connecting User: B User" is connected to the device node corresponding to "Device: B Model" and the order user node corresponding to "User: D User". The order user node corresponding to "User: D User" is connected to the merchant node corresponding to "Connecting User: B User", the device node corresponding to "Device: B Model", the order number node corresponding to "Order Number: 2***34", the value node corresponding to "Amount Involved: 20,000", and the platform node corresponding to "Using Platform: B Platform". The device user node corresponding to "User: E User" is connected to the device node corresponding to "Device: C Model", the value node corresponding to "Amount Involved: 20,000", and the platform node corresponding to "Using Platform: B Platform". "User: F User" is connected to the platform node corresponding to "Using Platform: B Platform" and the device node corresponding to "Device: C Model".

[0058] The third step is to generate a user association graph set based on the above object relationship structure network, wherein the user association graph includes: a central node with the order user node as the center of the graph and at least one association node adjacent to the above central node.

[0059] As an example, the execution subject may generate a user association graph according to the object relationship structure network, which may include the following steps:

[0060] Step 1: Determine the order user node set in the object relationship structure network.

[0061] Step 2: For each order user node in the order user node set, a user association graph is generated with the order user node as the central node and at least one associated node adjacent to the central node as a connecting node.

[0062] The fourth step is to perform unsupervised training on the initial user representation vector generation model based on the above user association graph set to obtain the above user representation vector generation model.

[0063] The user representation vector generation model may be a heterogeneous graph model for generating user representation vectors.

[0064] As an example, the execution subject may use the user association graph set as a training sample set to perform unsupervised training on the initial user representation vector generation model to obtain the user representation vector generation model.

[0065] Optionally, the constructing of the object relationship structure network for the historical order information set according to the predetermined preset object set and the first historical order value transformation data set may include the following steps:

[0066] The first step is to construct an initial object relationship structure network based on the object association relationship between each preset object in the above preset object set.

[0067] The second step is to extract a historical order user feature data set from the first historical order value transformation data set.

[0068] The third step is to add each feature data included in the above historical order user feature data set as an associated node of the corresponding order user node to the above initial object relationship structure network, and obtain the object relationship structure network after adding the feature data as the above object relationship structure network.

[0069] Optionally, performing unsupervised training on the initial user representation vector generation model according to the user association graph to obtain the user representation vector generation model may include the following steps:

[0070] In the first step, the user association graph set is divided to obtain a first user association graph subset and a second user association graph subset.

[0071] As an example, first, the execution subject may obtain the sample number ratio between the training sample set and the verification sample set. Then, based on the sample number ratio, the user association graph set is divided to obtain a first user association graph subset and a second user association graph subset. The number of first user association graphs included in the first user association graph subset is the same as the number of training samples included in the training sample set. The number of second user association graphs included in the second user association graph subset is the same as the number of verification samples included in the verification sample set.

[0072] The second step is to construct an unsupervised contrastive learning loss function as the loss function of the initial user representation vector generation model.

[0073] In the third step, the first user association graph subset is determined as a training sample set, and the second user association graph subset is determined as a verification sample set.

[0074] In the fourth step, according to the above training sample set and the above verification sample set, and according to the above unsupervised contrastive learning loss function, the initial user representation vector generation model is subjected to unsupervised training to obtain the above user representation vector generation model.

[0075] Step 204: input the concatenated vector into a pre-trained real-time risk information generation model to output risk information for the real-time order information.

[0076] In some embodiments, the execution entity may input the concatenated vector into a pre-trained real-time risk information generation model to output risk information for the real-time order information. The real-time risk information generation model may be a model for generating risk information in real time.

[0077] The above-mentioned risk information real-time generation model can be a lightweight model. The number of parameters of the risk information real-time generation model is small, and the corresponding model calculation amount is small. The above-mentioned risk information generation model can achieve millisecond-level information prediction in an online real-time risk control environment. For example, the risk information real-time generation model can be a lightweight machine learning classification model. For example, the risk information real-time generation model can be a support vector machine (SVM). For example, the risk information can be loan risk information. For example, the above-mentioned risk information can represent the risk information that the order user cannot repay the loan on time. In practice, the above-mentioned risk information can be a value between 0-100. The larger the value, the higher the risk coefficient, and the smaller the value, the smaller the risk coefficient.

