Third-party payment fund flow security tracing method and system

Abstract

The present application relates to a third-party payment fund flow security tracing method and system, belonging to the field of electronic digital data processing, more specifically to the field of internet security services, the method comprising: using a transaction state intelligent judgment model to intelligently judge whether a target third-party payee account currently belongs to a secure payee account according to fund flow analysis data of the target third-party payee account; and when it does not belong, transferring it to a suspicious third-party payee account observation sequence. Through the present application, the technical problem of the existing technology that it is difficult to perform real-time security tracing on the past fund flow of a third-party payee account, resulting in security vulnerabilities in third-party payments, can be solved by using a specifically designed fund flow security tracing mechanism to obtain past fund flow analysis data of a target third-party payee account, and then intelligently judging whether it currently belongs to a secure payee account.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of electric digital data processing, more particularly to the field of Internet security services, and especially to a third-party payment fund flow security tracing method and system. BACKGROUND

[0002] ‌The related technology of electric digital data processing can be used in Internet security services to improve the management level of Internet security services, especially the management level of Internet financial payments. At present, the third-party payment method has become the main payment channel in Internet financial payments. The third-party payment method refers to a payment mode in which a third-party institution temporarily stores transaction funds during the transaction process of the buyer and the seller, and completes the transfer after confirming the delivery of the goods. Its core function is to provide guaranteed payment for both parties. The third-party guaranteed payment supervises the transaction process as an intermediary, establishes a credit system, and changes the situation of information asymmetry and mutual distrust between customers, which is conducive to achieving a win-win situation for both buyers and sellers.

[0003] Although the third-party payment method not only overcomes the payment bottleneck in online transactions, but also greatly improves the success rate of online transactions, especially between individuals, it also provides convenience for network fund fraud. In the case of fraud on the third-party payment platform, the fund flow presents the characteristics of multiple channels and multiple steps. The funds obtained by fraud are first scattered to multiple accounts to confuse the situation, and then whitewashed through false transactions and other means to make them appear legal. Then it may flow to overseas accounts or be used to purchase virtual currencies, etc. These means are taken by fraudsters to evade legal sanctions and fund recovery, increasing the difficulty of tracing fund flow, and making it difficult to judge suspicious transactions during the third-party payment transaction process.

[0004] For example, Chinese invention patent publication CN110363522A proposes a secure payment method, device and computer equipment, which includes a secure payment method and device on the mobile terminal side, a secure payment method and device on the server side, and a computer equipment. The application is applied to mobile payment of multi-card mobile terminals. In the security verification process of the payment process, whether the order of the verification information sent by each SIM card, the device code, etc. is correct needs to be considered to improve the complexity of verification and thus improve the security of mobile payment. For example, even if the payment account and its corresponding password are obtained by a third party, when the third party uses the payment account and the corresponding password to make a payment or transfer funds, the third party can only provide the correct payment account and password, but cannot provide the verification information sent by the specified multiple SIM cards in the specified order carrying the correct device code. Therefore, the third party still cannot complete the payment or fund transfer, thereby improving the payment security of the mobile terminal.

[0005] For example, Chinese invention patent publication CN115187230A proposes a method and system for combining the account settlement of a third-party payment platform and the reconciliation, which includes the following steps: S1. Generate a to-be-settled file and transmit it to the third-party payment platform; S2. The third-party payment platform receives the to-be-settled file, performs the account settlement task after passing the signature verification, and generates an account settlement result file; S3. Obtain the account settlement result file generated by the third-party payment platform, and compare it with the to-be-settled file after passing the signature verification; if the comparison is successful, output the reconciliation result, and if the comparison fails, enter the account settlement exception order processing. The invention can effectively improve the financial reconciliation efficiency of the e-commerce platform, realize the combination of the fund flow and the information flow, and the transaction funds flow within the third-party payment platform from payment to account settlement. The whole process is safe and compliant, avoiding the "two clear" risk, effectively preventing the files transmitted between the comprehensive reconciliation engine module and the third-party payment platform module from being maliciously intercepted or information tampered, and having high safety performance.

[0006] However, the above technical solutions in the prior art only involve a transaction security judgment mechanism for applying the transaction data to perform security verification in the third-party payment transaction process, and cannot perform fund flow security tracing on each past collection fund of the third-party collection account when initiating the third-party payment transaction, for example, tracing the number of each collection fund diverted transaction and the amount of each diversion transaction, whether there is a virtual currency diversion transaction, and whether there is an overseas diversion transaction. Naturally, it is impossible to make a live judgment on whether the third-party collection account is suspicious and subsequent fund security protection based on the fund flow security tracing result, resulting in a probability of successful network fraud in the transaction process of the third-party payment transaction. SUMMARY

[0007] To solve the technical problems in the prior art, the present application provides a third-party payment fund flow security tracing method and system, which can perform fund flow security tracing on each past collection fund of the target third-party collection account of the current third-party payment transaction, and use a transaction state intelligent judgment model designed for the structure of the third-party transaction platform to intelligently judge whether the current third-party payment transaction is a safe collection account, and perform security protection on the current third-party payment transaction based on the intelligent judgment result, thereby reducing the probability of successful network fraud in the transaction process of the third-party payment transaction.

[0008] According to a first aspect of the present application, a third-party payment fund flow security tracing method is provided, which includes:

[0009] The third-party collection account of the current third-party payment transaction just initiated by the set third-party transaction platform is taken as the target third-party collection account, and a plurality of account association information of the target third-party collection account is obtained.

[0010] The target third-party payment account participates in each third-party payment transaction as a third-party payment account, and each piece of fund flow data corresponding to each third-party payment transaction is used as fund flow analysis data of the target third-party payment account. The fund flow data corresponding to each third-party payment transaction includes the number of sub-flow transactions initiated by the target third-party payment account as a payment account within a set time length after the third-party payment transaction, and each piece of amount data, virtual currency transaction initiation identifier, and overseas transaction initiation identifier corresponding to each sub-flow transaction;

[0011] The transaction state intelligent judgment model is used to intelligently judge whether the target third-party payment account currently belongs to a safe payment account according to a plurality of safety-related parameters of a third-party payment account of a current third-party payment transaction, a set time length, a plurality of account-related information of the target third-party payment account, and fund flow analysis data of the target third-party payment account.

[0012] When it is intelligently judged that the target third-party payment account currently does not belong to a safe payment account, the target third-party payment account is transferred to a suspicious third-party payment account observation sequence, and the current third-party payment transaction is judged as a suspicious third-party payment transaction and is refused to be executed.

[0013] The transaction state intelligent judgment model is a radial basis neural network after a plurality of learning actions.

[0014] According to a second aspect of the present application, a third-party payment fund flow security tracing system is provided, which comprises a memory and a plurality of processors. The memory stores a computer program configured to be executed by the plurality of processors to complete the following steps:

[0015] A third-party payment account of a current third-party payment transaction just initiated by a set third-party transaction platform is used as a target third-party payment account, and a plurality of account-related information of the target third-party payment account is obtained.

[0016] The target third-party payment account participates in each third-party payment transaction as a third-party payment account, and each piece of fund flow data corresponding to each third-party payment transaction is used as fund flow analysis data of the target third-party payment account. The fund flow data corresponding to each third-party payment transaction includes the number of sub-flow transactions initiated by the target third-party payment account as a payment account within a set time length after the third-party payment transaction, and each piece of amount data, virtual currency transaction initiation identifier, and overseas transaction initiation identifier corresponding to each sub-flow transaction;

[0017] The transaction state intelligent judgment model is used to intelligently judge whether the target third-party payment account currently belongs to a safe account according to the plurality of safety-related parameters of the third-party payment account, the set time length, the plurality of account-related information of the target third-party payment account, and the fund flow analysis data of the target third-party payment account.

