A terminal payment method and system

Abstract

The application provides a terminal payment method and system, and relates to the technical field of terminal payment. The method comprises the following steps: obtaining a plurality of completed payment software information of a user and to-be-paid software information; determining user login information of each completed payment software and payment verification information of each completed payment software based on the plurality of completed payment software information of the user; determining a plurality of different detailed pre-login information of the to-be-paid software based on the user login information of each completed payment software and the to-be-paid software information; obtaining and pre-logging in based on the login information of the to-be-paid software selected by the user; determining a target payment verification scheme of the to-be-paid software based on a plurality of initial payment verification schemes of the to-be-paid software; and performing payment verification on the to-be-paid software based on the target payment verification scheme of the to-be-paid software. The method can accurately determine a payment verification scheme that is most suitable for the user and takes into account safety and convenience.

Description

Technical Field

[0001] This invention relates to the field of terminal payment technology, and specifically to a terminal payment method and system. Background Technology

[0002] In today's rapidly growing digital payment landscape, users frequently switch between different payment applications to complete transactions in various scenarios. However, when making payments in a new application, users often face tedious processes of repeated registration, login, and manual entry of personal information, which is both time-consuming and detrimental to the payment experience. Meanwhile, balancing payment security and verification convenience remains a challenge in digital payment scenarios. If the verification process is too simple, payment security is compromised, potentially leading to information leaks and malicious payments. Conversely, if the process is too complex, it increases the user's workload and may even interrupt the payment process. Existing payment verification solutions often employ fixed verification models, failing to flexibly adapt to users' past payment habits and the specific characteristics of the application, thus failing to balance user needs for convenience with payment security. Furthermore, different users have varying levels of trust in payment applications, and existing solutions do not adequately consider the impact of this subjective factor on the verification process, resulting in a lack of personalization and targeting, and an inability to meet diverse payment scenarios and user needs.

[0003] Therefore, accurately determining the most suitable payment verification solution for users while also ensuring security and convenience is an urgent problem to be solved. Summary of the Invention

[0004] The main technical problem this invention addresses is how to accurately determine the payment verification scheme that is most suitable for users while also being secure and convenient.

[0005] According to a first aspect, the application provides a terminal payment method, comprising: obtaining a plurality of completed payment software information of a user, and to-be-paid software information; determining, based on the plurality of completed payment software information of the user, user login information of each completed payment software and payment verification information of each completed payment software; generating, based on the user login information of each completed payment software and the to-be-paid software information, pre-login information of the to-be-paid software in a plurality of different levels of detail; displaying the pre-login information of the to-be-paid software in a plurality of different levels of detail to the user, and performing pre-login based on the login information of the to-be-paid software selected by the user; determining, based on the login information of the to-be-paid software selected by the user, user trust degree of the to-be-paid software and payment security index of the to-be-paid software; determining, based on the user trust degree of the to-be-paid software, the payment security index of the to-be-paid software and the to-be-paid software information, a highest security verification complexity level acceptable to the user and a lowest convenient verification complexity level acceptable to the user; generating, based on the highest security verification complexity level acceptable to the user and the lowest convenient verification complexity level acceptable to the user, a plurality of preliminary payment verification schemes of the to-be-paid software; determining, based on the plurality of preliminary payment verification schemes of the to-be-paid software, a target payment verification scheme of the to-be-paid software; and performing payment verification on the to-be-paid software based on the target payment verification scheme of the to-be-paid software.

[0006] In a possible implementation, the determining, based on the plurality of preliminary payment verification schemes of the to-be-paid software, the target payment verification scheme of the to-be-paid software comprises: constructing a payment verification graph, the payment verification graph comprising a plurality of nodes and a plurality of edges between the plurality of nodes, the plurality of nodes comprising a plurality of preliminary payment verification nodes and a plurality of completed payment software nodes, wherein each preliminary payment verification node establishes an edge with each completed payment software node, the node feature of each preliminary payment verification node is each preliminary payment verification scheme of the to-be-paid software, the node feature of each completed payment software node is payment verification information of each completed payment software, and the edge between the preliminary payment verification node and the completed payment software node is similarity of the to-be-paid software information and the completed payment software information and similarity of the login information of the to-be-paid software and the user login information of the completed payment software; and determining, based on graph neural network processing on the payment verification graph, the target payment verification scheme of the to-be-paid software.

[0007] In a possible implementation, the generating, based on the highest security verification complexity level acceptable to the user and the lowest convenient verification complexity level acceptable to the user, the plurality of preliminary payment verification schemes of the to-be-paid software comprises: generating, based on the highest security verification complexity level acceptable to the user and the lowest convenient verification complexity level acceptable to the user, the plurality of preliminary payment verification schemes of the to-be-paid software using a generative adversarial network.

[0008] In a possible implementation, an input of the graph neural network is the payment verification graph, and an output of the graph neural network is the target payment verification scheme of the to-be-paid software.

