A risk account identification method and apparatus

Abstract

The application discloses a risk account identification method and device, and relates to the technical field of risk control. A specific implementation manner of the method comprises the following steps: acquiring transaction data corresponding to two payment systems respectively, wherein the transaction data comprises a plurality of transaction accounts and transaction information of each transaction account; generating a first transaction matrix and a second transaction matrix corresponding to the two payment systems respectively according to the transaction data; calculating a first similarity matrix corresponding to the two payment systems according to the first transaction matrix and the second transaction matrix; determining an account matching pair from the two payment systems according to the first similarity matrix and a matching algorithm; and determining a risk account according to the account matching pair. The implementation manner improves the accuracy of identifying the risk account.

Description

Technical Field

[0001] This invention relates to the field of risk control technology, and in particular to a method and apparatus for identifying risky accounts. Background Technology

[0002] With the development and popularization of technologies such as big data and blockchain, new forms of economic crime have emerged. Taking the "counter-trading" underground bank as an example, it colludes with individuals overseas to conduct cross-border fund transfers through domestic and international "counter-trading," effectively engaging in disguised foreign exchange trading. This method involves fictitious transactions between different payment systems and financial institutions to conceal the illegal source or transfer of funds; that is, it achieves "balance between two locations" through account offsetting, making it highly concealed.

[0003] Because wash trading typically involves large sums of money, risky accounts are currently identified based on transaction amount. For example, a threshold is set, and risky accounts are identified from those with transaction amounts exceeding that threshold. However, with the diversification of trading methods, identifying risky accounts solely based on transaction amount is less accurate. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method and apparatus for identifying risky accounts. Based on transaction data corresponding to two payment systems, a first similarity matrix is ​​calculated for the two payment systems. According to this first similarity matrix, a matching algorithm is used to determine account matching pairs from the two payment systems, and then risky accounts are identified based on these account matching pairs. Thus, based on the transaction similarity between the two payment systems, the matching algorithm accurately identifies account pairs with potential wash trading characteristics. Compared to related technologies that only use monetary amounts to identify risky accounts, this method improves the ability to identify complex transaction information and increases the accuracy of risky account identification.

[0005] To achieve the above objectives, according to one aspect of the present invention, a method for identifying risky accounts is provided. The method for identifying risky accounts according to this embodiment includes: acquiring transaction data corresponding to two payment systems, wherein the transaction data includes multiple transaction accounts and transaction information for each transaction account;

[0006] Based on the transaction data, a first transaction matrix and a second transaction matrix are generated for the two payment systems respectively; wherein, the rows of the first transaction matrix and the second transaction matrix represent the transaction accounts of their respective payment systems, and the columns of the first transaction matrix and the second transaction matrix represent the transaction information of their respective payment systems.

[0007] Based on the first transaction matrix and the second transaction matrix, calculate the first similarity matrix corresponding to the two payment systems;

[0008] Based on the first similarity matrix and the matching algorithm, account matching pairs are determined from the two payment systems;

[0009] Risky accounts are identified based on account matching pairs.

[0010] Optionally, based on the first similarity matrix and the matching algorithm, account matching pairs are determined from the two payment systems, including:

[0011] Based on a preset population size, multiple first populations are generated. Each first population includes multiple pairs of trading accounts, which are composed of trading accounts from a first trading matrix and a second trading matrix.

[0012] Calculate the fitness value of each first population based on the element values ​​in the first similarity matrix;

[0013] Based on the fitness value, the first group is iteratively optimized, and the account matching pairs are determined based on the results of the iterative optimization.

[0014] Optionally, based on the fitness value, the first population is iteratively optimized, including: taking multiple first populations as the current population, and repeatedly executing the following steps until the iterative optimization condition is met: selecting multiple second populations from the current population based on the fitness value, wherein the fitness value of the second population is not less than the fitness value of the other unselected first populations; performing crossover and mutation processing on the transaction account pairs in each second population based on the crossover and mutation strategies to generate a new current population; the crossover and mutation strategies are optimized based on the historical transaction data of the two payment systems.

[0015] Optionally, the method further includes: for each pair of trading accounts: calculating the distance between the trading account pairs based on the trading information, normalizing the distance, and obtaining the element values ​​in the first similarity matrix based on the processing result;

[0016] The fitness value of each first group is calculated based on the element values ​​in the first similarity matrix, including: determining the element values ​​corresponding to each pair of trading accounts in the first group from the first similarity matrix, and using the sum of the determined element values ​​as the fitness value of the first group.

[0017] Optionally, the method further includes: generating multiple third groups based on historical transaction data, wherein the third groups include multiple historical transaction account pairs, and the historical transaction account pairs are composed of accounts in two payment systems;

[0018] Based on the second similarity matrix corresponding to the two payment systems, the fitness value of the third population is calculated, and the third population is iteratively optimized based on the calculated fitness value.

[0019] The iterative optimization results of the third group are validated using pre-labeled risk accounts;

[0020] If validation fails, adjust the crossover and / or mutation strategies.

[0021] Optionally, based on the first transaction matrix and the second transaction matrix, the first similarity matrix corresponding to the two payment systems is calculated, including: calculating the distance between a pair of transaction accounts in the first transaction matrix and the second transaction matrix based on the transaction information, normalizing the distance, and using the reciprocal or negative value of the processing result as an element in the first similarity matrix;

[0022] Based on the first similarity matrix and the matching algorithm, account matching pairs are determined from the two payment systems, including: determining the distribution of zero elements of the first similarity matrix based on the minimum value of each row and the minimum value of each column in the first similarity matrix, and determining account matching pairs based on the distribution of zero elements.

[0023] Optionally, the distribution of zero elements of the first similarity matrix is ​​determined based on the minimum value of each row and the minimum value of each column in the first similarity matrix, and the account matching pairs are determined based on the distribution of zero elements, including: calculating the transformation matrix based on the difference between each element of each row and the minimum value of that row, and the difference between each element of each column and the minimum value of that column;

[0024] Each row of the transformation matrix is ​​labeled with a first matching label, and the other zero elements in the same column as the first matching label are labeled with a second matching label.

[0025] Determine whether the number of first matching tags is the same as the number of rows or columns of the first similarity matrix. If so, take the transaction account pair corresponding to the first matching tag as the account matching pair.

[0026] Optionally, based on the transaction data, a first transaction matrix and a second transaction matrix corresponding to the two payment systems are generated respectively, including:

[0027] Convert the transaction amounts in the transaction data corresponding to the two payment systems into the same currency;

[0028] Based on a preset time window, select transaction data from the converted transaction data that meets the time window and whose transaction amount is greater than the amount threshold;

[0029] Based on the selected transaction data, a first transaction matrix and a second transaction matrix are generated.