[0018] When it is intelligently judged that the target third-party payment account currently does not belong to a safe account, the target third-party payment account is transferred to a suspicious third-party payment account observation sequence, and the current third-party payment transaction is judged as a suspicious third-party payment transaction to be rejected.

[0019] The transaction state intelligent judgment model is a radial basis neural network after a plurality of learning actions.

[0020] According to a third aspect of the present application, a third-party payment fund flow security tracing system is provided, which comprises:

[0021] The third-party payment account of a current third-party payment transaction just initiated by a set third-party transaction platform is taken as a target third-party payment account, and a plurality of account-related information of the target third-party payment account is obtained.

[0022] The fund flow data corresponding to each third-party payment transaction in which the target third-party payment account participated in the past is taken as fund flow analysis data of the target third-party payment account, and the fund flow data corresponding to each third-party payment transaction includes the number of sub-transaction initiated by the target third-party payment account as a payment account within a set time length after the third-party payment transaction, and the amount data, virtual currency transaction initiation identifier and overseas transaction initiation identifier corresponding to each sub-transaction.

[0023] The transaction state intelligent judgment model is used to intelligently judge whether the target third-party payment account currently belongs to a safe account according to the plurality of safety-related parameters of the third-party payment account, the set time length, the plurality of account-related information of the target third-party payment account, and the fund flow analysis data of the target third-party payment account.

[0024] When it is intelligently judged that the target third-party payment account currently does not belong to a safe account, the target third-party payment account is transferred to a suspicious third-party payment account observation sequence, and the current third-party payment transaction is judged as a suspicious third-party payment transaction to be rejected.

[0025] The transaction state intelligent judgment model is a radial basis neural network after a plurality of learning actions.

[0026] Compared with the prior art, the present application has at least the following five key points:

[0027] The first place: the past each payment fund of the target third-party collection account of the current third-party payment transaction is executed respectively to obtain the fund flow analysis data of the target third-party collection account, and the fund flow security traceability result of each payment fund is the number of diversion transactions of the payment fund, the amount of each diversion transaction, the virtual currency transaction initiation identifier indicating whether the payment fund has subsequent virtual currency diversion transaction, and the overseas transaction initiation identifier indicating whether the payment fund has subsequent overseas diversion transaction, thereby providing key basic data for the subsequent intelligent judgment of whether the target third-party collection account currently belongs to a safe collection account;

[0028] The second place: when the intelligent judgment of the target third-party collection account currently does not belong to a safe collection account, the target third-party collection account is transferred to a suspicious third-party collection account observation sequence, the current third-party payment transaction is judged as a suspicious third-party payment transaction and is refused to be executed, a collection account replacement request for replacing the third-party collection account of the current third-party payment transaction is initiated to the third-party payment account of the current third-party payment transaction, and when the intelligent judgment of the target third-party collection account currently belongs to a safe collection account, the target third-party collection account is transferred to a safe third-party collection account observation sequence, and the current third-party payment transaction is allowed to be executed, thereby completing the automatic and intelligent management of the current third-party payment transaction based on the fund flow security traceability result of the third-party payment;

[0029] The third place: a transaction state intelligent judgment model with different customized structures is designed for different third-party transaction platforms, which is used to execute the intelligent judgment of whether the target third-party collection account currently belongs to a safe collection account. Specifically, the transaction state intelligent judgment model is a radial basis neural network after a plurality of learning actions, and the number of learning actions of the radial basis neural network is the same as the change trend of the average value of the number of third-party transactions of the set third-party transaction platform per day.

[0030] The fourth place: in addition to the target third-party collection account fund flow direction analysis data, a plurality of basic data are additionally introduced to ensure the comprehensiveness and sufficiency of the basic data for executing the judgment, the plurality of basic data additionally introduced including a plurality of safety-related parameters of the third-party payer account of the current third-party payment transaction, a set time length, a plurality of account association information of the target third-party collection account, wherein the plurality of account association information of the target third-party collection account includes a registration duration of the target third-party collection account, a number of times of participating in the third-party payment transaction, a suspicious number proportion of a transaction counterpart, and a number of past complaints, and the plurality of safety-related parameters of the third-party payer account of the current third-party payment transaction are a gender identification of an account belonging person of the third-party payer account of the current third-party payment transaction, age data, a number of times of participating in the third-party payment transaction per day, and a transaction amount average of the third-party payment transaction per day;

[0031] The fifth place: in each learning action performed on the radial basis neural network, a judgment result of whether a certain third-party collection account of a third-party transaction platform at a certain historical moment belongs to a safe collection account is taken as a single output content of the radial basis neural network, a plurality of safety-related parameters of a third-party payer account of a third-party payment transaction participated by the certain third-party collection account at the certain historical moment, a set time length, a plurality of account association information of the certain third-party collection account, and fund flow direction analysis data of the certain third-party collection account are taken as a plurality of input contents of the radial basis neural network, and the learning action is completed, so that the learning effect of each learning action of the radial basis neural network is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0032] The embodiments of the present application will be described below in conjunction with the accompanying drawings, in which:

[0033] Figure 1 It is a working scene schematic diagram of the third-party payment fund flow direction safety tracing method and system according to the present application.

[0034] Figure 2 It is a step flowchart of the third-party payment fund flow direction safety tracing method according to embodiment 1 of the present application.

[0035] Figure 3 It is a step flowchart of the third-party payment fund flow direction safety tracing method according to embodiment 2 of the present application.

[0036] Figure 4 It is a step flowchart of the third-party payment fund flow direction safety tracing method according to embodiment 3 of the present application.

[0037] Figure 5 It is a step flowchart of the third-party payment fund flow direction safety tracing method according to embodiment 4 of the present application.

[0038] Figure 6 A flow chart of steps of a third-party payment fund flow direction security tracing method according to an embodiment 5 of the present application.

[0039] Figure 7 A structural schematic diagram of a third-party payment fund flow direction security tracing system according to an embodiment 6 of the present application.

[0040] Figure 8 A structural schematic diagram of a third-party payment fund flow direction security tracing system according to an embodiment 7 of the present application. DETAILED DESCRIPTION

[0041] As Figure 1 shown, a working scenario schematic diagram of a third-party payment fund flow direction security tracing method and system according to the present application is given. The electronic digital data processing of the present application relates more particularly to the field of internet security services.

[0042] Currently, the fund flow direction in a third-party payment platform fraud case is often relatively complex and has certain characteristics. Network fraud personnel usually try every means to quickly transfer the fraud funds in order to evade tracking and recovery.

[0043] After the network fraud personnel frauds funds through the collection account of the third-party payment platform, the funds may be quickly distributed to multiple accounts, which may include other personal bank accounts controlled by them, third-party payment accounts, etc. By dispersing the funds, the difficulty of investigation is increased, for example, the funds may be dispersed to a dozen or even dozens of different accounts first. Then, the funds may be whitewashed through some seemingly normal transactions, for example, they may flow to some overseas accounts or purchase virtual currencies and other assets that are difficult to track, making the fund flow direction more confusing.

[0044] Therefore, when the payment account of the third-party payment platform initiates payment, it is necessary to determine in real time whether the collection account belongs to a suspicious account to avoid successful fraud. Compared with the conventional collection account suspicious judgment mechanism in the prior art, the present application proposes a suspicious judgment mechanism based on the past fund flow direction security tracking mechanism of the collection account, thereby ensuring the reliability of the judgment result and reducing the probability of successful fraud.

[0045] To achieve the above technical effects of the present application, the specific technical process of the present application is as follows:

[0046] Technical process A: when the third-party payment account initiates the current third-party payment transaction through the third-party payment transaction platform, the past each collection fund of the third-party collection account of the current third-party payment transaction, i.e. the target third-party collection account, is executed for fund flow direction security tracing, so as to obtain fund flow direction analysis data of the target third-party collection account,Figure 1 the target third-party payee account, including fund flow direction analysis data of the target third-party payee account;

[0047] Specifically, the fund flow direction security traceability result of each payment fund is the number of payment diversion transactions of the payment fund and the amount of each payment diversion transaction, a virtual currency transaction initiation identifier indicating whether the payment fund is followed by a virtual currency payment diversion transaction, and an overseas transaction initiation identifier indicating whether the payment fund is followed by an overseas payment diversion transaction.