[0009] According to a second aspect, the present application provides a terminal payment system, comprising: an acquisition module configured to acquire a plurality of completed payment software information of a user and to-be-paid software information; a payment information determination module configured to determine, based on the plurality of completed payment software information of the user, user login information of each completed payment software and payment verification information of each completed payment software; a pre-login information determination module configured to generate a plurality of pre-login information of different levels of detail of the to-be-paid software based on the user login information of each completed payment software and the to-be-paid software information; a pre-login module configured to display the plurality of pre-login information of different levels of detail of the to-be-paid software to the user, acquire and pre-login based on login information of the to-be-paid software selected by the user; a trust and security index determination module configured to determine, based on the login information of the to-be-paid software selected by the user, a user trust degree of the to-be-paid software and a payment security index of the to-be-paid software; a complexity level determination module configured to determine, based on the user trust degree of the to-be-paid software, the payment security index of the to-be-paid software and the to-be-paid software information, a highest security verification complexity level acceptable to the user and a lowest convenient verification complexity level acceptable to the user; a preliminary scheme generation module configured to generate a plurality of preliminary payment verification schemes of the to-be-paid software based on the highest security verification complexity level acceptable to the user and the lowest convenient verification complexity level acceptable to the user; a target scheme determination module configured to determine a target payment verification scheme of the to-be-paid software based on the plurality of preliminary payment verification schemes of the to-be-paid software; and a payment verification module configured to perform payment verification on the to-be-paid software based on the target payment verification scheme of the to-be-paid software.

[0010] In a possible implementation, the target scheme determination module is further configured to: construct a payment verification graph, the payment verification graph comprising a plurality of nodes and a plurality of edges between the plurality of nodes, the plurality of nodes comprising a plurality of preliminary payment verification nodes and a plurality of completed payment software nodes, wherein each preliminary payment verification node establishes an edge with each completed payment software node, a node feature of each preliminary payment verification node is each preliminary payment verification scheme of the to-be-paid software, a node feature of each completed payment software node is payment verification information of each completed payment software, and an edge between a preliminary payment verification node and a completed payment software node is a similarity between to-be-paid software information and completed payment software information and a similarity between login information of the to-be-paid software and user login information of the completed payment software; and determine, based on processing of the payment verification graph by a graph neural network, the target payment verification scheme of the to-be-paid software.

[0011] In a possible implementation, the preliminary scheme generation module is further configured to generate a plurality of preliminary payment verification schemes of the to-be-paid software based on the highest security verification complexity level acceptable to the user and the lowest convenience verification complexity level acceptable to the user using a generative adversarial network.

[0012] In a possible implementation, an input of the graph neural network is the payment verification graph, and an output of the graph neural network is the target payment verification scheme of the to-be-paid software.

[0013] According to a third aspect, embodiments of the present application provide an electronic device, comprising: a processor; a memory; and a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement the method as described above, the method comprising: obtaining a plurality of completed payment software information of a user, to-be-paid software information; determining user login information of each completed payment software and payment verification information of each completed payment software based on the plurality of completed payment software information of the user; generating a plurality of pre-login information of different levels of detail of the to-be-paid software based on the user login information of each completed payment software and the to-be-paid software information; displaying the plurality of pre-login information of different levels of detail of the to-be-paid software to the user, and obtaining and pre-login based on the login information of the to-be-paid software selected by the user; determining user trust degree of the to-be-paid software and payment security index of the to-be-paid software based on the login information of the to-be-paid software selected by the user; determining a highest security verification complexity level acceptable to the user and a lowest convenience verification complexity level acceptable to the user based on the user trust degree of the to-be-paid software, the payment security index of the to-be-paid software, and the to-be-paid software information; generating a plurality of preliminary payment verification schemes of the to-be-paid software based on the highest security verification complexity level acceptable to the user and the lowest convenience verification complexity level acceptable to the user; determining a target payment verification scheme of the to-be-paid software based on the plurality of preliminary payment verification schemes of the to-be-paid software; and performing payment verification on the to-be-paid software based on the target payment verification scheme of the to-be-paid software.

[0014] According to a fourth aspect, the embodiments provide a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the terminal payment method provided above, the method comprising: obtaining a plurality of completed payment software information of a user, and to-be-paid software information; determining, based on the plurality of completed payment software information of the user, user login information of each completed payment software and payment verification information of each completed payment software; generating, based on the user login information of each completed payment software and the to-be-paid software information, pre-login information of the to-be-paid software in a plurality of different levels of detail; displaying the pre-login information of the to-be-paid software in a plurality of different levels of detail to the user, and obtaining and pre-logging based on login information of the to-be-paid software selected by the user; determining, based on the login information of the to-be-paid software selected by the user, a user trust degree of the to-be-paid software and a payment security index of the to-be-paid software; determining, based on the user trust degree of the to-be-paid software, the payment security index of the to-be-paid software and the to-be-paid software information, a highest security verification complexity level acceptable to the user and a lowest convenient verification complexity level acceptable to the user; generating, based on the highest security verification complexity level acceptable to the user and the lowest convenient verification complexity level acceptable to the user, a plurality of preliminary payment verification schemes of the to-be-paid software; determining, based on the plurality of preliminary payment verification schemes of the to-be-paid software, a target payment verification scheme of the to-be-paid software; and performing payment verification on the to-be-paid software based on the target payment verification scheme of the to-be-paid software.