[0078] In some optional implementations of some embodiments, the above risk information real-time generation model is trained by the following steps:

[0079] The first step is to obtain a pre-stored historical user representation vector set and a corresponding second historical order value transformation data set.

[0080] Among them, there is a one-to-one correspondence between the historical user representation vector set and the second historical order value transformation data set. The generation time of the user representation vector set is later than that of the historical user representation vector set. That is, the user representation vector set is generated by vector updating and / or vector addition to the above-mentioned historical user representation vector set. The above-mentioned second historical order value transformation data may be historical loan transaction data for orders corresponding to the historical order information. The above-mentioned historical loan transaction data may be historical loan transaction records. For example, the historical loan transaction data may include: loan platform information, loan amount information, loan user information, loan time information, and loan interest rate information.

[0081] The second step is to process each second historical order value transformation data in the second historical order value transformation data set to generate a historical data processing vector and obtain a historical data processing vector set.

[0082] As an example, the execution entity may perform data vector conversion processing on each second historical order value transformation data in the second historical order value transformation data set to generate a historical data processing vector and obtain a historical data processing vector set.

[0083] The third step is to perform corresponding vector splicing on the above historical user representation vector set and the above historical data processing vector set to obtain a spliced ​​vector set.

[0084] As an example, the execution subject may perform corresponding vector horizontal splicing of the historical user representation vectors in the historical user representation vector set and the corresponding historical data processing vectors in the historical data processing vector set to generate a splicing vector and obtain a splicing vector set.

[0085] The fourth step is to determine the risk label set corresponding to the above historical user representation vector set.

[0086] The risk label represents the loan risk information of the user corresponding to the historical user representation vector. In practice, the risk label can be one of the following: high risk label, medium risk label, low risk label.

[0087] In the fifth step, each splicing vector in the above splicing vector set is input into the real-time generation model of initial risk information to generate predicted risk information, thereby obtaining a predicted risk information set.

[0088] In the sixth step, supervised training is performed on the initial risk information real-time generation model based on the predicted risk information set and the risk label set to obtain a risk information real-time generation model.

[0089] As an example, first, the execution subject may generate a loss value set based on the predicted risk information set and the risk label set, and then, based on the loss value set, perform supervised training on the initial risk information real-time generation model to obtain a risk information real-time generation model.

[0090] In some optional implementations of some embodiments, after step 204, the steps further include:

[0091] The execution entity may perform risk control operations for the order user based on the risk information. In practice, the risk control operations may include but are not limited to at least one of the following: lending operations, credit warning operations, and loan collection operations.

[0092] As an example, in response to determining that the risk information is greater than a predetermined risk value, a credit warning operation is performed for the order user. In response to determining that the risk information is less than or equal to the predetermined risk value, a loan operation is performed for the order user.

[0093] In some optional implementations of some embodiments, after step 204, the steps further include:

[0094] In the first step, in response to determining that the target user representation vector does not exist in the user representation vector set, an order user representation vector for the order user is generated by using the user representation vector generation model.

[0095] As an example, first, the execution subject may obtain a real-time order information set within a period corresponding to the real-time order information. Then, a real-time user association graph set is generated based on the real-time order information set. Finally, the real-time user association graph corresponding to the order user in the real-time user association graph set is input into the user representation vector generation model to output an order user representation vector for the order user.

[0096] The second step is to add the above order user representation vector to the above user representation vector set.