[0048] In this way, key basic data is provided for subsequent intelligent judgment of whether the target third-party payee account currently belongs to a secure payee account.

[0049] Technical process B: to perform intelligent judgment of whether the target third-party payee account currently belongs to a secure payee account, a transaction state intelligent judgment model with a customized structure is designed, as shown in Figure 1 , wherein different third-party transaction platforms have different transaction state intelligent judgment models with customized structures.

[0050] Specifically, the customized structure design of the transaction state intelligent judgment model mainly reflects in the following aspects:

[0051] Aspect one: the transaction state intelligent judgment model is a radial basis neural network after multiple learning actions, that is, the transaction state intelligent judgment model uses the network architecture of a radial basis neural network.

[0052] Aspect two: the number of learning actions of the radial basis neural network is the same as the change trend of the average number of third-party transactions per day of the set third-party transaction platform.

[0053] In this way, the greater the average number of third-party transactions per day of the set third-party transaction platform, the more the number of learning actions, thereby designing different transaction state intelligent judgment models with customized structures for different third-party transaction platforms.

[0054] Aspect three: in each learning action performed on the radial basis neural network, the judgment result of whether a certain third-party payee account of the set third-party transaction platform has been identified as a secure payee account at a certain historical time is taken as a single output content of the radial basis neural network, and multiple security-related parameters of a third-party payee account participating in a third-party payment transaction of the certain third-party payee account at the certain historical time, a set time length, multiple account-related information of the certain third-party payee account, and fund flow direction analysis data of the certain third-party payee account are taken as multiple input contents of the radial basis neural network, thereby completing the learning action and ensuring the learning effect of each learning action of the radial basis neural network.

[0055] In this way, the effectiveness and stability of the intelligent judgment result are ensured through the customized structural design of the above aspects.

[0056] Technical process C: In addition to the fund flow direction analysis data of the target third-party collection account, a plurality of basic data are additionally introduced to ensure the comprehensiveness and sufficiency of the basic data used for performing the judgment, so that Figure 1 the basic data used for intelligent judgment are obtained.

[0057] Specifically, the plurality of basic data additionally introduced include a plurality of security-related parameters of the third-party payer account of the current third-party payment transaction, a set time length, and a plurality of account association information of the target third-party collection account.

[0058] Further specifically, the plurality of account association information of the target third-party collection account include a registration duration of the target third-party collection account, a number of times of participating in the third-party payment transaction, a suspicious proportion of a transaction counterpart, and a number of past complaints, and the plurality of security-related parameters of the third-party payer account of the current third-party payment transaction are a gender identification of an account belonging person of the third-party payer account of the current third-party payment transaction, age data, a number of times of participating in the third-party payment transaction per day, and a transaction amount average per day of participating in the third-party payment transaction.

[0059] In this way, the effectiveness and stability of the intelligent judgment result are further ensured through the comprehensive and sufficient screening of the above basic data.

[0060] Technical process D: using the transaction state intelligent judgment model customized and structurally designed for the set third-party transaction platform according to the comprehensive and sufficient screening of the basic data in technical process A and technical process B, completing the intelligent judgment of whether the target third-party collection account is a safe collection account at present, that is, obtaining Figure 1 the intelligent judgment result.

[0061] In this way, in the processing process of each third-party payment transaction, based on the past fund flow direction safety tracing result of the payment counterpart, that is, the third-party collection account, the transaction state intelligent judgment model customized and structurally designed for the set third-party transaction platform is used to intelligently judge whether the payment counterpart, that is, the third-party collection account, is a trusted account, so as to provide key information for subsequent third-party payment transaction safety management.

[0062] Technical process E: according to the intelligent judgment result of technical process D, a safety management strategy for the target third-party collection account is formulated, as shown in Figure 1 to ensure the safety of the funds of the current third-party payer account initiating the third-party payment transaction.

[0063] Specifically, when the target third-party payment account is intelligently judged as not currently belonging to a safe payment account, the target third-party payment account is transferred to a suspicious third-party payment account observation sequence, the current third-party payment transaction is judged as a suspicious third-party payment transaction, execution of the current third-party payment transaction is refused, and a payment account replacement request for replacing the third-party payment account of the current third-party payment transaction is initiated to a third-party payment account of the current third-party payment transaction;

[0064] Specifically, when the target third-party payment account is intelligently judged as currently belonging to a safe payment account, the target third-party payment account is transferred to a safe third-party payment account observation sequence, and execution of the current third-party payment transaction is allowed;

[0065] Therefore, through the cooperative operation of the above five technical processes, the automatic and intelligent management of the current third-party payment transaction based on the safe tracing result of the third-party payment fund flow direction is completed.

[0066] The key points of the present application are: targeted analysis of the fund flow direction analysis data of the target third-party payment account, customized structural design of different transaction state intelligent judgment models of different third-party transaction platforms, comprehensive and sufficient screening of various basic data for intelligent judgment, and matching formulation of the safety management strategy for the target third-party payment account based on the intelligent judgment result.

[0067] In the following, the third-party payment fund flow direction safety tracing method and system of the present application will be described in detail in the form of an embodiment.

[0068] Embodiment 1

[0069] Figure 2 The step flow chart of the third-party payment fund flow direction safety tracing method according to Embodiment 1 of the present application is shown.

[0070] As shown in Figure 2 , the third-party payment fund flow direction safety tracing method includes the following specific steps:

[0071] Step S21: taking the third-party payment account of the current third-party payment transaction just initiated by the set third-party transaction platform as a target third-party payment account, and obtaining multiple account association information of the target third-party payment account;

[0072] For example, taking the third-party payment account of the current third-party payment transaction just initiated by the set third-party transaction platform as a target third-party payment account, and obtaining multiple account association information of the target third-party payment account includes: the set third-party transaction platform is one of Venmo, Cash App, Alipay or WeChat Pay;

[0073] Step S22: taking each piece of fund flow data corresponding to each piece of third-party payment transaction in which the target third-party payee account participates as a third-party payee account as fund flow analysis data of the target third-party payee account, each piece of fund flow data corresponding to each piece of third-party payment transaction including the number of sub-flow transactions initiated by the target third-party payee account as a payer within a set time length after the third-party payment transaction, and each piece of sub-flow transaction corresponding to each piece of amount data, virtual currency transaction initiation identifier and cross-border transaction initiation identifier;

[0074] For example, in determining the target third-party payee account participates in each piece of third-party payment transaction as a third-party payee account, each piece of third-party payment transaction can be obtained by using a time-limited manner, for example, each piece of third-party payment transaction in which the target third-party payee account participates as a third-party payee account within two months is taken as each piece of third-party payment transaction in which the target third-party payee account participates as a third-party payee account;

[0075] Here, the above exemplary time-limited manner and the set time length belong to different concepts, one is used to limit the target third-party payee account participates in each piece of third-party payment transaction as a third-party payee account, and the other is used to determine each piece of sub-flow transaction corresponding to each piece of third-party payment transaction in which the target third-party payee account participates as a third-party payee account;

[0076] Specifically, different binary values can be selected to represent the virtual currency transaction initiation identifier to represent whether a virtual currency sub-flow transaction occurs, and the same, different binary values can be selected to represent the cross-border transaction initiation identifier to represent whether a cross-border sub-flow transaction occurs;

[0077] Step S23: using a transaction state intelligent judgment model to intelligently judge whether the target third-party payee account currently belongs to a safe payee account according to the plurality of safety-related parameters of the third-party payer account of the current third-party payment transaction, the set time length, the plurality of account-related information of the target third-party payee account and the fund flow analysis data of the target third-party payee account;

[0078] For example, the set time length is used to determine each piece of sub-flow transaction in which the target third-party payee account exists as a third-party payer within the set time length after each piece of third-party payment transaction in which the target third-party payee account participates as a third-party payee, and the value of the set time length is relatively short, because network financial fraudsters are eager to sub-flow the amount of successful fraud, for example, the value can be 8 hours, 12 hours or 24 hours;

[0079] Step S24: when the intelligent judgment result shows that the target third-party payment account does not belong to the safe payment account, the target third-party payment account is transferred to the suspicious third-party payment account observation sequence, and the current third-party payment transaction is judged as a suspicious third-party payment transaction and is rejected for execution;

[0080] In this way, the matching establishment of the safety management strategy for the target third-party payment account based on the intelligent judgment result is realized.