[0015] The application provides a terminal payment method and system, the method comprises the following steps: obtaining a plurality of completed payment software information of a user and to-be-paid software information; determining user login information of each completed payment software and payment verification information of each completed payment software based on the plurality of completed payment software information of the user; generating a plurality of different detailed pre-login information of the to-be-paid software based on the user login information of each completed payment software and the to-be-paid software information; displaying the plurality of different detailed pre-login information of the to-be-paid software to the user, and obtaining and pre-logging the login information of the to-be-paid software selected by the user; determining user trust degree of the to-be-paid software and payment security index of the to-be-paid software based on the login information of the to-be-paid software selected by the user; determining a highest security verification complexity level acceptable by the user and a lowest convenient verification complexity level acceptable by the user based on the user trust degree of the to-be-paid software, the payment security index of the to-be-paid software and the to-be-paid software information; generating a plurality of primary payment verification schemes of the to-be-paid software based on the highest security verification complexity level acceptable by the user and the lowest convenient verification complexity level acceptable by the user; determining a target payment verification scheme of the to-be-paid software based on the plurality of primary payment verification schemes of the to-be-paid software; and performing payment verification on the to-be-paid software based on the target payment verification scheme of the to-be-paid software, so that the payment verification scheme most suitable for the user and taking into account the security and convenience can be accurately determined. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A flowchart of a terminal payment method provided by the embodiment of the application is shown in the figure;

[0017] Figure 2 A schematic diagram of a local terminal provided by the embodiment of the application is shown in the figure;

[0018] Figure 3 A flowchart of determining a target payment verification scheme of to-be-paid software provided by the embodiment of the application is shown in the figure;

[0019] Figure 4 A schematic diagram of a terminal payment system provided by the embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0020] The application will be described in further detail below with specific reference being made to the drawings. Like elements are referenced with like numerals throughout the several figures of the drawings. In the following description, numerous specific details are discussed to provide a thorough understanding of the application. However, it will be apparent to one of ordinary skill in the art that the application can be practiced without these specific details. In other instances, well-known methods have not been described in detail in order to avoid unnecessarily obscuring the application. Also, some terminology can be used in the description for the sake of clarity.

[0021] In the embodiment of the application, a terminal payment method is provided, as shown in Figure 1 The terminal payment method comprises steps S1-S9.

[0022] In step S1, a plurality of completed payment software information and to-be-paid software information of a user are acquired.

[0023] The plurality of completed payment software of the user are a plurality of software that have been used and payment operations have been completed in the user equipment.

[0024] The to-be-paid software is software that needs to be paid by the user and is not in a valid login state.

[0025] The software information comprises a software name, a software unique application ID and a software version.

[0026] In step S2, user login information of each completed payment software and payment verification information of each completed payment software are determined based on the plurality of completed payment software information of the user.

[0027] The user login information of each completed payment software is personal information left by the user after logging in the corresponding completed payment software, which is acquired by reading the local terminal. The user login information of each completed payment software comprises a user number, identity information, an email address, a user ID and the like. Figure 2 A schematic diagram of a local terminal provided in the embodiment of the application is shown in Figure 2 The terminal can be a notebook computer or a desktop computer.

[0028] The payment verification information of each completed payment software is verification mode information required by the user when making payment in the corresponding completed payment software, which is acquired by reading the local terminal.

[0029] The payment verification information specifically includes a verification method habitually used by the user, such as a short message verification code, biometric identification, third-party payment platform authorization verification, and payment password verification.

[0030] The payment verification information can reflect the payment verification habits and preferences of the user.

[0031] In step S3, pre-login information of multiple different levels of detail for the to-be-paid software is generated based on the user login information of each completed payment software and the to-be-paid software information.

[0032] In some embodiments, the pre-login information of multiple different levels of detail for the to-be-paid software can be generated using a pre-login information generation model. The pre-login information generation model is a generative adversarial network. The input of the pre-login information generation model is the user login information of each completed payment software and the to-be-paid software information, and the output of the pre-login information generation model is the pre-login information of multiple different levels of detail for the to-be-paid software.

[0033] A generative adversarial network (GAN) is a model composed of two sub-networks, a generator and a discriminator. Both the generator and the discriminator are trained through mutual game playing. The task of the generator is to learn the distribution of real data and generate new data samples similar to the real data. The task of the discriminator is to judge whether the input data is real or generated by the generator. During the training process, the generator constantly strives to generate more realistic data, while the discriminator constantly improves its discrimination ability. Finally, when the discriminator cannot effectively distinguish between real and generated data, it means that the generator has been able to generate high-quality data.

[0034] The pre-login information of multiple different levels of detail for the to-be-paid software is a set of quick login information schemes for the to-be-paid software generated by the pre-login information generation model, which can be selected by the user.

[0035] Each pre-login information of different levels of detail is generated by extracting and combining the user login information of other completed payment software. The levels of detail and privacy sensitivity of the information in different pre-login information are different. For example, one pre-login information may only include the user's nickname and avatar, another pre-login information may include the mobile phone number and email, and a more detailed pre-login information may include de-sensitized identity information.

[0036] The user login information of each completed payment software contains the user's verified identity credentials in different application scenarios, such as mobile phone number, email, user ID, etc. These information constitutes the user's digital identity portrait. The to-be-paid software information can define the target application scenario that needs to be pre-logged in. These inputs provide the model with raw materials and target constraints for generating pre-login information, enabling the model to generate information combinations that meet both user identity characteristics and the requirements of the to-be-paid software login.

[0037] The generative adversarial network can learn the combination rules and correlations within the user login information of each completed payment software, such as the user's mobile phone number being usually bound with specific real-name information. The goal of the generator is to create multiple sets of new user information combinations, i.e., multiple different detailed pre-login information of the to-be-paid software. These generated information combinations are set to multiple gradients in terms of detail, from low-detail combinations containing only basic nicknames, to medium-detail combinations containing mobile phone numbers and emails, to high-detail combinations containing de-identified identity information. The discriminator can evaluate the rationality and security of the pre-login information combinations created by the generator. The discriminator can determine whether a generated information combination is suitable for the current pre-login scenario based on the pre-set security policy and the risk level of the application scenario defined by the to-be-paid software information. For example, if the to-be-paid software is a high-risk financial application, the discriminator will give low scores to those overly simple information combinations, and thus consider them as not meeting the requirements. Through continuous adversarial training, the generator can continuously adjust its generation strategy, and ultimately create multiple different detailed pre-login information that can pass the discriminator's evaluation and form a clear gradient in information detail.