[0030] Optionally, if the number of transaction accounts selected by the two payment systems is different, a first transaction matrix and a second transaction matrix are generated based on the selected transaction data, including:

[0031] For payment systems with a small number of transaction accounts, virtual accounts and corresponding virtual transaction information are configured. The transaction amount in the virtual transaction information is less than the amount threshold, and the number of virtual accounts is the difference between the number of transaction accounts selected by the two payment systems. Based on the virtual accounts, virtual transaction information, and selected transaction data, a first transaction matrix and a second transaction matrix are generated.

[0032] To achieve the above objectives, according to another aspect of the present invention, a risk account identification device is provided. One such risk account identification device includes: a data acquisition module, a matrix calculation module, and an account identification module; wherein,

[0033] The data acquisition module is configured to acquire transaction data corresponding to the two payment systems, including multiple transaction accounts and transaction information for each account.

[0034] The matrix calculation module is configured to: generate a first transaction matrix and a second transaction matrix corresponding to the two payment systems based on the transaction data; wherein the rows of the first transaction matrix and the second transaction matrix represent the transaction accounts of their respective payment systems, and the columns of the first transaction matrix and the second transaction matrix represent the transaction information of their respective payment systems; and calculate a first similarity matrix corresponding to the two payment systems based on the first transaction matrix and the second transaction matrix.

[0035] The account identification module is configured to: determine account matching pairs from two payment systems based on the first similarity matrix and the matching algorithm; and identify risky accounts based on the account matching pairs.

[0036] To achieve the above objectives, according to another aspect of the present invention, an electronic device for identifying risky accounts is provided. An electronic device for identifying users according to an embodiment of the present invention includes: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement a risky account identification method according to an embodiment of the present invention.

[0037] To achieve the above objectives, according to another aspect of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium of the present invention stores a computer program thereon, which, when executed by a processor, implements a risk account identification method according to the present invention.

[0038] To achieve the above objectives, according to another aspect of the present invention, a computer program product is provided, including a computer program that, when executed by a processor, implements a risk account identification method according to an embodiment of the present invention.

[0039] One embodiment of the above invention has the following advantages or beneficial effects: A first similarity matrix is ​​calculated between the two payment systems based on their respective transaction data. Based on this first similarity matrix, a matching algorithm is used to determine account matching pairs from the two payment systems, and then risky accounts are identified based on these account matching pairs. Thus, based on the transaction similarity between the two payment systems, the matching algorithm accurately identifies account pairs with potential wash trading characteristics. Compared to related technologies that only use monetary amounts to identify risky accounts, this improves the ability to identify complex transaction information and increases the accuracy of risky account identification.

[0040] The further effects of the aforementioned unconventional alternative methods will be explained below in conjunction with specific implementation methods. Attached Figure Description

[0041] The accompanying drawings are provided to better understand the invention and are not intended to unduly limit the scope of the invention. Wherein:

[0042] Figure 1 This is a schematic diagram of the main steps of a risk account identification method according to an embodiment of the present invention;

[0043] Figure 2 This is a schematic diagram illustrating the main steps of another risk account identification method according to an embodiment of the present invention;

[0044] Figure 3 This is a schematic diagram of the main steps of another risk account identification method according to an embodiment of the present invention;

[0045] Figure 4 This is a schematic diagram of the main modules of a risk account identification device according to an embodiment of the present invention;

[0046] Figure 5 This is an exemplary system architecture diagram in which embodiments of the present invention can be applied;

[0047] Figure 6 This is a schematic diagram of the structure of a computer system suitable for implementing terminal devices or servers of the present invention. Detailed Implementation

[0048] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of the present invention, including various details to aid understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0049] It should be noted that, unless otherwise specified, the embodiments of the present invention and the technical features thereof can be combined with each other.

[0050] Figure 1 This invention relates to a risk account identification method, which mainly includes the following steps:

[0051] Step S101: Obtain the transaction data corresponding to the two payment systems respectively. The transaction data includes multiple transaction accounts and transaction information of each transaction account.

[0052] The two payment systems can be the domestic payment system M (hereinafter referred to as System M) and the stablecoin system N (hereinafter referred to as System N). The transaction data refers to the transaction data generated by System M and System N within a certain period of time. This transaction data includes transaction accounts and transaction information such as whether each transaction account has large transactions, the number of transactions transferred in, the number of transactions transferred out, and the transaction amount.

[0053] Step S102: Based on the transaction data, generate a first transaction matrix and a second transaction matrix corresponding to the two payment systems respectively; wherein, the rows of the first transaction matrix and the second transaction matrix represent the transaction accounts of the corresponding payment systems, and the columns of the first transaction matrix and the second transaction matrix represent the transaction information of the corresponding payment systems respectively.

[0054] In the process of forming the first and second transaction matrices, considering that the transaction currencies of different payment systems may differ, the transaction amounts in the transaction data corresponding to the two payment systems are converted to the same currency to facilitate subsequent calculations. For example, the transaction amounts of both payment systems are converted to RMB or USD. A time window is preset to account for time differences between different payment systems. For example, the time window is set to the day of risk account identification and the day before and after. If the risk account is to be identified on January 2, 2024, the time window is set to January 1, 2024 to January 3, 2024. Then, based on the preset time window, transaction data that meets the time window and whose transaction amount is greater than a threshold is selected from the converted transaction data. For example, if the initial transaction data obtained from the two payment systems is data from the most recent month, then in the process of generating the first and second transaction matrices, transaction data from January 1, 2024 to January 3, 2024 with transaction amounts greater than the threshold are selected based on the time window and the threshold. The criteria for setting the threshold can be the same as or different from the criteria for determining whether a large transaction has occurred. If both criteria are the same, an account that has made a large transaction within a time window can be included as an element in the transaction matrix.