[0081] The transaction state intelligent judgment model is a radial basis neural network after multiple learning actions.

[0082] The transaction state intelligent judgment model is a radial basis neural network after multiple learning actions, including that the number of learning actions is the same as the change trend of the average number of third-party transactions per day of the set third-party transaction platform.

[0083] Specifically, when the change trend of the average number of third-party transactions per day of the set third-party transaction platform is a certain nonlinear change function, the change trend of the corresponding number of learning actions has the same change curve as the change curve corresponding to the certain nonlinear change function.

[0084] For example, when the average number of third-party transactions per day of the set third-party transaction platform is 5, the number of selected learning actions is 500, when the average number of third-party transactions per day of the set third-party transaction platform is 6, the number of selected learning actions is 600, when the average number of third-party transactions per day of the set third-party transaction platform is 7, the number of selected learning actions is 700, when the average number of third-party transactions per day of the set third-party transaction platform is 8, the number of selected learning actions is 800, and so on.

[0085] In each learning action performed on the radial basis neural network, the judgment result of whether a certain third-party payment account of the set third-party transaction platform has been identified as a safe payment account at a certain historical time is taken as a single output content of the radial basis neural network, and a plurality of safety-related parameters of a third-party payment account participating in a third-party payment transaction of the certain third-party payment account at the certain historical time, a set time length, a plurality of account-related information of the certain third-party payment account, and fund flow analysis data of the certain third-party payment account are taken as a plurality of input contents of the radial basis neural network, to complete the learning action.

[0086] The plurality of account-related information of the target third-party payment account includes the registration duration of the target third-party payment account, the number of times of participating in third-party payment transactions, the suspicious proportion of transaction counterparts, and the number of past complaints.

[0087] For example, a plurality of different information capturing units can be selected to obtain the registration duration of the target third-party payment account, the number of times of participating in the third-party payment transaction, the suspicious account proportion of the transaction counterpart, and the number of past complaints, respectively.

[0088] Among them, the third-party payment account of the current third-party payment transaction initiated by the set third-party transaction platform is taken as the target third-party payment account, and the registration duration of the target third-party payment account, the number of times of participating in the third-party payment transaction, the suspicious account proportion of the transaction counterpart, and the number of past complaints are obtained as the multiple account association information of the target third-party payment account. The suspicious account proportion of the transaction counterpart of the target third-party payment account is obtained by taking the number of third-party payment accounts judged as suspicious accounts in each transaction counterpart corresponding to each third-party payment transaction participated by the target third-party payment account.

[0089] For example, the suspicious account proportion of the transaction counterpart of the target third-party payment account is obtained by taking the number of third-party payment accounts judged as suspicious accounts in each transaction counterpart corresponding to each third-party payment transaction participated by the target third-party payment account. For example, the number of each transaction counterpart corresponding to each third-party payment transaction participated by the target third-party payment account is 20, and the number of third-party payment accounts judged as suspicious accounts is 5. The suspicious account proportion of the transaction counterpart of the target third-party payment account is 1 / 4, that is, 1 / 4 is taken as the suspicious account proportion of the transaction counterpart of the target third-party payment account.

[0090] Among them, the multiple security-related parameters of the third-party payment account of the current third-party payment transaction are the gender identification, age data, number of times of participating in the third-party payment transaction per day, and average transaction amount of the third-party payment account of the current third-party payment transaction.

[0091] For example, a plurality of numerical value analysis units can be selected to analyze the gender identification, age data, number of times of participating in the third-party payment transaction per day, and average transaction amount of the third-party payment account of the current third-party payment transaction, respectively.

[0092] and wherein the fund flow direction data corresponding to each third-party payment transaction comprises the number of sub-transaction initiated by the target third-party payment account as a payee account within a set time length after the third-party payment transaction, and each part of the amount data, the virtual currency transaction initiation identifier and the cross-border transaction initiation identifier corresponding to each sub-transaction respectively, the fund flow direction data corresponding to each third-party payment transaction comprises the number of sub-transaction initiated by the target third-party payment account as a payee account within a set time length after the third-party payment transaction, and each part of the amount data, the virtual currency transaction initiation identifier and the cross-border transaction initiation identifier corresponding to each sub-transaction respectively, the virtual currency transaction initiation identifier and the cross-border transaction initiation identifier respectively indicating whether the target third-party payment account initiates purchase of virtual currency and initiates cross-border transaction as a payee account within a set time length after the third-party payment transaction.

[0093] Embodiment 2

[0094] Figure 3 The step flow chart of the third-party payment fund flow direction security tracing method according to Embodiment 2 of the present application is shown.

[0095] As shown in Figure 3 , unlike the embodiment in Figure 2 , when it is intelligently judged that the target third-party payment account does not currently belong to a secure payment account, the target third-party payment account is transferred to a suspicious third-party payment account observation sequence, and after the current third-party payment transaction is judged to be a suspicious third-party payment transaction and the execution is refused, i.e. after step S24, the method further comprises:

[0096] Step S25: after judging the current third-party payment transaction to be a suspicious third-party payment transaction, initiating a payee account replacement request for replacing the third-party payment account of the current third-party payment transaction to the third-party payee account of the current third-party payment transaction;

[0097] For example, after judging the current third-party payment transaction to be a suspicious third-party payment transaction, initiating a payee account replacement request for replacing the third-party payment account of the current third-party payment transaction to the third-party payee account of the current third-party payment transaction comprises displaying the payee account replacement request on the display screen of the transaction initiation terminal of the current third-party payment transaction.

[0098] Embodiment 3

[0099] Figure 4 The step flow chart of the third-party payment fund flow direction security tracing method according to Embodiment 3 of the present application is shown.

[0100] As shown in Figure 4 , unlike the embodiment in Figure 2Unlike the previous implementation, when the intelligent system determines that the target third-party receiving account is not currently a secure receiving account, the method moves the target third-party receiving account to the suspicious third-party receiving account observation sequence, and rejects the current third-party payment transaction as a suspicious third-party payment transaction, i.e., after step S24, the method further includes:

[0101] Step S26: When the intelligent system determines that the target third-party receiving account is currently a secure receiving account, the target third-party receiving account is transferred to the secure third-party receiving account observation sequence, and the current third-party payment transaction is allowed to be executed;

[0102] For example, when the intelligent system determines that the target third-party receiving account is currently a secure receiving account, the target third-party receiving account is transferred to the secure third-party receiving account observation sequence, and the execution of the current third-party payment transaction is allowed. This includes: the secure third-party receiving account observation sequence stores the account information corresponding to each secure receiving account in the order of transfer.

[0103] Example 4

[0104] Figure 5 This is a flowchart illustrating the steps of a third-party payment fund flow security tracing method according to Embodiment 4 of the present invention.