[0097] The above-mentioned various embodiments of the present disclosure have the following beneficial effects: the risk information generation method of some embodiments of the present disclosure can generate risk information for real-time order information efficiently and in real time. Specifically, the reason why risk information cannot be predicted in real time is that deep learning neural networks are often complex and have a large amount of calculation, which makes the network unable to perform real-time prediction of risk information in an online real-time risk control environment. Based on this, the risk information generation method of some embodiments of the present disclosure first obtains real-time order information and corresponding real-time order value transformation data as input source data of the real-time generation model of risk information for subsequent risk information generation. Then, the real-time order value transformation data is processed to obtain a data processing vector. By converting the real-time order value transformation data into a vector form, it is convenient for the feature learning of the subsequent real-time generation model of risk information. Then, in response to determining that the pre-stored user representation vector set has a target user representation vector, the target user representation vector and the data processing vector are vector spliced ​​to obtain a spliced ​​vector. Among them, the user corresponding to the target user representation vector is the same user as the order user corresponding to the real-time order information, and the user representation vector set is generated offline based on the pre-trained user representation vector generation model. Here, the user representation vector set is pre-generated offline through the user representation vector generation model, which reduces the amount of calculation for generating user representation vectors in the real-time risk control prediction environment, and avoids the situation where the network cannot perform real-time prediction of risk information in the online real-time risk control environment due to the complexity of the online deep learning neural network and the large amount of calculation. By determining the target user representation vector corresponding to the real-time order information from the pre-stored user representation vector set, the operation is simple and does not require more complex calculations in the real-time risk control environment, which improves the prediction speed of risk information. In addition, the target user representation vector and the above-mentioned data processing vector are vector spliced ​​to facilitate subsequent input into the real-time prediction model of risk information. Finally, the above-mentioned spliced ​​vector is input into the pre-trained real-time generation model of risk information to output risk information for the above-mentioned real-time order information. In summary, the user representation vector set generated offline by the user representation vector generation model reduces the amount of calculation for generating the user representation vector corresponding to the real-time order information in the real-time risk control environment, reduces the prediction review of the online real-time generation model of risk information, and can effectively realize the real-time prediction of risk information.

[0098] Further references Figure 4 , shows a process 400 of another embodiment of the risk information generation method according to the present disclosure. The risk information generation method comprises the following steps:

[0099] Step 401, obtaining real-time order information and corresponding real-time order value transformation data.

[0100] Step 402, performing data processing on the above-mentioned real-time order value transformation data to obtain a data processing vector.

[0101] Step 403 , in response to determining that the pre-stored user representation vector set has a target user representation vector, vector concatenation is performed on the target user representation vector and the data processing vector to obtain a concatenated vector.

[0102] Step 404: input the concatenated vector into a pre-trained real-time risk information generation model to output risk information for the real-time order information.

[0103] In some embodiments, the specific implementation of steps 401-404 and the technical effects thereof can be referred to in Figure 2 The steps 201-204 in the corresponding embodiment are not described in detail here.

[0104] Step 405: Generate an order user representation vector for the order user using the user representation vector generation model.

[0105] In some embodiments, an execution entity (e.g. Figure 1 The electronic device 101 shown can use the above-mentioned user representation vector generation model to generate an order user representation vector for the above-mentioned order user.

[0106] As an example, first, the execution subject may obtain a real-time order information set within a period corresponding to the real-time order information. Then, a real-time user association graph set is generated based on the real-time order information set. Finally, the real-time user association graph corresponding to the order user in the real-time user association graph set is input into the user representation vector generation model to output an order user representation vector for the order user.

[0107] Step 406: Update the target user representation vector in the user representation vector set using the order user representation vector.

[0108] In some embodiments, the execution subject may update the target user representation vector in the user representation vector set by using the order user representation vector, that is, replace the target user representation vector with the order user representation vector.

[0109] from Figure 4 It can be seen that Figure 2 Compared with the description of some corresponding embodiments, Figure 4 The risk information generation method in some corresponding embodiments continuously updates the user representation vector generation model to ensure the high quality of the stored user representation vectors, thereby making the subsequent online real-time prediction of risk information more accurate.

[0110] Further references Figure 5 As an implementation of the methods shown in the above figures, the present disclosure provides some embodiments of a risk information generating device. These device embodiments are similar to Figure 2 Corresponding to the method embodiments shown, the risk information generating device can be specifically applied to various electronic devices.

[0111] like Figure 5 As shown, a risk information generating device 500 includes: an acquisition unit 501, a data processing unit 502, a vector splicing unit 503 and an input unit 504. The acquisition unit 501 is configured to acquire real-time order information and corresponding real-time order value transformation data; the data processing unit 502 is configured to perform data processing on the real-time order value transformation data to obtain a data processing vector; the vector splicing unit 503 is configured to, in response to determining that a target user representation vector exists in a pre-stored user representation vector set, perform vector splicing on the target user representation vector and the data processing vector to obtain a splicing vector, wherein the user corresponding to the target user representation vector is the same user as the order user corresponding to the real-time order information, and the user representation vector set is generated offline based on a pre-trained user representation vector generation model; the input unit 504 is configured to input the splicing vector into a pre-trained risk information real-time generation model to output risk information for the real-time order information.