[0038] Step S4, displaying the multiple different detailed pre-login information of the to-be-paid software to the user, and pre-logging in based on the user-selected login information of the to-be-paid software.

[0039] The user-selected login information of the to-be-paid software is the information that the user autonomously selects from the displayed multiple different detailed pre-login information of the to-be-paid software for pre-login of the to-be-paid software.

[0040] Step S5, determining the user trust degree of the to-be-paid software and the payment security index of the to-be-paid software based on the user-selected login information of the to-be-paid software.

[0041] In some embodiments, a security analysis model can be used to determine the user trust degree of the to-be-paid software and the payment security index of the to-be-paid software. The security analysis model is a deep neural network model. The input of the security analysis model is the user-selected login information of the to-be-paid software, and the output of the security analysis model is the user trust degree of the to-be-paid software and the payment security index of the to-be-paid software.

[0042] A deep neural network model includes a deep neural network (DNN). The deep neural network is a complex computational model capable of imitating the structure of human brain neuron network to process information. The deep neural network is composed of multiple processing layers, which can include an input layer, one or more hidden layers, and an output layer, each layer containing a number of neurons, and the layers are connected through weights. Data enters from the input layer, undergoes nonlinear transformation and layer-by-layer feature extraction in each hidden layer, and finally obtains a result in the output layer. The deep neural network can perform advanced tasks such as classification, regression, and prediction by learning complex patterns and relationships in a large amount of data.

[0043] The user trust degree of the to-be-paid software is an index reflecting the user's trust degree for the to-be-paid software, which is determined by analyzing the login information of the to-be-paid software selected by the user through a security analysis model. The user trust degree of the to-be-paid software is embodied by a numerical value, and the higher the numerical value, the higher the user's trust degree.

[0044] The numerical value of the user trust degree of the to-be-paid software can directly reflect the degree of the user's willingness to disclose personal information to the software.

[0045] The payment security index of the to-be-paid software is a numerical index reflecting the security degree of the payment process of the to-be-paid software, which is obtained by analyzing the login information of the to-be-paid software selected by the user through a security analysis model.

[0046] The payment security index of the to-be-paid software depends on the importance and sensitivity of the login information selected by the user.

[0047] The login information of the to-be-paid software selected by the user directly reflects the range and detail of the personal information that the user is willing to provide to the to-be-paid software, and this selection behavior of the user can reflect the user's trust tendency for the software. At the same time, the importance of login information of different detail degrees is different, and the content and quantity of sensitive information contained are also different, which can provide direct and key basis for the model to judge the user trust degree and payment security index of the to-be-paid software.

[0048] The deep neural network model can perform data preprocessing on the login information of the selected to-be-paid software to convert it into a format that can be processed by the model. The deep neural network can transmit the preprocessed data into the network through the input layer, and then extract key features in the login information through layer-by-layer processing of multiple hidden layers, such as information detail level, sensitive information inclusion, information integrity, etc. The model can map the correspondence between the detail level of the login information of the selected to-be-paid software and the user trust level according to the learned feature rules. For example, when the user selects high-level pre-login information containing a large amount of sensitive information, the model can determine that the user has a high degree of trust in the software, and then output a corresponding high trust value. If the user only selects basic-level pre-login information, the model will output a lower corresponding trust value. When calculating the payment security index of the to-be-paid software, the model can focus on analyzing the number and importance of important information in the selected login information, and at the same time combining the security impact weight of such information in the payment scenario, the model can calculate the specific index reflecting the payment security degree through the operation and fitting of the multi-layer neural network.

[0049] In some embodiments, the determination of the user trust level of the to-be-paid software and the payment security index of the to-be-paid software based on the login information of the selected to-be-paid software includes steps S21-S23:

[0050] Step S21, determining the information inclusion type, login information simplification level, information sensitivity rating, and session identity trust level based on the login information of the selected to-be-paid software.

[0051] In some embodiments, a deep neural network can be used to determine the information inclusion type, login information simplification level, information sensitivity rating, and session identity trust level.

[0052] The information inclusion type is the specific content category covered by the login information of the to-be-paid software output by the deep neural network, such as only basic information (nickname / avatar), containing a mobile phone number, containing de-identified identity information, etc.

[0053] The login information simplification level is the simplification level of the login information output by the deep neural network, and the login information simplification level is divided into categories such as minimal, basic, and detailed.

[0054] The information sensitivity rating is the level of the sensitivity of private information in the login information of the selected to-be-paid software determined by the deep neural network.

[0055] The session identity trust level is a level that can reflect the reliability of user identity verification, which is evaluated by the deep neural network according to the completeness, validity, and closeness to user identity of the login information of the selected to-be-paid software.