[0105] like Figure 5 As shown, with Figure 2 The implementation differs from the previous one. Instead of using the fund flow data corresponding to each third-party payment transaction in which the target third-party receiving account previously participated as a third-party receiving account as fund flow analysis data for the target third-party receiving account, the fund flow data corresponding to each third-party payment transaction includes the number of diversion transactions initiated by the target third-party receiving account as a payment account within a set time period after that third-party payment transaction, as well as the amount data corresponding to each diversion transaction, the virtual currency transaction initiation identifier, and the overseas transaction initiation identifier. That is, after step S22, the method further includes:

[0106] Step S27: Perform multiple learning actions on the radial basis neural network to obtain the radial basis neural network after multiple learning actions and output it as the intelligent judgment model for transaction status.

[0107] For example, numerical simulation mode can be used to complete the process of performing multiple learning actions on the radial basis neural network to obtain the radial basis neural network after multiple learning actions and use it as the output of the intelligent judgment model of transaction status for testing and simulation.

[0108] The process of obtaining the radial basis function neural network after multiple learning actions and using it as the output of the intelligent judgment model for trading status includes: using the various model parameters of the intelligent judgment model for trading status to complete the model representation of the intelligent judgment model for trading status.

[0109] Example 5

[0110] Figure 6 The flowchart illustrates the steps of a third-party payment fund flow security tracing method according to Embodiment 5 of the present invention.

[0111] like Figure 6 As shown, with Figure 2 Unlike the previous implementation, when the intelligent system determines that the target third-party receiving account is not currently a secure receiving account, the method moves the target third-party receiving account to the suspicious third-party receiving account observation sequence, and rejects the current third-party payment transaction as a suspicious third-party payment transaction, i.e., after step S24, the method further includes:

[0112] Step S28: After binding the suspicious third-party receiving account observation sequence with the set third-party trading platform, send it together to the remote fund supervision server;

[0113] Specifically, a blockchain service node, a big data service node, or a cloud computing service node can be selected to implement the remote fund monitoring server;

[0114] The process of binding the suspicious third-party payment account observation sequence with the set third-party trading platform and sending them together to the remote fund supervision server includes: packaging each suspicious third-party payment account in the suspicious third-party payment account observation sequence and the platform code value of the set third-party trading platform together into a network data packet and then sending it to the remote fund supervision server.

[0115] Next, the various method embodiments of the present invention will be described in detail.

[0116] In the third-party payment fund flow security tracing method according to various method embodiments of the present invention:

[0117] The intelligent transaction status judgment model uses multiple security-related parameters of the third-party payment account in the current third-party payment transaction, the set time length, multiple account association information of the target third-party receiving account, and the fund flow analysis data of the target third-party receiving account to intelligently determine whether the target third-party receiving account is currently a safe receiving account. This includes: synchronously inputting multiple security-related parameters of the third-party payment account in the current third-party payment transaction, the set time length, multiple account association information of the target third-party receiving account, and the fund flow analysis data of the target third-party receiving account into the intelligent transaction status judgment model;

[0118] For example, an ASIC chip can be selected to synchronously input the plurality of security-related parameters of the third-party payment account of the current third-party payment transaction, the set time length, the plurality of account-related information of the target third-party payment account, and the fund flow analysis data of the target third-party payment account into the transaction state intelligent judgment model;

[0119] The transaction state intelligent judgment model intelligently judges whether the target third-party payment account currently belongs to a safe payment account according to the plurality of security-related parameters of the third-party payment account of the current third-party payment transaction, the set time length, the plurality of account-related information of the target third-party payment account, and the fund flow analysis data of the target third-party payment account, and the method further includes: executing the transaction state intelligent judgment model to obtain an account security identifier output by the transaction state intelligent judgment model, the account security identifier indicating whether the target third-party payment account currently belongs to a safe payment account;

[0120] The transaction state intelligent judgment model intelligently judges whether the target third-party payment account currently belongs to a safe payment account according to the plurality of security-related parameters of the third-party payment account of the current third-party payment transaction, the set time length, the plurality of account-related information of the target third-party payment account, and the fund flow analysis data of the target third-party payment account, and the method further includes: executing the transaction state intelligent judgment model to obtain an account security identifier output by the transaction state intelligent judgment model, the account security identifier indicating whether the target third-party payment account currently belongs to a safe payment account;

[0121] Specifically, the numerical normalization processing includes binary numerical conversion processing on the plurality of security-related parameters of the third-party payment account of the current third-party payment transaction, the set time length, the plurality of account-related information of the target third-party payment account, and the fund flow analysis data of the target third-party payment account;

[0122] The transaction state intelligent judgment model intelligently judges whether the target third-party payment account currently belongs to a safe payment account according to the plurality of security-related parameters of the third-party payment account of the current third-party payment transaction, the set time length, the plurality of account-related information of the target third-party payment account, and the fund flow analysis data of the target third-party payment account, and the method further includes: executing the transaction state intelligent judgment model to obtain an account security identifier output by the transaction state intelligent judgment model, the account security identifier indicating whether the target third-party payment account currently belongs to a safe payment account;

[0123] Specifically, the account security identifier indicating whether the target third-party receiving account is currently a secure receiving account is represented by a normalized numerical value, including: the account security identifier indicating whether the target third-party receiving account is currently a secure receiving account is represented by a binary numerical value after conversion.

[0124] Example 6

[0125] Figure 7 This is a schematic diagram of the structure of a third-party payment fund flow security tracking system according to Embodiment 6 of the present invention.

[0126] like Figure 7 As shown, the third-party payment fund flow security tracking system includes a memory and multiple processors. The memory stores a computer program, which is configured to be executed by the multiple processors to complete the following steps:

[0127] Step S21: Take the third-party receiving account of the current third-party payment transaction initiated by the third-party transaction platform as the target third-party receiving account, and obtain multiple account association information of the target third-party receiving account;

[0128] For example, the third-party payment account that was just initiated by the third-party transaction platform is used as the target third-party payment account. The target third-party payment account is obtained by obtaining multiple account association information of the target third-party payment account, including: setting the third-party transaction platform as one of Venmo, Cash App, Alipay or WeChat Pay.

[0129] Step S22: Use the fund flow data corresponding to each third-party payment transaction that the target third-party receiving account has participated in when it was acting as a third-party receiving account as the fund flow analysis data of the target third-party receiving account. The fund flow data corresponding to each third-party payment transaction includes the number of diversion transactions initiated by the target third-party receiving account as a payment account within a set time period after the third-party payment transaction, as well as the amount data, virtual currency transaction initiation identifier and overseas transaction initiation identifier corresponding to each diversion transaction.

[0130] For example, when determining the third-party payment transactions that the target third-party payment account participated in when it was acting as a third-party payment account, the third-party payment transactions can be obtained by using a time-limited method. For example, the third-party payment transactions that the target third-party payment account participated in within the past two months when it was acting as a third-party payment account can be used as the third-party payment transactions that the target third-party payment account participated in when it was acting as a third-party payment account.

[0131] Here, the above exemplary time-limited manner and the set time length belong to different concepts, one is used to limit the target third-party payee account in the past participating in each third-party payment transaction as a third-party payee account, and the other is used to determine each corresponding diversion transaction for each third-party payment transaction in the past participating in;

[0132] Specifically, different binary values can be selected to represent the virtual currency transaction initiation identifier to represent whether a virtual currency diversion transaction has occurred, and the same, different binary values can be selected to represent the cross-border transaction initiation identifier to represent whether a cross-border diversion transaction has occurred.

[0133] Step S23: using a transaction state intelligent judgment model to intelligently judge whether the target third-party payee account currently belongs to a safe payee account according to the plurality of safety-related parameters of the third-party payer account of the current third-party payment transaction, the set time length, the plurality of account association information of the target third-party payee account and the fund flow direction analysis data of the target third-party payee account;

[0134] For example, the set time length is used to determine each diversion transaction of the target third-party payee account as a third-party payer account within the set time length after the completion of each third-party payment transaction in which the target third-party payee account has participated in the past, and the value of the set time length is relatively short, because network financial fraudsters are eager to divert the amount of successful fraud, for example, the value can be 8 hours, 12 hours or 24 hours.