[0112] In some optional implementations of some embodiments, the risk information generating device 500 further includes: a first generating unit and an updating unit (not shown in the figure). The first generating unit may be configured to: generate an order user representation vector for the order user using the user representation vector generation model; the updating unit may be configured to: update the target user representation vector in the user representation vector set using the order user representation vector.

[0113] In some optional implementations of some embodiments, the risk information generating device 500 further includes: a second generating unit and an adding unit (not shown in the figure). The second generating unit may be configured to: in response to determining that the target user representation vector does not exist in the user representation vector set, generate an order user representation vector for the order user using the user representation vector generation model; and the adding unit may be configured to: add the order user representation vector to the user representation vector set.

[0114] In some optional implementations of some embodiments, the risk information generating device 500 further includes: an execution unit (not shown in the figure). The execution unit may be configured to: execute a risk control operation for the order user according to the risk information.

[0115] It is understandable that the various units recorded in the risk information generating device 500 and the reference Figure 2 Therefore, the operations, features and beneficial effects described above for the method are also applicable to the risk information generating device 500 and the units included therein, and will not be described in detail here.

[0116] Reference below Figure 6 , which shows an electronic device (eg, Figure 1 Schematic diagram of the structure of the electronic device 101)600. Figure 6 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0117] like Figure 6 As shown, the electronic device 600 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory 602 or a program loaded from a storage device 608 into a random access memory 603. Various programs and data required for the operation of the electronic device 600 are also stored in the random access memory 603. The processing device 601, the read-only memory 602, and the random access memory 603 are connected to each other via a bus 604. An input / output interface 605 is also connected to the bus 604.

[0118] Typically, the following devices may be connected to the input / output interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 6 The electronic device 600 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead. Figure 6 Each block shown in the figure may represent one device, or may represent multiple devices as required.

[0119] In particular, according to some embodiments of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, some embodiments of the present disclosure include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In some such embodiments, the computer program can be downloaded and installed from a network through a communication device 609, or installed from a storage device 608, or installed from a read-only memory 602. When the computer program is executed by the processing device 601, the above-mentioned functions defined in the method of some embodiments of the present disclosure are executed.

[0120] It should be noted that the computer-readable medium in some embodiments of the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In some embodiments of the present disclosure, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In some embodiments of the present disclosure, the computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0121] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0122] The computer-readable medium may be included in the electronic device; or it may exist independently without being installed in the electronic device. The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: obtains real-time order information and corresponding real-time order value transformation data; processes the real-time order value transformation data to obtain a data processing vector; in response to determining that a target user representation vector exists in a pre-stored user representation vector set, vector splices the target user representation vector and the data processing vector to obtain a spliced ​​vector, wherein the user corresponding to the target user representation vector is the same user as the order user corresponding to the real-time order information, and the user representation vector set is generated offline based on a pre-trained user representation vector generation model; the spliced ​​vector is input into a pre-trained risk information real-time generation model to output risk information for the real-time order information.

[0123] Computer program code for performing the operations of some embodiments of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0124] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0125] The units described in some embodiments of the present disclosure may be implemented by software or by hardware. The described units may also be arranged in a processor, for example, may be described as: a processor includes an acquisition unit, a data processing unit, a vector splicing unit and an input unit. The names of these units do not constitute a limitation on the units themselves in some cases, for example, the acquisition unit may also be described as "a unit for acquiring real-time order information and corresponding real-time order value transformation data".

[0126] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0127] Some embodiments of the present disclosure further provide a computer program product, including a computer program, which implements any of the above-mentioned risk information generation methods when executed by a processor.

[0128] The above descriptions are only some preferred embodiments of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the embodiments of the present disclosure (but not limited to) and the technical solutions formed.