[0056] Deep neural networks have strong feature extraction and pattern recognition capabilities. Through training a large number of samples, deep neural networks can learn the field composition rules of different login information, so as to accurately divide the information containing types, such as distinguishing the categories containing only basic information or containing sensitive information. At the same time, deep neural networks can quantitatively analyze the number of information fields and redundancy, and automatically determine the login information simplification level according to the preset standard. The model can identify the feature patterns of private information, and output the information sensitivity rating level according to the pre-trained sensitive weight system. In addition, the model can also comprehensively evaluate the integrity of the information, the closeness of the association with the user identity, and the effectiveness, so as to generate a session identity trust level reflecting the reliability of identity verification.

[0057] Step S22, determining the user's preference degree for privacy sensitivity, the user's acceptance degree for cross-software information reuse, and the user's demand intensity for quick login based on the information containing type, the information sensitivity rating, and the session identity trust level.

[0058] In some embodiments, deep neural networks can be used to determine the user's preference degree for privacy sensitivity, the user's acceptance degree for cross-software information reuse, and the user's demand intensity for quick login.

[0059] The user's preference degree for privacy sensitivity is a numerical indicator of the user's tendency to value personal information privacy protection in the login scenario determined by the deep neural network.

[0060] The user's acceptance degree for cross-software information reuse is a numerical indicator of the user's willingness to reuse information of other software to the current software login determined by the deep neural network.

[0061] The user's demand intensity for quick login is a numerical indicator of the user's urgency to quickly complete login by simplifying the login process determined by the deep neural network.

[0062] Deep neural networks have strong implicit association mining and rule learning capabilities. Through continuous training of real scene samples, deep neural networks can capture the potential association features between information containing types, information sensitivity ratings, and session identity trust levels, and can deeply mine the deep corresponding logic between these attributes and user subjective tendencies. This learning ability verified by a large amount of data enables deep neural networks to effectively integrate the objective information of input data and accurately extract the key elements affecting user subjective preferences, and thus can realize accurate conversion from objective information attributes to user subjective preference indicators.

[0063] Step S23, determining the user trust degree of the to-be-paid software, the payment security index of the to-be-paid software based on the information containing type, the information sensitivity rating, the session identity trust level, the user's privacy sensitivity preference, the user's cross-software information reuse acceptance, and the user's fast login demand intensity.

[0064] In some embodiments, a deep neural network can be used to determine the user trust degree of the to-be-paid software and the payment security index of the to-be-paid software.

[0065] The deep neural network can quantitatively disassemble the content category attribute of the information containing type, the risk level of the information sensitivity rating, and the verification strength of the session identity trust level, and simultaneously convert the user's privacy sensitivity preference, cross-software information reuse acceptance, and fast login demand intensity into a calculable feature vector. Subsequently, the deep neural network can filter and fuse these features layer by layer through a multi-layer network structure, and can combine the scene rules learned by the model in the training samples to further establish the corresponding relationship between different parameter combinations and the user trust degree and the payment security index. The deep neural network can complete the conversion from the input information to the user trust degree of the to-be-paid software and the payment security index of the to-be-paid software through nonlinear transformation.

[0066] Step S6, determining the user-acceptable highest security verification complexity level and the user-acceptable lowest convenient verification complexity level based on the user trust degree of the to-be-paid software, the payment security index of the to-be-paid software, and the to-be-paid software information.

[0067] In some embodiments, a complexity determination model can be used to determine the user-acceptable highest security verification complexity level and the user-acceptable lowest convenient verification complexity level. The complexity determination model is a Transformer model. The input of the complexity determination model is the user trust degree of the to-be-paid software, the payment security index of the to-be-paid software, and the to-be-paid software information, and the output of the complexity determination model is the user-acceptable highest security verification complexity level and the user-acceptable lowest convenient verification complexity level.

[0068] The Transformer model is a deep learning architecture that uses a self-attention mechanism. The Transformer model can process sequence data in parallel and effectively capture long-range dependencies within the sequence. The Transformer model mainly consists of an encoder and a decoder, and one part of the model can also be used alone. The self-attention mechanism of the Transformer model allows the model to weigh the importance of all other elements in the input sequence when processing an element, thereby better understanding the context relationship. Through the multi-head attention mechanism, the Transformer model can learn relevant information from different representation subspaces.

[0069] The highest security verification complexity level acceptable to the user is determined by the complexity determination model, which defines the highest level of security verification complexity that the user can accept during the payment verification process of the to-be-paid software. The higher the level, the more complex the steps and methods of security verification, and the higher the level of security protection.

[0070] The lowest convenient verification complexity level acceptable to the user is determined by the complexity determination model, which defines the lowest level of convenient verification complexity that the user can accept during the payment verification process of the to-be-paid software. The lower the level, the more simple the steps and methods of security verification, and the more convenient the user operation.

[0071] The user trust level of the to-be-paid software, the payment security index of the to-be-paid software, and the to-be-paid software information provide a decision basis for the model to determine the verification complexity level. A high user trust level indicates that the user has a higher acceptance of the software and is more inclined to accept simplified verification methods under the premise of ensuring basic security. A low payment security index indicates that the security basis based on the user's selected login information is weak, and the verification strength needs to be improved to enhance the risk prevention and control capability of the payment link. The to-be-paid software information provides a scenario-based risk background.