[0135] Step S24: when intelligently judging that the target third-party payee account does not currently belong to a safe payee account, transferring the target third-party payee account to a suspicious third-party payee account observation sequence, and judging the current third-party payment transaction as a suspicious third-party payment transaction and refusing to execute;

[0136] In this way, the matching formulation of the security management strategy for the target third-party payee account based on the intelligent judgment result is realized.

[0137] The transaction state intelligent judgment model is a radial basis neural network after a plurality of learning actions.

[0138] The transaction state intelligent judgment model is a radial basis neural network after a plurality of learning actions, including that the number of learning actions is the same as the change trend of the average number of third-party transactions per day of the set third-party transaction platform.

[0139] Specifically, the change trend of the number of learning actions is the same as the change trend of the average number of third-party transactions per day of the third-party transaction platform.

[0140] For example, when the average number of third-party transactions per day of the third-party transaction platform is set to 5, the number of selected learning actions is 500, when the average number of third-party transactions per day of the third-party transaction platform is set to 6, the number of selected learning actions is 600, when the average number of third-party transactions per day of the third-party transaction platform is set to 7, the number of selected learning actions is 700, when the average number of third-party transactions per day of the third-party transaction platform is set to 8, the number of selected learning actions is 800, and so on.

[0141] In each learning action performed on the radial basis neural network, the judgment result of whether a certain third-party payment account of the third-party transaction platform at a certain historical time is determined to belong to a safe payment account is set as a single output content of the radial basis neural network, and a plurality of safety-related parameters of a third-party payment account participating in a third-party payment transaction at the certain historical time, a set time length, a plurality of account-related information of the certain third-party payment account, and fund flow analysis data of the certain third-party payment account are set as a plurality of input contents of the radial basis neural network, to complete the learning action.

[0142] In the method, the plurality of account-related information of the target third-party payment account includes a registration duration of the target third-party payment account, a number of times of participating in third-party payment transactions, a suspicious account proportion of a transaction counterpart, and a number of past complaints.

[0143] For example, a plurality of different information capturing units can be selected to respectively obtain the registration duration of the target third-party payment account, the number of times of participating in third-party payment transactions, the suspicious account proportion of the transaction counterpart, and the number of past complaints.

[0144] In the method, the plurality of account-related information of the target third-party payment account includes a registration duration of the target third-party payment account, a number of times of participating in third-party payment transactions, a suspicious account proportion of a transaction counterpart, and a number of past complaints.

[0145] For example, the percentage of third-party payment / receiving accounts identified as suspicious accounts in each transaction pair corresponding to each third-party payment transaction previously participated in by the target third-party receiving account is used as the percentage of suspicious accounts in the target third-party receiving account's transaction pair. This includes the following: if the number of each transaction pair corresponding to each third-party payment transaction previously participated in by the target third-party receiving account is 20, and the number of third-party payment / receiving accounts identified as suspicious accounts is 5, then the percentage of third-party payment / receiving accounts identified as suspicious accounts in each transaction pair corresponding to each third-party payment transaction previously participated in by the target third-party receiving account is 1 / 4, and 1 / 4 is output as the percentage of suspicious accounts in the target third-party receiving account's transaction pair.

[0146] Among them, the security-related parameters of the third-party payment account in the current third-party payment transaction are the gender identifier, age data, number of times participating in third-party payment transactions in a single day, and average transaction amount of the third-party payment account in a single day;

[0147] For example, multiple numerical parsing units can be selected to parse the gender identifier, age data, number of times a person participates in third-party payment transactions per day, and average transaction amount of a person participating in third-party payment transactions per day for the account holder of the current third-party payment transaction.

[0148] The fund flow data for each third-party payment transaction includes the number of diversion transactions initiated by the target third-party receiving account as the payment account within a set time period after the third-party payment transaction, as well as the amount data corresponding to each diversion transaction, virtual currency transaction initiation identifier, and overseas transaction initiation identifier. The data includes: the number of diversion transactions initiated by the target third-party receiving account as the payment account within a set time period after the third-party payment transaction, as well as the amount data corresponding to each diversion transaction, and virtual currency transaction initiation identifier and overseas transaction initiation identifier indicating whether the target third-party receiving account as the payment account initiated the purchase of virtual currency and whether it initiated an overseas transaction within a set time period after the third-party payment transaction.

[0149] like Figure 7 As shown, for example, S processors are given, where S is a natural number greater than or equal to 1.

[0150] Example 7

[0151] Figure 8 This is a schematic diagram of the structure of a third-party payment fund flow security tracking system according to Embodiment 7 of the present invention.

[0152] As Figure 8 shown, the third-party payment fund flow direction security tracing system comprises the following components:

[0153] An information capturing device is configured to capture a plurality of account-related information of a target third-party payee account, which is a third-party payee account of a current third-party payment transaction initiated via a designated third-party transaction platform.

[0154] For example, the target third-party payee account is a third-party payee account of a current third-party payment transaction initiated via a designated third-party transaction platform, and the plurality of account-related information of the target third-party payee account is captured, wherein the designated third-party transaction platform is one of Venmo, Cash App, Alipay, or WeChat Pay.

[0155] A security tracing device is configured to capture a plurality of fund flow direction data corresponding to each of a plurality of third-party payment transactions in which the target third-party payee account participated as a third-party payee account, as fund flow direction analysis data of the target third-party payee account, wherein the fund flow direction data corresponding to each of the plurality of third-party payment transactions comprises a number of sub-transaction initiated by the target third-party payee account as a payer account within a designated time length after the third-party payment transaction, and a plurality of amount data, a virtual currency transaction initiation identifier, and an overseas transaction initiation identifier corresponding to each of the sub-transaction.

[0156] For example, in determining the plurality of third-party payment transactions in which the target third-party payee account participated as a third-party payee account, a time-limited manner can be used to capture the plurality of third-party payment transactions, for example, the plurality of third-party payment transactions in which the target third-party payee account participated as a third-party payee account within two months are determined as the plurality of third-party payment transactions in which the target third-party payee account participated as a third-party payee account.

[0157] Here, the above-mentioned exemplary time-limited manner and the designated time length belong to different concepts, one is used to limit the plurality of third-party payment transactions in which the target third-party payee account participated as a third-party payee account, and the other is used to determine the corresponding sub-transaction for each of the plurality of third-party payment transactions.

[0158] Specifically, different binary values can be selected to represent whether a virtual currency sub-transaction has occurred, and different binary values can also be selected to represent whether an overseas sub-transaction has occurred.

[0159] The intelligent judgment device is connected with the information capturing device and the security tracing device, and is configured to intelligently judge, by using a transaction state intelligent judgment model, whether the target third-party payee account currently belongs to a safe payee account according to a plurality of safety-related parameters of a third-party payment account of the current third-party payment transaction, a set time length, a plurality of account-related information of the target third-party payee account, and fund flow analysis data of the target third-party payee account.

[0160] For example, the set time length is used to determine, for each third-party payment transaction in which the target third-party payee account has participated in the past, a plurality of diversion transactions in which the target third-party payee account acts as a third-party payment account within the set time length after the completion of the third-party payment transaction. The set time length has a relatively short value because network financial fraudsters are eager to divert the amount of successful fraud, for example, the value can be 8 hours, 12 hours or 24 hours.

[0161] The account processing device is connected with the intelligent judgment device, and is configured to, when the intelligent judgment device determines that the target third-party payee account currently does not belong to a safe payee account, transfer the target third-party payee account to a suspicious third-party payee account observation sequence, and determine that the current third-party payment transaction is a suspicious third-party payment transaction and refuse to execute the current third-party payment transaction.

[0162] In this way, the matching formulation of the security management strategy for the target third-party payee account based on the intelligent judgment result is realized.