Claims

1. A method for generating risk information, comprising: Obtain real-time order information and corresponding real-time order value transformation data; Performing data processing on the real-time order value transformation data to obtain a data processing vector; In response to determining that a target user representation vector exists in the pre-stored user representation vector set, vector concatenation is performed on the target user representation vector and the data processing vector to obtain a concatenated vector, wherein a user corresponding to the target user representation vector and an order user corresponding to the real-time order information are the same user, and the user representation vector set is generated offline based on a pre-trained user representation vector generation model; The concatenated vector is input into a pre-trained real-time risk information generation model to output risk information for the real-time order information.

2. The method according to claim 1, wherein: The method further comprises: Generate an order user representation vector for the order user by using the user representation vector generation model; The target user representation vector in the user representation vector set is updated using the order user representation vector.

3. The method according to claim 1, wherein: The method further comprises: In response to determining that the target user representation vector does not exist in the user representation vector set, generating an order user representation vector for the order user by using the user representation vector generation model; The order user representation vector is added to the user representation vector set.

4. The method according to claim 1, wherein: The user representation vector generation model is trained by the following steps: Obtaining a first historical order value transformation data set for the historical order information set; According to a predetermined preset object set and the first historical order value transformation data set, an object relationship structure network for the historical order information set is constructed, wherein the object relationship structure network includes: an order user node; Generate a user association graph set according to the object relationship structure network, wherein the user association graph includes: a central node with an order user node as the graph center and at least one association node adjacent to the central node; According to the user association graph set, unsupervised training is performed on the initial user representation vector generation model to obtain the user representation vector generation model.

5. The method according to claim 4, wherein: The step of constructing an object relationship structure network for the historical order information set according to a predetermined preset object set and the first historical order value transformation data set includes: Constructing an initial object relationship structure network according to the object association relationship between each preset object in the preset object set; Extracting a historical order user feature data set from the first historical order value transformation data set; Each feature data included in the historical order user feature data set is used as an associated node of the corresponding order user node and added to the initial object relationship structure network to obtain the object relationship structure network after adding the feature data as the object relationship structure network.

6. The method according to claim 4, wherein: The step of performing unsupervised training on the initial user representation vector generation model according to the user association graph set to obtain the user representation vector generation model includes: Dividing the user association graph set to obtain a first user association graph subset and a second user association graph subset; Construct an unsupervised contrastive learning loss function as the loss function of the initial user representation vector generation model; Determining the first user association graph subset as a training sample set, and determining the second user association graph subset as a verification sample set; According to the training sample set and the verification sample set, and according to the unsupervised contrastive learning loss function, the initial user representation vector generation model is subjected to model unsupervised training to obtain the user representation vector generation model.

7. The method according to claim 1, wherein: The risk information real-time generation model is trained by the following steps: Obtaining a pre-stored historical user representation vector set and a corresponding second historical order value transformation data set; Performing data processing on each second historical order value transformation data in the second historical order value transformation data set to generate a historical data processing vector, thereby obtaining a historical data processing vector set; Performing corresponding vector splicing on the historical user representation vector set and the historical data processing vector set to obtain a spliced ​​vector set; Determining a risk label set corresponding to the historical user representation vector set; Inputting each splicing vector in the splicing vector set into the initial risk information real-time generation model to generate predicted risk information, thereby obtaining a predicted risk information set; According to the predicted risk information set and the risk label set, supervised training is performed on the initial risk information real-time generation model to obtain the risk information real-time generation model.

8. The method according to claim 1, wherein: The method further comprises: Based on the risk information, a risk control operation is performed on the order user.

9. A risk information generating device, comprising: an acquisition unit, configured to acquire real-time order information and corresponding real-time order value transformation data; A data processing unit, configured to perform data processing on the real-time order value transformation data to obtain a data processing vector; a vector concatenation unit configured to, in response to determining that a target user representation vector exists in a pre-stored user representation vector set, concatenate the target user representation vector and the data processing vector to obtain a concatenated vector, wherein the user corresponding to the target user representation vector and the order user corresponding to the real-time order information are the same user, and the user representation vector set is generated offline based on a pre-trained user representation vector generation model; The input unit is configured to input the concatenated vector into a pre-trained real-time risk information generation model to output risk information for the real-time order information.

10. An electronic device comprising: one or more processors; a storage device having one or more programs stored thereon, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 8.

11. A computer readable medium having a computer program stored thereon, wherein: When the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

12. A computer program product, comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 8.

Images