[0055] For the selected transaction data, the transaction accounts corresponding to the two payment systems may be different. For example, system M may have 100 transaction accounts with transaction amounts exceeding a threshold within a preset time window, while system N may only have 90 transaction accounts with transaction amounts exceeding the threshold within the preset time window. In this case, virtual accounts and corresponding virtual transaction information are configured for the payment system with fewer transaction accounts. The transaction amounts in the virtual transaction information are less than the threshold. The number of virtual accounts is the difference between the number of transaction accounts selected by the two payment systems. Then, based on the virtual accounts, virtual transaction information, and selected transaction data, a first transaction matrix and a second transaction matrix are generated. Referring again to the example where system M selects 100 transaction accounts and system N selects 90 transaction accounts, 10 virtual accounts are configured for system N, and corresponding virtual transaction information is configured for these 10 virtual accounts. The transaction amounts in this virtual transaction information are less than the threshold, making them different from the transaction characteristics of other actual transaction data. This ensures that these virtual transactions and their virtual transaction information will not interfere with the identification process when calculating matrix similarity and using matching algorithms based on the similarity matrix to identify risky accounts.

[0056] In this embodiment of the invention, the transaction information can also be standardized to reduce the impact of inconsistencies in the dimensions of different features. For example, the first transaction matrix corresponding to system M is shown below, where M1, M2...Mk represent transaction accounts, mij represents transaction information, i represents the column number of the corresponding transaction account, and j represents the row number of the transaction information.

[0057]

[0058] Similarly, the second transaction matrix corresponding to system N can be referenced from the following matrix N, where N1, N2...Nk represent transaction accounts, nij represent transaction information, i represents the column number of the corresponding transaction account, and j represents the row number of the transaction information. Among these k transaction accounts, 10 are virtual accounts, and their corresponding transaction information is virtual transaction information.

[0059]

[0060] Step S103: Calculate the first similarity matrix corresponding to the two payment systems based on the first transaction matrix and the second transaction matrix.

[0061] In one embodiment of the present invention, the elements in the first similarity matrix can be determined based on the distance between a pair of trading accounts in the first and second trading matrices. In other words, the elements in the first similarity matrix are obtained based on the distance between a pair of trading accounts. For example, the elements in the first row of the first similarity matrix are the distances between the first trading account in the first trading matrix and each trading account in the second trading matrix, the elements in the second row are the distances between the second trading account in the first trading matrix and each trading account in the second trading matrix, and so on, to obtain each element in the first similarity matrix.

[0062] For example, the distance between a pair of trading accounts in the first and second trading matrices can be calculated based on transaction information. This distance is then normalized, and the reciprocal or negative value of the result is used as an element in the first similarity matrix. For instance, for a pair of trading accounts consisting of trading account M1 in the first trading matrix M and trading account N1 in the second trading matrix N, the distance between trading account M1 and trading account N1 can be calculated based on their transaction information using methods such as Euclidean distance, Manhattan distance, or Chebyshev distance. This distance is then normalized to obtain the result M1N1, and 1 / M1N1 or -M1N1 is used as an element in the first similarity matrix. Similarly, processing other trading accounts in the first and second trading matrices N yields the first similarity matrix S of the matching algorithm.

[0063] or

[0064] Alternatively, when the matching algorithm is a population matching algorithm, the distance between trading account pairs can be calculated based on transaction information. This distance can then be normalized, and the element values ​​in the first similarity matrix can be obtained based on the processing result. Taking the trading account pair consisting of trading account M1 in the first transaction matrix M and trading account N1 in the second transaction matrix N as an example, the normalized processing result M1N1 can be used as an element in the first similarity matrix. Thus, the first similarity matrix S' of the matching algorithm can be obtained:

[0065]

[0066] Step S104: Based on the first similarity matrix and the matching algorithm, determine the account matching pairs from the two payment systems.

[0067] In one embodiment of the present invention, to improve the efficiency and accuracy of identifying risky accounts, the matching algorithm can be iteratively optimized before determining account matching pairs using the matching algorithm. For example, before identifying risky accounts, the matching algorithm can be iteratively optimized using historical transaction data. Specifically, based on the historical transaction data of the two payment systems, multiple third groups are generated, each including multiple historical transaction account pairs, each consisting of transaction accounts from the two payment systems. The fitness value of the third group is calculated based on the second similarity matrix corresponding to the historical transaction data of the two payment systems. The third group is then iteratively optimized based on the calculated fitness value. The iterative optimization results of the third group can then be verified using pre-labeled risky accounts. If the verification fails, the crossover strategy and / or mutation strategy in the matching algorithm are adjusted.

[0068] In this embodiment, transaction matrices for the two payment systems can also be generated first based on historical transaction data. The method for generating the transaction matrices is basically the same as that for generating the first and second transaction matrices. For example, after currency conversion of the transaction amounts in the historical transaction data, transaction data that meets a preset time window and whose transaction amounts are greater than a threshold can be selected to generate transaction matrices X and Y for the two payment systems. Transaction matrices X and Y can be as follows: X1, X2...Xk represent transaction accounts, xij represents transaction information, i represents the column number of the corresponding transaction account, and j represents the row number of the transaction information; Y1, Y2...Yk represent transaction accounts, yij represents transaction information, i represents the column number of the corresponding transaction account, and j represents the row number of the transaction information.

[0069]

[0070] Then, based on historical transaction information, the distance between transaction account pairs in transaction matrices X and Y can be calculated. This distance is then normalized, and the result yields a second similarity matrix corresponding to the historical transaction data. Each element in the second similarity matrix represents the normalized distance. Distance calculation methods can also include Euclidean distance, Manhattan distance, or Chebyshev distance. In short, the historical transaction data undergoes essentially the same processing as the currently acquired transaction data to obtain the second similarity matrix corresponding to the historical transaction data. For example, if two payment systems each have four accounts, the second similarity matrix S" obtained based on Euclidean distance and normalization would be:

[0071]

[0072] With a preset population size of 4, the four randomly generated third populations each contain four trading account pairs. For example, third population 1 is [(X1,Y1),(X2,Y2),(X3,Y3),(X4,Y4)], third population 2 is [(X1,Y1),(X2,Y3),(X3,Y2),(X4,Y4)], third population 3 is [(X1,Y2),(X2,Y1),(X3,Y3),(X4,Y4)], and third population 4 is [(X1,Y3),(X2,Y1),(X3,Y2),(X4,Y4)]. Where X1, X2, X3, X4 are account elements in transaction matrix X, Y1, Y2, Y3, Y4 are account elements in transaction matrix Y, and (Xe, Yc) is the historical transaction account pair, where e can be 1, 2, 3, 4, and c can also be 1, 2, 3, 4.