[0072] The Transformer model can calculate the correlation weights between the three input elements, i.e., the user trust degree of the to-be-paid software, the payment security index of the to-be-paid software, and the to-be-paid software information, through the self-attention mechanism. For example, the model can find that when the payment security index is low, the influence weights of the user trust degree and the to-be-paid software information on the final complexity level will change. Through the processing of multiple layers of encoders, the model can deeply understand the complex patterns of these input combinations. For example, the Transformer model can learn that the combination of "high user trust degree" and "high payment security index" can correspond to a relatively low minimum convenient verification complexity level, because the user trusts the application and the information provided by the user is also safe enough. At the same time, the model can also learn that when the "payment security index" is extremely low, no matter how high the "user trust degree" is, a very high maximum security verification complexity level must be set to ensure security. Finally, the output layer of the model will accurately map the user-acceptable maximum security verification complexity level and the user-acceptable minimum convenient verification complexity level according to these deeply learned correlation relationships.

[0073] Step S7, generating a plurality of preliminary payment verification schemes of the to-be-paid software based on the user-acceptable maximum security verification complexity level and the user-acceptable minimum convenient verification complexity level.

[0074] In some embodiments, the plurality of preliminary payment verification schemes of the to-be-paid software can be generated based on the user-acceptable maximum security verification complexity level and the user-acceptable minimum convenient verification complexity level using a generative adversarial network. The input of the generative adversarial network is the user-acceptable maximum security verification complexity level and the user-acceptable minimum convenient verification complexity level, and the output of the generative adversarial network is the plurality of preliminary payment verification schemes of the to-be-paid software.

[0075] The payment verification scheme refers to a series of specific executable identity verification and payment authorization processes combined that the user needs to perform when initiating a payment request after completing the pre-login, binding a bank card or a payment account, and other payment pre-operation of the to-be-paid software.

[0076] The payment verification scheme can be used to confirm that the payment behavior is operated by the user himself and prevent payment risks.

[0077] The plurality of preliminary payment verification schemes of the to-be-paid software are a plurality of payment verification schemes generated by the generative adversarial network according to the complexity level range, which meet the scene risk and user acceptance.

[0078] Each of the preliminary payment verification schemes contains one or more verification steps, such as "entering an SMS verification code", "conducting fingerprint recognition", "answering a security question", and the like. The multiple preliminary payment verification schemes differ in security and convenience, but are all within the interval defined by the highest security verification complexity level and the lowest convenient verification complexity level.

[0079] The generator in the generative adversarial network can create diversified verification schemes within the interval defined by the highest security verification complexity level and the lowest convenient verification complexity level. The generator attempts to combine different verification methods, such as SMS, fingerprint, facial recognition, and payment password, and can assign a theoretical complexity score to each combination. For example, a scheme containing double verification of "fingerprint recognition" and "SMS verification code" has a higher complexity score than a single verification scheme containing only "payment password". The goal of the generator is to generate a series of preliminary payment verification schemes, and the complexity scores are distributed between the lowest convenient verification complexity level and the highest security verification complexity level acceptable to the user. The discriminator is responsible for evaluating the effectiveness and rationality of these generated schemes. The discriminator can learn a large number of effective payment verification combinations in real environments and determine whether the schemes generated by the generator are logical and have security vulnerabilities. For example, the discriminator can know that the combination of "answering a security question" and "email verification" is not secure in some scenarios. Through this adversarial training, the generator can continuously optimize its scheme generation strategy, and ultimately output multiple preliminary payment verification schemes for the to-be-paid software. These schemes not only meet the complexity requirements, but are also secure and feasible in actual application.

[0080] Step S8, determining a target payment verification scheme for the to-be-paid software based on the multiple preliminary payment verification schemes for the to-be-paid software.

[0081] In some embodiments, Figure 3 A flowchart of a process for determining a target payment verification scheme for a to-be-paid software is provided in some embodiments of the present application. The process for determining a target payment verification scheme for a to-be-paid software includes steps S31-S32:

[0082] Step S31, a payment verification graph is constructed, the payment verification graph includes a plurality of nodes and a plurality of edges between the plurality of nodes, the plurality of nodes includes a plurality of preliminary payment verification nodes and a plurality of completed payment software nodes, wherein each preliminary payment verification node respectively establishes an edge with each completed payment software node, the node feature of each preliminary payment verification node is each preliminary payment verification scheme of the to-be-paid software, the node feature of each completed payment software node is the payment verification information of each completed payment software, and the edge between the preliminary payment verification node and the completed payment software node is the similarity of the to-be-paid software information and the completed payment software information and the similarity of the login information of the to-be-paid software and the user login information of the completed payment software.

[0083] The payment verification graph is a network structured data composed of nodes and edges, and can represent the complex relationship between the preliminary verification scheme of the to-be-paid software and the user historical payment habit. The payment verification graph is composed of preliminary payment verification nodes, completed payment software nodes and edges connecting them. Each preliminary payment verification node respectively establishes an edge with each completed payment software node, the node feature of each preliminary payment verification node is each preliminary payment verification scheme of the to-be-paid software, the node feature of each completed payment software node is the payment verification information of each completed payment software, and the edge between the preliminary payment verification node and the completed payment software node is the similarity of the to-be-paid software information and the completed payment software information and the similarity of the login information of the to-be-paid software and the user login information of the completed payment software.

[0084] By constructing the payment verification graph, the preliminary payment verification scheme of the to-be-paid software can be structurally associated with the payment behavior characteristics of the user using the completed payment software in the past, thereby forming an associated network that can be quantitatively analyzed.

[0085] Step S32, the target payment verification scheme of the to-be-paid software is determined based on processing of the payment verification graph by a graph neural network.