[0163] The transaction state intelligent judgment model is a radial basis neural network after a plurality of learning actions.

[0164] The transaction state intelligent judgment model is a radial basis neural network after a plurality of learning actions, and the number of learning actions is the same as the change trend of the average number of third-party transactions per day of the set third-party transaction platform.

[0165] Specifically, when the change trend of the average number of third-party transactions per day of the set third-party transaction platform is a certain nonlinear change function, the change trend of the corresponding number of learning actions has the same change curve as the change curve corresponding to the certain nonlinear change function.

[0166] For example, when the average number of third-party transactions per day of the set third-party transaction platform is 5, the number of selected learning actions is 500, when the average number of third-party transactions per day of the set third-party transaction platform is 6, the number of selected learning actions is 600, when the average number of third-party transactions per day of the set third-party transaction platform is 7, the number of selected learning actions is 700, when the average number of third-party transactions per day of the set third-party transaction platform is 8, the number of selected learning actions is 800, and so on.

[0167] In each learning action performed on the radial basis neural network, the judgment result of whether a certain third-party payment account of the third-party transaction platform is a safe payment account at a certain historical time is set as the single output content of the radial basis neural network, and a plurality of safety-related parameters of a third-party payment account participating in a third-party payment transaction with the certain third-party payment account at the certain historical time, a set time length, a plurality of account-related information of the certain third-party payment account, and fund flow analysis data of the certain third-party payment account are set as the multiple input contents of the radial basis neural network, and the learning action is completed;

[0168] In the method, the plurality of account-related information of the target third-party payment account includes a registration duration of the target third-party payment account, a number of times of participating in the third-party payment transaction, a suspicious account ratio of a transaction counterpart, and a number of past complaints.

[0169] For example, a plurality of different information capturing units can be selected to respectively obtain the registration duration of the target third-party payment account, the number of times of participating in the third-party payment transaction, the suspicious account ratio of the transaction counterpart, and the number of past complaints.

[0170] In the method, the plurality of account-related information of the target third-party payment account includes a registration duration of the target third-party payment account, a number of times of participating in the third-party payment transaction, a suspicious account ratio of a transaction counterpart, and a number of past complaints.

[0171] For example, the suspicious account ratio of the transaction counterpart of the target third-party payment account includes that, when the number of the transaction counterparts corresponding to each of the third-party payment transactions participated in by the target third-party payment account is 20 and the number of the third-party payment accounts or the third-party payment accounts judged as suspicious accounts is 5, the suspicious account ratio of the transaction counterpart of the target third-party payment account is 1 / 4.

[0172] The plurality of security-related parameters of the third-party payment account in the current third-party payment transaction include gender identification, age data, number of times of participating in third-party payment transactions per day, and average transaction amount per day of the account owner of the third-party payment account in the current third-party payment transaction.

[0173] For example, a plurality of numerical analysis units can be selected to analyze the gender identification, age data, number of times of participating in third-party payment transactions per day, and average transaction amount per day of the account owner of the third-party payment account in the current third-party payment transaction, respectively.

[0174] The fund flow direction data corresponding to each third-party payment transaction includes the number of sub-transaction transactions initiated by the target third-party payment account as a payment account within a set time length after the third-party payment transaction, and each piece of amount data, virtual currency transaction initiation identification, and overseas transaction initiation identification corresponding to each sub-transaction transaction.

[0175] In addition, the present application can also refer to the following technical contents to further show the outstanding substantial progress of the present application:

[0176] The number of learning actions is the same as the change trend of the average number of third-party transactions per day of the set third-party transaction platform, including: the smaller the average number of third-party transactions per day of the set third-party transaction platform, the fewer the number of selected learning actions.

[0177] The number of learning actions is the same as the change trend of the average number of third-party transactions per day of the set third-party transaction platform, including: the smaller the average number of third-party transactions per day of the set third-party transaction platform, the fewer the number of selected learning actions.

[0178] For example, in the numerical mapping function, the average number of third-party transactions per day of the set third-party transaction platform is the input value of the numerical mapping function, and the number of learning actions that is the same as the change trend of the average number of third-party transactions per day of the set third-party transaction platform is the output value of the numerical mapping function.

[0179] While the above discussion discusses various embodiments of the application, the application is not limited thereto, but can be modified and varied by those skilled in the art in light of the above teachings without departing from the spirit and scope of the application as claimed in the appended claims.

Claims

1. A method for secure tracing of third party payment fund flow direction, characterized in that, The method comprises: taking a third-party payment account of a current third-party payment transaction just initiated by a set third-party transaction platform as a target third-party payment account, and obtaining multiple pieces of account association information of the target third-party payment account; taking multiple pieces of fund flow direction data corresponding to each third-party payment transaction in which the target third-party payment account participated as a third-party payment account as fund flow direction analysis data of the target third-party payment account, and each piece of fund flow direction data corresponding to each third-party payment transaction comprising the number of sub-transaction initiated by the target third-party payment account as a payment account within a set time length after the third-party payment transaction, and multiple pieces of amount data, virtual currency transaction initiation identifiers and overseas transaction initiation identifiers corresponding to each sub-transaction; adopting a transaction state intelligent judgment model to intelligently judge whether the target third-party payment account currently belongs to a safe payment account according to multiple pieces of security-related parameters of a third-party payment account of a current third-party payment transaction, a set time length, multiple pieces of account association information of the target third-party payment account and fund flow direction analysis data of the target third-party payment account; when intelligently judging that the target third-party payment account currently does not belong to a safe payment account, moving the target third-party payment account to a suspicious third-party payment account observation sequence, and judging the current third-party payment transaction as a suspicious third-party payment transaction to refuse execution; wherein the transaction state intelligent judgment model is a radial basis neural network after multiple learning actions.

2. The third-party payment fund flow direction security tracing method according to claim 1, wherein: the radial basis neural network after multiple learning actions comprises that the number of learning actions is the same as the change trend of the average number of third-party transactions of the set third-party transaction platform per day; wherein in each learning action performed on the radial basis neural network, a judgment result of whether a certain third-party payment account of the set third-party transaction platform at a certain historical time has been identified as a safe payment account is taken as a single output content of the radial basis neural network, and multiple pieces of security-related parameters of a third-party payment account of a third-party payment transaction in which the certain third-party payment account participated at the certain historical time, a set time length, multiple pieces of account association information of the certain third-party payment account and fund flow direction analysis data of the certain third-party payment account are taken as multiple input contents of the radial basis neural network, and the learning action is completed.

3. The third-party payment fund flow direction security tracing method according to claim 2, wherein: obtaining multiple pieces of account association information of the target third-party payment account comprises obtaining a registration duration, a number of third-party payment transaction participations, a suspicious number proportion of transaction counterparts and a number of past complaints of the target third-party payment account as multiple pieces of account association information of the target third-party payment account; The target third-party payment account is obtained by setting the third-party payment account of the current third-party payment transaction initiated by the third-party transaction platform as the target third-party payment account, and the registration duration, the number of times of participating in the third-party payment transaction, the suspicious number proportion of the transaction counterpart, and the number of past complaints of the target third-party payment account are obtained as the multiple account association information of the target third-party payment account, including: the number proportion of the third-party payment / collection accounts of each transaction counterpart judged as suspicious accounts in each third-party payment transaction participated by the target third-party payment account in the past is taken as the suspicious number proportion of the transaction counterpart of the target third-party payment account. The multiple security-related parameters of the third-party payment account of the current third-party payment transaction are the gender identification, age data, number of times of participating in the third-party payment transaction per day, and average transaction amount of the third-party payment account of the current third-party payment transaction per day. The fund flow direction data corresponding to each third-party payment transaction includes the number of sub-flow transactions initiated by the target third-party payment account as a payment account within a set time length after the third-party payment transaction, and each piece of amount data, a virtual currency transaction initiation identifier, and a cross-border transaction initiation identifier corresponding to each sub-flow transaction, including: the fund flow direction data corresponding to each third-party payment transaction includes the number of sub-flow transactions initiated by the target third-party payment account as a payment account within a set time length after the third-party payment transaction, and each piece of amount data, a virtual currency transaction initiation identifier, and a cross-border transaction initiation identifier corresponding to each sub-flow transaction, respectively representing whether the target third-party payment account initiates the purchase of virtual currency and whether the target third-party payment account initiates cross-border transactions as a payment account within a set time length after the third-party payment transaction.