[0073] Then, the element values ​​corresponding to each pair of trading accounts in the third group are determined from the second similarity matrix S". The sum of the determined element values ​​is used as the fitness value of the third group. For example, for the third group 1, the element value corresponding to the trading account pair (X1,Y1) is 0.9, the element value corresponding to the trading account pair (X2,Y2) is 0.8, the element value corresponding to the trading account pair (X3,Y3) is 0.7, and the element value corresponding to the trading account pair (X4,Y4) is 0.9. Therefore, the fitness value of the third group 1 is 3.3. Similarly, the fitness values ​​of the third group 2 are calculated as 0.9+0.3+0.6+0.9=2.7, the fitness values ​​of the third group 3 are 0.2+0.1+0.7+0.9=1.9, and the fitness values ​​of the third group 4 are 0.4+0.1+0.3+0.9=1.7.

[0074] In the iterative optimization of the third population based on its fitness value, multiple iterations can be performed until the iterative optimization conditions are met. Specifically, the third population is used as the current population, and the following steps are executed repeatedly until the iterative optimization conditions are met:

[0075] Based on the fitness value, select multiple fourth populations from the current population, where the fitness value of the fourth population is not less than the fitness value of the other unselected third populations; based on the preset crossover and mutation strategies, perform crossover and mutation processing on the trading account pairs in each third population to generate a new current population.

[0076] The iterative optimization condition is that the number of iterations is greater than a threshold or the fitness of the current population is greater than a fitness threshold. Taking the four third populations mentioned above as an example, two fourth populations can be selected from the four third populations using the roulette wheel selection method. In this example, the fitness values ​​of third population 1 and third population 2 are greater than those of third population 3 and third population 4; therefore, third population 1 and third population 2 are selected as the fourth populations.

[0077] Then, according to the preset crossover and mutation strategies, the trading account pairs in the fourth group are subjected to crossover and mutation processing. For example, the preset crossover strategy is to cross two accounts, and the mutation strategy is to mutate two adjacent account pairs. Then, during the crossover, from the third group 1 [(X1,Y1),(X2,Y2),(X3,Y3),(X4, Y4 )] and the third group 2[(X1,Y1),(X2,Y3),(X3, Y2 Choose one account from [(X1,Y1),(X2,Y2),(X3,Y3),(X4,Y4)] and perform crossover. For example, crossover Y4 and Y2 to obtain the new current population 5 as [(X1,Y1),(X2,Y2),(X3,Y3),(X4,Y4)]. Y2 The new current population 6 is [(X1,Y1),(X2,Y3),(X3, Y4 In the new current population 5, since information about trading account Y2 appears in both trading account pairs, to improve the diversity of information in the population, the new current population 5 can be modified to avoid duplicate trading accounts, such as modifying the new current population 5 to [(X1,Y1),(X2,Y4),(X3,Y3),)(X4,Y2)]. Similarly, the new current population 6 can be modified to [(X1,Y1),(X2,Y3),(X3,Y4),(X4,Y2)]. Additionally, mutations can be performed on an account in the third population 1 [(X1,Y1),(X2,Y2),(X3,Y3),(X4,Y4)] and the third population 2 [(X1,Y1),(X2,Y3),(X3,Y2),(X4,Y4)] to obtain a new current population. For example, mutations are performed on (X1,Y1) and (X2,Y2) in the third population 1 to obtain a new current population 7 [(X1,Y2),(X2,Y1),(X3,Y3),(X4,Y4)]. Mutations are performed on (X1,Y1) and (X2,Y3) in the third population 2 to obtain a new current population 8 [(X1,Y3),(X2,Y1),(X3,Y2),(X4,Y4)].

[0078] Then, the fitness values ​​of the new current population 5-8 can be calculated based on the second similarity matrix S". Then, the population with larger fitness values ​​can be selected. This process is repeated to select the third population that is globally optimal based on the fitness values.

[0079] To further improve the accuracy of the matching algorithm in identifying account matching pairs, in one embodiment of the present invention, the iterative results of the third population can be verified by combining pre-labeled risk accounts. If the verification passes, the matching algorithm optimization is determined to be complete; if the verification fails, the crossover strategy and / or mutation strategy are adjusted to change the parameters for further optimization of the matching algorithm. For example, after determining that the iterative optimization results of the third population meet the iterative optimization conditions, the pre-labeled risk accounts are used to verify the trading account pairs in the new current population, such as verifying whether the trading account pairs match the risk accounts. If the matching rate exceeds a preset matching threshold, the verification is determined to pass; otherwise, the verification fails. If the verification fails, the crossover strategy and / or mutation strategy of the population can be adjusted. For example, in the first iterative optimization process, if after the number of iterations exceeds a threshold, the matching rate between the trading account pairs and risk accounts in the first population is lower than the matching threshold, the crossover strategy and mutation strategy can be adjusted.

[0080] Therefore, after iterative optimization and parameter adjustment, an optimized matching algorithm is obtained. When determining account matching pairs from payment system M and payment system N, the optimized matching algorithm can be used to accurately determine the account matching pairs based on the first similarity matrix S′, which helps to improve the accuracy of identifying risky accounts.

[0081] Specifically, in the process of determining account matching pairs, multiple first populations can be generated according to the preset population size. Each first population includes multiple transaction account pairs, and each transaction account pair is composed of transaction accounts in the first transaction matrix and the second transaction matrix. The fitness value of each first population is calculated based on the element values ​​in the first similarity matrix. The first population is iteratively optimized based on its fitness value, and the account matching pairs are determined based on the iterative optimization results.

[0082] In one embodiment of the present invention, when calculating the fitness value of the first group, the element values ​​corresponding to each pair of trading accounts in the first group can be determined from the first similarity matrix, and the sum of the determined element values ​​is used as the fitness value of the first group.

[0083] The iterative optimization process for the first population is essentially the same as that for the third population, with the main difference being that the crossover and mutation strategies used in the first population's iterative optimization are derived from historical transaction data of the two payment systems, i.e., from the iterative optimization results of the third population. Specifically, the iterative optimization process for the first population may include the following steps: taking multiple first populations as the current population, and repeatedly executing the following steps until the iterative optimization conditions are met: selecting multiple second populations from the current population based on their fitness values, wherein the fitness values ​​of the second populations are not less than the fitness values ​​of the other unselected current populations; performing crossover and mutation processing on the transaction account pairs in each second population based on the crossover and mutation strategies to generate a new current population; wherein the crossover and mutation strategies are derived from historical transaction data of the two payment systems.