[0086] The graph neural network (GNN) is a deep learning model capable of processing graph structured data. The graph neural network can directly operate on the nodes and edges of the graph. The graph neural network can learn and reason using the node features, edges and topological structure information in the graph. The graph neural network can realize classification, regression or generation of nodes in the graph by aggregating the features of the node itself and its neighbor nodes. The input of the graph neural network is the payment verification graph, and the output of the graph neural network is the target payment verification scheme of the to-be-paid software.

[0087] The target payment verification scheme of the to-be-paid software is an optimal payment verification scheme selected from multiple preliminary payment verification schemes of the to-be-paid software through processing of a payment verification graph by a graph neural network.

[0088] The target payment verification scheme of the to-be-paid software meets the verification complexity requirement acceptable by the user, has a high degree of adaptation to the payment verification habit of the user in using the completed payment software, and can guarantee the security and convenience of the payment process.

[0089] By constructing the payment verification graph, the association network between the preliminary payment verification scheme of the to-be-paid software and the completed payment software can be clearly presented. Since the acceptance of the new payment verification scheme by the user is often associated with the verification habit of the user in using the payment software, this association information plays an important role in determining the target payment verification scheme. The verification characteristics of the preliminary payment verification scheme and the payment verification information of the completed payment software are taken as node features, and the information similarity and login information similarity between the to-be-paid software and the completed payment software are taken as edge features, so that the related data value can be more fully mined and utilized. This helps the model to better understand the adaptation logic of the preliminary scheme to the historical verification habit of the user and the association rule of different payment scenarios of software, and to improve the accuracy of screening of the target verification scheme.

[0090] The graph neural network can aggregate the features of each preliminary payment verification node and the features of the completed payment software nodes connected thereto. In the aggregation process, the model can give different weights according to the similarity of the edges. The higher the similarity of the completed payment software node, the greater the influence weight of the node feature on the preliminary payment verification node. For example, if the information of a completed payment software is highly similar to the information of the to-be-paid software, and the login information of the completed payment software is also highly similar to the login information of the to-be-paid software, the payment verification information of the completed payment software as a node feature will be given a higher weight, and thus will have a greater impact on the corresponding preliminary payment verification node in the aggregation. The graph neural network can continuously integrate global feature information and gradually mine deep information such as the adaptation rule of the preliminary payment verification scheme to the historical payment verification habit of the user and the balance between security and convenience of the preliminary scheme. Then, the graph neural network can decode the features of each preliminary payment verification node to obtain a comprehensive evaluation score of each preliminary scheme. The evaluation dimensions of the model include the adaptation degree to the user habit, the security verification effect, and the operation convenience. The model can select the preliminary payment verification scheme with the highest comprehensive evaluation score as the target payment verification scheme of the to-be-paid software.

[0091] Step S9, performing payment verification on the to-be-paid software based on the target payment verification scheme of the to-be-paid software.

[0092] Once the target payment verification scheme for the software to be paid is determined, when the user initiates a payment operation, the preset verification process and operation steps in the scheme are executed sequentially based on the target payment verification scheme for the software to be paid, so as to ensure the security and compliance of the payment operation.

[0093] Based on the same inventive concept Figure 4 This is a schematic diagram of a terminal payment system provided in an embodiment of the present invention. The terminal payment system includes:

[0094] Module 41 is used to obtain information on multiple completed payment software and pending payment software for the user;

[0095] The payment information determination module 42 is used to determine the user login information and payment verification information of each completed payment software based on the user's multiple completed payment software information;

[0096] The pre-login information determination module 43 is used to generate various levels of pre-login information for the software to be paid based on the user login information of each completed payment software and the information of the software to be paid.

[0097] The pre-login module 44 is used to display various levels of pre-login information of the software to be paid to the user, and to obtain and perform pre-login based on the login information of the software to be paid selected by the user.

[0098] The trust security index determination module 45 is used to determine the user trust level and the payment security index of the software to be paid based on the login information of the software to be paid selected by the user.

[0099] Complexity level determination module 46 is used to determine the highest acceptable security verification complexity level and the lowest acceptable convenience verification complexity level for the user based on the user trust level of the software to be paid, the payment security index of the software to be paid, and the information of the software to be paid.

[0100] The initial scheme generation module 47 is used to generate multiple initial payment verification schemes for the software to be paid based on the highest security verification complexity level acceptable to the user and the lowest convenience verification complexity level acceptable to the user.

[0101] The target scheme determination module 48 is used to determine the target payment verification scheme for the software to be paid based on multiple preliminary payment verification schemes of the software to be paid.

[0102] The payment verification module 49 is used to perform payment verification on the software to be paid based on the target payment verification scheme of the software to be paid.

[0103] Furthermore, the order of the processing elements and sequence of the steps described in this specification are not intended to be limiting, unless otherwise indicated, and is presented according to the order in which they are believed to be performed, from the view of the applicant, and / or as is practical in light of the functionality provided. Moreover, the use of numbering or letters in the examples, if any, are not intended to limit their application to the narrow set of examples described. Rather, such numbering and / or letters are merely artifacts of the inventor's attempts to describe the embodiment in a detailed or orderly manner. Also, although the above-disclosed subject matter has been described in language specific to structural features, methodical steps, and / or computational acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features, steps, and / or acts described. Rather, the specific features, steps, and / or acts described above are disclosed as example forms of implementing the claims.

[0104] Similarly, it is to be noticed that the term "comprising", used in the description, should not be interpreted as being restricted to the means listed thereafter; it does not exclude other elements or steps. It is thus to be interpreted as specifying the presence of the stated features, steps or components as referred to, but does not preclude the presence or addition of one or more other features, steps or components, or groups thereof. Thus, the scope of the expression should be commensurate to the disclosure and defined by the claims.