4. The method of claim 3, wherein, After intelligently determining that the target third-party payment account does not currently belong to a safe collection account, transferring the target third-party payment account to a suspicious third-party payment account observation sequence, and determining that the current third-party payment transaction is a suspicious third-party payment transaction and refusing to execute, the method further includes: After determining that the current third-party payment transaction is a suspicious third-party payment transaction, initiating a collection account replacement request to replace the third-party payment account of the current third-party payment transaction to the third-party payment account of the current third-party payment transaction.

5. The method of claim 3, wherein, After intelligently determining that the target third-party payment account does not currently belong to a safe collection account, transferring the target third-party payment account to a suspicious third-party payment account observation sequence, and determining that the current third-party payment transaction is a suspicious third-party payment transaction and refusing to execute, the method further includes: After intelligently determining that the target third-party payment account currently belongs to a safe collection account, transferring the target third-party payment account to a safe third-party payment account observation sequence, and allowing the current third-party payment transaction to be executed.

6. The method of claim 3, wherein, corresponding to each of the third-party payment transactions in which the target third-party payment account has participated as a third-party payment account, the fund flow direction data corresponding to each of the third-party payment transactions including the number of sub-flow transactions initiated by the target third-party payment account as a payment account within a set time length after the third-party payment transaction, and the amount data, the virtual currency transaction initiation identifier, and the cross-border transaction initiation identifier corresponding to each of the sub-flow transactions, the method further comprising: performing multiple learning actions on the radial basis neural network to obtain the radial basis neural network after the multiple learning actions and output the radial basis neural network as the transaction state intelligent judgment model; wherein obtaining the radial basis neural network after the multiple learning actions and outputting the radial basis neural network as the transaction state intelligent judgment model comprises: completing the model representation of the transaction state intelligent judgment model using the model parameters of the transaction state intelligent judgment model.

7. The method of claim 3, wherein, after the target third-party payment account is transferred to the suspicious third-party payment account observation sequence and the current third-party payment transaction is judged as a suspicious third-party payment transaction and refused to be executed, the method further comprising: binding the suspicious third-party payment account observation sequence and the set third-party transaction platform and sending them to the remote fund supervision server together; wherein binding the suspicious third-party payment account observation sequence and the set third-party transaction platform and sending them to the remote fund supervision server together comprises: packaging each suspicious third-party payment account in the suspicious third-party payment account observation sequence and the platform code value of the set third-party transaction platform into a network data packet and then sending it to the remote fund supervision server.

8. The third-party payment fund flow security tracing method of claim 3, wherein: the intelligent judgment of whether the target third-party payment account currently belongs to a safe payment account by the transaction state intelligent judgment model according to the multiple safety-related parameters of the third-party payment account of the current third-party payment transaction, the set time length, the multiple account-related information of the target third-party payment account, and the fund flow direction analysis data of the target third-party payment account comprises: synchronously inputting the multiple safety-related parameters of the third-party payment account of the current third-party payment transaction, the set time length, the multiple account-related information of the target third-party payment account, and the fund flow direction analysis data of the target third-party payment account into the transaction state intelligent judgment model; wherein the intelligent judgment of whether the target third-party payment account currently belongs to a safe payment account by the transaction state intelligent judgment model according to the multiple safety-related parameters of the third-party payment account of the current third-party payment transaction, the set time length, the multiple account-related information of the target third-party payment account, and the fund flow direction analysis data of the target third-party payment account further comprises: executing the transaction state intelligent judgment model to obtain the account security identifier output by the transaction state intelligent judgment model representing whether the target third-party payment account currently belongs to a safe payment account. The plurality of security-related parameters of the third-party payment account of the current third-party payment transaction, the set time length, the plurality of account-related information of the target third-party payment account, and the fund flow analysis data of the target third-party payment account are synchronously input into the transaction state intelligent judgment model, including: before the plurality of security-related parameters of the third-party payment account of the current third-party payment transaction, the set time length, the plurality of account-related information of the target third-party payment account, and the fund flow analysis data of the target third-party payment account are synchronously input into the transaction state intelligent judgment model, the plurality of security-related parameters of the third-party payment account of the current third-party payment transaction, the set time length, the plurality of account-related information of the target third-party payment account, and the fund flow analysis data of the target third-party payment account are respectively subjected to numerical normalization processing; The transaction state intelligent judgment model is executed to obtain the account security identifier output by the transaction state intelligent judgment model, which indicates whether the target third-party payment account currently belongs to a safe payment account.

9. A third party payment fund flow security trace system, characterized in that, The system includes a memory and a plurality of processors, the memory stores a computer program, the computer program is configured to be executed by the plurality of processors to complete the following steps: The target third-party payment account is obtained as the target third-party payment account, and the plurality of account-related information of the target third-party payment account is obtained. The fund flow analysis data of the target third-party payment account when participating in each third-party payment transaction as a third-party payment account is obtained as the fund flow analysis data of the target third-party payment account, and the fund flow analysis data corresponding to each third-party payment transaction includes the number of sub-transaction initiated by the target third-party payment account as a payment account within a set time length after the third-party payment transaction, and each sub-transaction corresponds to each part of the amount data, the virtual currency transaction initiation identifier and the overseas transaction initiation identifier. The transaction state intelligent judgment model is used to intelligently judge whether the target third-party payment account currently belongs to a safe payment account according to the plurality of security-related parameters of the third-party payment account of the current third-party payment transaction, the set time length, the plurality of account-related information of the target third-party payment account, and the fund flow analysis data of the target third-party payment account. When it is intelligently judged that the target third-party payment account does not currently belong to a safe payment account, the target third-party payment account is transferred to a suspicious third-party payment account observation sequence, and the current third-party payment transaction is judged as a suspicious third-party payment transaction and is refused to be executed. The transaction state intelligent judgment model is a radial basis neural network after a plurality of learning actions.

10. A third party payment fund flow security trace system, characterized in that, The system includes: An information capturing device is used to obtain the target third-party payment account as the target third-party payment account, and the plurality of account-related information of the target third-party payment account is obtained. The security tracing device is used for taking each part of fund flow direction data corresponding to each third-party payment transaction in which the target third-party collection account participates as a third-party collection account in the past as fund flow direction analysis data of the target third-party collection account. The fund flow direction data corresponding to each third-party payment transaction includes the number of sub-flow transactions initiated by the target third-party collection account as a payment account within a set time length after the third-party payment transaction, and each part of amount data, virtual currency transaction initiation identifier and overseas transaction initiation identifier corresponding to each sub-flow transaction. The intelligent judgment device is connected with the information capturing device and the security tracing device respectively, and is used for intelligently judging whether the target third-party collection account currently belongs to a secure collection account according to a third-party payment transaction state intelligent judgment model, a plurality of safety-related parameters of a third-party payment account of a current third-party payment transaction, a set time length, a plurality of account-related information of the target third-party collection account and the fund flow direction analysis data of the target third-party collection account. The account processing device is connected with the intelligent judgment device, and is used for transferring the target third-party collection account to a suspicious third-party collection account observation sequence when the intelligent judgment device judges that the target third-party collection account currently does not belong to a secure collection account, and judging the current third-party payment transaction as a suspicious third-party payment transaction to refuse execution. The transaction state intelligent judgment model is a radial basis neural network after a plurality of learning actions.

Images