[0084] For example, when both the first and second transaction matrices include four transaction accounts, the calculated first similarity matrix S' is:

[0085]

[0086] With a preset population size of 4, the four randomly generated first populations each contain four trading account pairs. For example, the first population 1 is [(M1,N1),(M2,N2),(M3,N3),(M4,N4)], the first population 2 is [(M1,N1),(M2,N3),(M3,N2),(M4,N4)], the first population 3 is [(M1,N2),(M2,N1),(M3,N3),(M4,N4)], and the first population 4 is [(M1,N3),(M2,N1),(M3,N2),(M4,N4)]. Where M1, M2, M3, M4 are account elements in the first transaction matrix M, N1, N2, N3, N4 are account elements in the second transaction matrix N, and (Me, Nc) is the transaction account pair, where e can be 1, 2, 3, 4, and c can also be 1, 2, 3, 4.

[0087] Then, the element values ​​corresponding to each pair of trading accounts in the first group are determined from the first similarity matrix S', and the sum of the determined element values ​​is used as the fitness value of the first group. For example, for the first group 1, the element value corresponding to the trading account pair (M1, N1) is 0.2, the element value corresponding to the trading account pair (M2, N2) is 0.7, the element value corresponding to the trading account pair (M3, N3) is 0.9, and the element value corresponding to the trading account pair (M4, N4) is 0.2, so the fitness value of the first group 1 is 2. Similarly, the fitness value of the first group 2 is calculated to be 0.2 + 0.1 + 0.1 + 0.2 = 0.6, the fitness value of the first group 3 is 0.6 + 0.1 + 0.9 + 0.2 = 1.8, and the fitness value of the first group 4 is 0.3 + 0.1 + 0.1 + 0.2 = 0.7.

[0088] In the iterative optimization process of the first population, multiple iterations can be performed until the iterative optimization conditions are met. These conditions include either the number of iterations exceeding a threshold or the fitness of the current population exceeding a fitness threshold. Taking the four first populations as an example, two second populations can be selected using a roulette wheel selection method. In this example, the fitness values ​​of first population 1 and 3 are higher than those of first population 2 and 4; therefore, first population 1 and 3 are selected as the second populations. Then, based on the optimized crossover and mutation strategies, first population 1 and 3 are crossovered and mutated to obtain a new current population. The fitness value of the new current population is then calculated. This process is repeated until the iterative optimization conditions are met. The trading account pair from the first population with the highest fitness value is then selected as the account matching pair.

[0089] In another embodiment of the present invention, account matching pairs can be determined by another matching algorithm. Specifically, the distribution of zero elements of the first similarity matrix can be determined based on the minimum value of each row and the minimum value of each column in the first similarity matrix, and account matching pairs can be determined based on the distribution of zero elements.

[0090] Taking the first transaction matrix M and the second transaction matrix N corresponding to payment system M and payment system N respectively as examples, the resulting first similarity matrix S is: or If both the first transaction matrix M and the second transaction matrix N contain four accounts, and the negative value of the normalized distance between them is used as an element in the first similarity matrix, the resulting first similarity matrix is ​​exemplarily as follows:

[0091]

[0092] Then, based on the difference between each element in each row of the first similarity matrix S and the minimum value of that row, and the difference between each element in each column and the minimum value of that column, the transformation matrix is ​​calculated; the zero elements in the transformation matrix are marked as the first matching label row by row, and the other zero elements in the same column as the first matching label are marked as the second matching label; it is determined whether the number of the first matching labels is the same as the number of rows or columns of the first similarity matrix. If so, the transaction account pairs corresponding to the first matching labels are taken as account matching pairs.

[0093] In this example, based on the first similarity matrix S mentioned above, by subtracting the minimum value of each element in each row from the minimum value of that row, and then subtracting the minimum value of each element in each column from the minimum value of that column, we can obtain the transformation matrix Z as follows:

[0094]

[0095] The zero elements in the transformation matrix Z are labeled, for example, 1 and 2 are used as the first matching label and the second matching label, respectively. Starting from the first row, if there is only one zero element in the row, then the zero element is labeled as 1. For each matching label 1, the other zero elements in the same column are labeled as 2. The resulting labeled transformation matrix is ​​as follows:

[0096]

[0097] Then determine whether the number of first matching tags (4) is the same as the number of rows or columns (4) of the first similarity matrix. In this example, they are the same. Then the transaction account pair corresponding to the first matching tag is taken as the account matching pair. That is, the determined account matching pairs are: (M1,N1), (M2N2), (M3,N3), (M4,N4).

[0098] Furthermore, if the number of first matching tags is less than the number of rows or columns of the first similarity matrix, a transformation can be performed based on the first similarity matrix. For example, if the reciprocal of the distance normalization result is used as the element of the first similarity matrix, the resulting first similarity matrix S is:

[0099]

[0100] The transformation matrix Z is calculated based on the first similarity matrix, and the result after labeling is as follows:

[0101]

[0102] The number of first matching tags 1 is 3, which is less than the number of rows or columns of the first similarity matrix (4). In this case, all zeros are first covered with the fewest horizontal and vertical lines. Then, a minimum element is found among the uncovered elements. This minimum element is subtracted from each element in the row containing the uncovered element. At this time, negative elements may appear among the covered elements. Then, this minimum element is added to each element in the column containing the negative element. Transforming the above-labeled matrix, the following matrix Z' can be obtained:

[0103]

[0104] In this case, if the number of first matching tags (3) in matrix Z' is still less than the number of rows or columns (4) of the first similarity matrix, then the matrix is ​​drawn and its elements are transformed again: first, all zeros are covered with the fewest horizontal and vertical lines, then a minimum element is found among the uncovered elements, and this minimum element is subtracted from each element in the row containing the uncovered element. At this time, negative elements may appear among the covered elements, so this minimum element is added to each element in the column containing the negative element. This process is repeated until each row in the matrix has an element marked as the first matching tag, that is, until the number of first matching tags is equal to the number of rows or columns of the first similarity matrix. Then, the transaction account pairs corresponding to the first matching tags are taken as account matching pairs.

[0105] Step S105: Identify risky accounts based on account matching pairs.

[0106] After determining account matching pairs based on the first similarity matrix using the matching algorithm, the account matching pairs can be directly identified as risky accounts, or the account matching pairs can be further filtered. For example, if the fitness values ​​of multiple populations obtained based on the matching algorithm are all higher than the fitness threshold, account matching pairs with higher fitness values ​​can be further filtered as risky accounts.

[0107] The risk account identification method provided in this embodiment of the invention will be further explained below using two matching algorithms as examples.

[0108] Example 1

[0109] like Figure 2 As shown, this risk account identification method includes the following steps:

[0110] Step S201: Obtain historical transaction data from the two payment systems and generate transaction matrices corresponding to the two payment systems respectively.