[0105] Finally, it should be noted that the language used in the specification has been principally selected for readability and instructional purposes and can not have been selected to delineate or circumscribe the subject matter disclosed herein. Accordingly, the disclosure of the present subject matter is intended to be illustrative, but not limiting, of the subject matter.

Claims

1. A terminal payment method, characterized in that, include: Obtain information on multiple completed payment software programs and pending payment software programs for the user; Based on the user's information on multiple completed payment software, determine the user login information and payment verification information for each completed payment software; Based on the user login information of each completed payment software and the information of the software to be paid, generate various levels of pre-login information for the software to be paid. The system displays various levels of pre-login information for the software to be paid to the user, and obtains and performs pre-login based on the login information of the software selected by the user. The user trust level and payment security index of the software to be paid are determined based on the login information of the software selected by the user. Based on the user's trust level in the software to be paid, the payment security index of the software to be paid, and the information of the software to be paid, determine the highest level of security verification complexity acceptable to the user and the lowest level of convenience verification complexity acceptable to the user. Based on the highest security verification complexity level acceptable to the user and the lowest convenience verification complexity level acceptable to the user, multiple preliminary payment verification schemes for the software to be paid are generated. The target payment verification scheme for the software to be paid is determined based on multiple preliminary payment verification schemes. This determination includes: A payment verification graph is constructed, comprising multiple nodes and multiple edges between them. The multiple nodes include multiple preliminary payment verification nodes and multiple completed payment software nodes. Each preliminary payment verification node establishes an edge with each completed payment software node. The node feature of each preliminary payment verification node is each preliminary payment verification scheme of the software to be paid. The node feature of each completed payment software node is the payment verification information of each completed payment software. The edges between the preliminary payment verification nodes and the completed payment software nodes represent the similarity between the information of the software to be paid and the information of the completed payment software, and the similarity between the login information of the software to be paid and the user login information of the completed payment software. The target payment verification scheme for the software to be paid is determined by processing the payment verification graph using a graph neural network. The target payment verification scheme for the software to be paid is used to verify the payment of the software to be paid.

2. The terminal payment method as described in claim 1, characterized in that, The process of generating multiple preliminary payment verification schemes for the software to be paid based on the highest acceptable level of security verification complexity and the lowest acceptable level of convenience verification complexity includes: Based on the highest acceptable level of security verification complexity and the lowest acceptable level of convenience verification complexity, generative adversarial networks are used to generate multiple initial payment verification schemes for the software to be paid.

3. The terminal payment method as described in claim 1, characterized in that, The input of the graph neural network is the payment verification graph, and the output of the graph neural network is the target payment verification scheme of the software to be paid.

4. A terminal payment system, characterized in that, include: The acquisition module is used to acquire information on multiple completed payment software and pending payment software for the user. The payment information determination module is used to determine the user login information and payment verification information of each completed payment software based on the user's multiple completed payment software information; The pre-login information determination module is used to generate various levels of pre-login information for the software to be paid based on the user login information of each completed payment software and the information of the software to be paid. The pre-login module is used to display various levels of pre-login information of the software to be paid to the user, and to obtain and perform pre-login based on the login information of the software to be paid selected by the user. The trust security index determination module is used to determine the user trust level and the payment security index of the software to be paid based on the login information of the software selected by the user. The complexity level determination module is used to determine the highest acceptable security verification complexity level and the lowest acceptable convenience verification complexity level for the user based on the user's trust level in the software to be paid, the payment security index of the software to be paid, and the information of the software to be paid. The initial scheme generation module is used to generate multiple initial payment verification schemes for the software to be paid based on the highest security verification complexity level acceptable to the user and the lowest convenience verification complexity level acceptable to the user. The target scheme determination module is used to determine a target payment verification scheme for the software to be paid based on multiple preliminary payment verification schemes. The target scheme determination module is further used for: A payment verification graph is constructed, comprising multiple nodes and multiple edges between them. The multiple nodes include multiple preliminary payment verification nodes and multiple completed payment software nodes. Each preliminary payment verification node establishes an edge with each completed payment software node. The node feature of each preliminary payment verification node is each preliminary payment verification scheme of the software to be paid. The node feature of each completed payment software node is the payment verification information of each completed payment software. The edges between the preliminary payment verification nodes and the completed payment software nodes represent the similarity between the information of the software to be paid and the information of the completed payment software, and the similarity between the login information of the software to be paid and the user login information of the completed payment software. The target payment verification scheme for the software to be paid is determined by processing the payment verification graph using a graph neural network. The payment verification module is used to perform payment verification on the software to be paid based on the target payment verification scheme of the software to be paid.

5. The terminal payment system as described in claim 4, characterized in that, The preliminary selection scheme generation module is also used for: Based on the highest acceptable level of security verification complexity and the lowest acceptable level of convenience verification complexity, generative adversarial networks are used to generate multiple initial payment verification schemes for the software to be paid.

6. The terminal payment system as described in claim 4, characterized in that, The input of the graph neural network is the payment verification graph, and the output of the graph neural network is the target payment verification scheme of the software to be paid.

7. An electronic device, characterized in that, include: processor; Memory; And a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement the terminal payment method as described in any one of claims 1 to 3.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the terminal payment method as described in any one of claims 1 to 3.

Images