[0111] In this transaction matrix, rows represent account information, and columns represent transaction information. During the generation of the transaction matrix, the transaction amounts corresponding to the two payment systems are first converted to the same currency. Then, based on a preset time window, historical transaction data that meets the time window requirements and whose transaction amount is greater than a threshold is selected from the converted historical transaction data. If the number of transaction accounts differs between the two payment systems in the selected historical transaction data, virtual accounts and corresponding virtual transaction information can be configured for the payment system with the smaller number of accounts. The transaction amount in the virtual transaction information is less than the threshold, ensuring that the transaction matrices for the two payment systems have the same number of transaction accounts.

[0112] Step S202: Generate multiple third populations according to the preset population size. Each third population includes multiple historical trading account pairs. Each historical trading account pair consists of trading accounts from two trading matrices.

[0113] Step S203: Calculate the second similarity matrix corresponding to the historical transaction data based on the transaction matrices corresponding to the two payment systems respectively.

[0114] Here, the distance between trading accounts in the two trading matrices can be calculated based on the trading information in the trading matrix. This distance is then normalized, and the reciprocal or negative value of the result is used as an element in the second similarity matrix to obtain the second similarity matrix.

[0115] Step S204: Calculate the fitness value of each third population based on the element values ​​in the second similarity matrix.

[0116] Step S205: Use the third population as the current population.

[0117] Step S206: Select multiple fourth populations with higher fitness values ​​from the current population.

[0118] Here, the fitness value of the fourth population is no less than the fitness values ​​of the other unselected current populations.

[0119] Step S207: Based on the preset crossover and mutation strategies, perform crossover and mutation processing on the trading accounts in each fourth group to generate a new current group.

[0120] Step S208: Use the new current population as the iteration result of the third population to determine whether it meets the iterative optimization conditions. If it does, proceed to step S209; otherwise, proceed to step S206.

[0121] Step S209: Validate the iteration results of the third group using the pre-labeled risk account; if the validation passes, proceed to step S211; otherwise, proceed to step S210.

[0122] Step S210: Adjust the crossover and mutation strategies, and proceed to step S205.

[0123] If the verification fails, the similarity calculation method can be adjusted and iterative optimization can be performed again. For example, if the second similarity matrix is ​​calculated using Euclidean distance in this iteration, and the verification fails, the calculation method can be adjusted to Manhattan distance or another similarity algorithm, and iterative optimization can be performed again based on the new second similarity matrix. After adjusting the similarity algorithm, step S203 can be executed after step S210.

[0124] Step S211: Determine that the matching algorithm optimization is complete.

[0125] Step S212: Obtain the transaction data corresponding to the two payment systems respectively, and generate the first transaction matrix and the second transaction matrix corresponding to the two payment systems respectively based on the transaction data.

[0126] The transaction data includes multiple transaction accounts and transaction information for each account. Specifically, the rows of the first and second transaction matrices represent the transaction accounts of the two payment systems, and the columns of the first and second transaction matrices represent the transaction information of the two payment systems.

[0127] Step S213: Calculate the first similarity matrix corresponding to the two payment systems based on the first transaction matrix and the second transaction matrix.

[0128] Step S214: Generate multiple first populations according to the preset population size. Each first population includes multiple trading account pairs, and each trading account pair is composed of account elements in the first trading matrix and the second trading matrix.

[0129] Step S215: Determine the element values ​​corresponding to each pair of trading accounts in the first group from the first similarity matrix, and use the sum of the determined element values ​​as the fitness value of the first group.

[0130] Step S216: Use the first population as the current population.

[0131] Step S217: Select multiple second populations with higher fitness values ​​from the current population.

[0132] Step S218: Based on the optimized crossover and mutation strategies, perform crossover and mutation processing on the trading accounts in each second population to generate a new current population.

[0133] Here, the fitness value of the second population is no less than the fitness values ​​of the other unselected current populations.

[0134] Step S219: Use the new current population as the iteration result of the first population to determine whether it meets the iterative optimization conditions. If it does, proceed to step S220; otherwise, proceed to step S217.

[0135] Step S220: Determine account matching pairs based on the iteration results of the first group, and determine risky accounts based on the account matching pairs.

[0136] It should be understood that after optimizing the matching algorithm in steps S201-S211, when identifying risky accounts, it is not necessary to execute the optimization step of the matching algorithm every time. Instead, step S212 can be executed directly based on the transaction data to identify the risky accounts corresponding to the current transaction data. After a certain period, the identified risky accounts can be used as historical transaction data to further optimize the matching algorithm.

[0137] Example 2

[0138] like Figure 3 As shown, this risk account identification method includes the following steps:

[0139] Step S301: Obtain the transaction data corresponding to the two payment systems respectively.

[0140] The transaction data includes multiple trading accounts and transaction information for each account.

[0141] Step S302: Based on the transaction data, generate the first transaction matrix and the second transaction matrix corresponding to the two payment systems respectively.

[0142] In this system, the rows of the first transaction matrix and the second transaction matrix represent the transaction accounts of the two payment systems, and the columns of the first transaction matrix and the second transaction matrix represent the transaction information of the two payment systems.

[0143] Step S303: Based on the transaction information of each transaction account in the first transaction matrix and the second transaction matrix, calculate the distance between the transaction account pairs in the first transaction matrix and the second transaction matrix, normalize the distance, and use the negative number of the processing result as an element in the first similarity matrix to obtain the first similarity matrix.

[0144] Step S304: Calculate the transformation matrix based on the differences between each element in each row of the first similarity matrix and the minimum value of that row, and the differences between each element in each column and the minimum value of that column.

[0145] Step S305: Mark the zero elements in the transformation matrix row by row as the first matching label, and mark the other zero elements in the same column as the first matching label as the second matching label.

[0146] Step S306: Determine whether the number of first matching tags is the same as the number of rows in the first similarity matrix. If yes, proceed to step S308; otherwise, proceed to step S307.

[0147] Step S307: Cover all zero elements in the transformation matrix with the fewest horizontal and vertical lines. Determine a minimum element among the uncovered elements. Subtract the minimum element from each element in the row containing the uncovered elements. Based on the calculation results, for the negative elements that appear among the covered elements, add the minimum element to each element in the column containing the negative element, and then execute step S305.

[0148] Step S308: Take the trading account pair corresponding to the first matching tag as the account matching pair, and determine the risk account based on the account matching pair.

[0149] According to an embodiment of the present invention, a risk account identification method involves calculating a first similarity matrix between two payment systems based on their respective transaction data. Based on this first similarity matrix, a matching algorithm is used to determine account matching pairs from the two payment systems, and then risk accounts are identified based on these matching pairs. Thus, by using the matching algorithm to accurately identify account pairs with potential wash trading characteristics based on the transaction similarity between the two payment systems, this method improves the ability to identify complex transaction information and increases the accuracy of risk account identification compared to related technologies that only use monetary amounts. This provides strong support for tracing risk clues in payment systems.

[0150] Based on the same inventive concept as the above embodiments, this embodiment of the invention also provides a risk account identification device, such as... Figure 4 As shown, the risk account identification device 400 includes: a data acquisition module 401, a matrix calculation module 402, and an account identification module 403; wherein,

[0151] The data acquisition module 401 is configured to acquire transaction data corresponding to the two payment systems respectively. The transaction data includes multiple transaction accounts and transaction information of each transaction account.

[0152] The matrix calculation module 402 is configured to: generate a first transaction matrix and a second transaction matrix corresponding to the two payment systems respectively, based on the transaction data; wherein, the rows of the first transaction matrix and the second transaction matrix represent the transaction accounts of their respective payment systems, and the columns of the first transaction matrix and the second transaction matrix represent the transaction information of their respective payment systems; and calculate a first similarity matrix corresponding to the two payment systems based on the first transaction matrix and the second transaction matrix.

[0153] The account identification module 403 is configured to: determine account matching pairs from two payment systems based on the first similarity matrix and the matching algorithm; and determine risky accounts based on the account matching pairs.

[0154] In one embodiment of the present invention, the account identification module 403: generates multiple first populations according to a preset population size, each first population including multiple transaction account pairs, each transaction account pair being composed of transaction accounts in a first transaction matrix and a second transaction matrix; calculates the fitness value of each first population based on the element values ​​in the first similarity matrix; performs iterative optimization on the first population based on the fitness value; and determines account matching pairs based on the iterative optimization results.

[0155] In one embodiment of the present invention, the account identification module 403: takes multiple first populations as the current population and repeatedly executes the following steps until the iterative optimization condition is met: select multiple second populations from the current population based on fitness values, wherein the fitness values ​​of the second populations are not less than the fitness values ​​of other unselected current populations; perform crossover and mutation processing on the transaction account pairs in each second population according to the crossover and mutation strategies to generate a new current population; the crossover and mutation strategies are optimized based on the historical transaction data of the two payment systems.

[0156] In one embodiment of the present invention, the account identification module 403: for each transaction account pair: calculates the distance between the transaction account pairs according to the transaction information, normalizes the distance, and obtains the element values ​​in the first similarity matrix according to the processing result; determines the element values ​​corresponding to each transaction account pair in the first group from the first similarity matrix, and uses the sum of the determined element values ​​as the fitness value of the first group.

[0157] In one embodiment of the present invention, the risk account identification device further includes an optimization module 404, which is further configured to: generate multiple third groups based on historical transaction data, wherein the third groups include multiple historical transaction account pairs, and the historical transaction account pairs are composed of transaction accounts from two payment systems; calculate the fitness value of the third groups based on the second similarity matrix corresponding to the historical transaction data of the two payment systems, and iteratively optimize the third groups based on the calculated fitness value; verify the iterative optimization result of the third groups using pre-labeled risk accounts; and adjust the crossover strategy and / or mutation strategy if the verification fails.

[0158] In one embodiment of the present invention, the matrix calculation module 402 is configured to: calculate the distance between a pair of transaction accounts in the first transaction matrix and the second transaction matrix according to the transaction information, normalize the distance, and use the reciprocal or negative value of the processing result as an element in the first similarity matrix;

[0159] The account identification module 403 is configured to: determine the distribution of zero elements of the first similarity matrix based on the minimum value of each row and the minimum value of each column in the first similarity matrix, and determine the account matching pair based on the distribution of zero elements.

[0160] In one embodiment of the present invention, the account identification module 403 is configured to: calculate a transformation matrix based on the difference between each element in each row of the first similarity matrix and the minimum value of that row, and the difference between each element in each column and the minimum value of that column; mark the zero elements in the transformation matrix row by row as first matching labels, and mark the other zero elements in the same column as the first matching labels as second matching labels; determine whether the number of first matching labels is the same as the number of rows or columns of the first similarity matrix, and if so, take the transaction account pair corresponding to the first matching label as the account matching pair.

[0161] In one embodiment of the present invention, the matrix calculation module 402 is configured to: convert the transaction amounts in the transaction data corresponding to the two payment systems into the same currency; select transaction data from the converted transaction data that meets the time window and whose transaction amount is greater than the amount threshold according to a preset time window; and generate a first transaction matrix and a second transaction matrix based on the selected transaction data.

[0162] In one embodiment of the present invention, the matrix calculation module 402 is configured to: when the number of transaction accounts selected by the two payment systems is different, configure virtual accounts and virtual transaction information corresponding to the virtual accounts for the payment system with a smaller number of transaction accounts, wherein the transaction amount in the virtual transaction information is less than the amount threshold, and the number of virtual accounts is the difference between the number of transaction accounts selected by the two payment systems; and generate a first transaction matrix and a second transaction matrix based on the virtual accounts, virtual transaction information, and selected transaction data.

[0163] As can be seen from the risk account identification device provided in the above embodiments, a first similarity matrix is ​​calculated for the two payment systems based on their respective transaction data. Based on this first similarity matrix, a matching algorithm is used to determine account matching pairs from the two payment systems, and then risk accounts are identified based on these account matching pairs. Therefore, based on the transaction similarity between the two payment systems, the matching algorithm accurately identifies account pairs with potential wash trading characteristics. Compared to related technologies that only use monetary amounts to identify risk accounts, this method improves the ability to identify complex transaction information, increases the accuracy of risk account identification, and provides strong support for tracking risk clues in payment systems.

[0164] Figure 5 An exemplary system architecture 500 is shown that can be applied to the risk account identification method or risk account identification device of the present invention.

[0165] like Figure 5 As shown, the system architecture 500 may include servers 501 and 502 for payment, server 503 for identifying risky accounts, and network 504. Network 504 serves as the medium for providing communication links between servers 501, 502, and 503. Network 504 may include various connection types, such as wired or wireless communication links or fiber optic cables. Servers 501 and 502 for payment can interact with payment information through network 504 to facilitate transactions between different accounts and store transaction data. Server 503 for identifying risky accounts can obtain transaction data from servers 501 and 502 through network 504 and identify risky accounts based on the transaction data.

[0166] It should be understood that Figure 5 The number of servers and networks shown is merely illustrative. Depending on implementation needs, there can be any number of networks and servers.

[0167] The following is for reference. Figure 6 It shows a schematic diagram of the structure of a computer system 600 suitable for implementing a terminal device of the present invention. Figure 6 The terminal device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0168] like Figure 6 As shown, the computer system 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 602 or programs loaded from storage section 608 into random access memory (RAM) 603. The RAM 603 also stores various programs and data required for the operation of the system 600. The CPU 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0169] The following components are connected to I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to I / O interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 610 as needed so that computer programs read from it can be installed into storage section 608 as needed.

[0170] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 609, and / or installed from removable medium 611. When the computer program is executed by central processing unit (CPU) 601, it performs the functions defined above in the system of this invention.

[0171] It should be noted that the computer-readable medium shown in this invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0172] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0173] The modules described in the embodiments of the present invention can be implemented in software or hardware. The described system can also be housed in a processor; for example, it can be described as a processor including a data acquisition module, a matrix calculation module, and an account identification module. The names of these modules do not necessarily limit the module itself; for example, the account identification module can also be described as a "module for identifying risky accounts."

[0174] In another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiments; or it may exist independently and not assembled into the device. The computer-readable medium carries one or more programs, which, when executed by the device, cause the device to include: acquiring transaction data corresponding to two payment systems, the transaction data including multiple transaction accounts and transaction information of each transaction account; generating a first transaction matrix and a second transaction matrix corresponding to the two payment systems based on the transaction data; wherein the rows of the first and second transaction matrices respectively represent the transaction accounts of their corresponding payment systems, and the columns of the first and second transaction matrices respectively represent the transaction information of their corresponding payment systems; calculating a first similarity matrix corresponding to the two payment systems based on the first and second transaction matrices; determining account matching pairs from the two payment systems based on the first similarity matrix and a matching algorithm; and determining risk accounts based on the account matching pairs.

[0175] According to the technical solution of this invention, a first similarity matrix is ​​calculated for the two payment systems based on their respective transaction data. Based on this first similarity matrix, a matching algorithm is used to determine account matching pairs from the two payment systems, and then risky accounts are identified based on these account matching pairs. Thus, based on the transaction similarity between the two payment systems, the matching algorithm accurately identifies account pairs with potential wash trading characteristics. Compared to related technologies that only use monetary amounts to identify risky accounts, this method improves the ability to identify complex transaction information, increases the accuracy of risky account identification, and provides strong support for tracking risk clues in payment systems.

[0176] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for identifying risky accounts, characterized in that, include: Obtain transaction data corresponding to the two payment systems respectively. The transaction data includes multiple transaction accounts and transaction information of each transaction account. Based on the transaction data, a first transaction matrix and a second transaction matrix corresponding to the two payment systems are generated respectively; wherein, the rows of the first transaction matrix and the second transaction matrix represent the transaction accounts of their respective payment systems, and the columns of the first transaction matrix and the second transaction matrix represent the transaction information of their respective payment systems; Based on the first transaction matrix and the second transaction matrix, calculate the first similarity matrix corresponding to the two payment systems; Based on the first similarity matrix and the matching algorithm, account matching pairs are determined from the two payment systems; Risky accounts are identified based on the account matching pairs.

2. The method according to claim 1, characterized in that, The step of determining account matching pairs from the two payment systems based on the first similarity matrix and the matching algorithm includes: Based on a preset population size, multiple first populations are generated. Each first population includes multiple pairs of trading accounts, and each pair of trading accounts is composed of trading accounts from the first trading matrix and the second trading matrix. Calculate the fitness value of each of the first populations based on the element values ​​in the first similarity matrix; Based on the fitness value, the first population is iteratively optimized, and the account matching pairs are determined based on the iterative optimization results.

3. The method according to claim 2, characterized in that, The step of iteratively optimizing the first population based on the fitness value includes: Using the multiple first populations as the current population, the following steps are executed iteratively until the iterative optimization condition is met: Based on the fitness value, multiple second populations are selected from the current population, wherein the fitness value of the second population is not less than the fitness value of the other unselected current populations; Based on the crossover and mutation strategies, crossover and mutation processes are performed on the transaction account pairs in each of the second groups to generate a new current group; the crossover and mutation strategies are optimized based on the historical transaction data of the two payment systems.

4. The method according to claim 2, characterized in that, Also includes: For each of the transaction account pairs: calculate the distance between the transaction account pairs based on the transaction information, normalize the distance, and obtain the element values ​​in the first similarity matrix based on the processing result; The step of calculating the fitness value of each of the first populations based on the element values ​​in the first similarity matrix includes: The element values ​​corresponding to each pair of trading accounts in the first population are determined from the first similarity matrix, and the sum of the determined element values ​​is used as the fitness value of the first population.

5. The method according to claim 3, characterized in that, Also includes: Based on the historical transaction data, multiple third groups are generated, each including multiple historical transaction account pairs, which are composed of transaction accounts from two payment systems. Based on the second similarity matrix corresponding to the historical transaction data of the two payment systems, the fitness value of the third population is calculated, and the third population is iteratively optimized based on the calculated fitness value. The iterative optimization results of the third group are verified using pre-labeled risk accounts; If the verification fails, adjust the crossover strategy and / or the mutation strategy.

6. The method according to claim 1, characterized in that, The step of calculating the first similarity matrix corresponding to the two payment systems based on the first transaction matrix and the second transaction matrix includes: Based on the transaction information, the distance between a pair of transaction accounts in the first transaction matrix and the second transaction matrix is ​​calculated, the distance is normalized, and the reciprocal or negative value of the processing result is used as an element in the first similarity matrix; Based on the first similarity matrix and the matching algorithm, account matching pairs are determined from the two payment systems, including: The distribution of zero elements in the first similarity matrix is ​​determined based on the minimum value of each row and the minimum value of each column, and the account matching pairs are determined based on the distribution of zero elements.

7. The method according to claim 6, characterized in that, The step of determining the zero-element distribution of the first similarity matrix based on the minimum value of each row and the minimum value of each column in the first similarity matrix, and determining the account matching pair based on the zero-element distribution, includes: Calculate the transformation matrix based on the differences between each element in each row of the first similarity matrix and the minimum value of that row, and the differences between each element in each column and the minimum value of that column; Each row of the transformation matrix is ​​labeled with a first matching label, and the other zero elements in the same column as the first matching label are labeled with a second matching label. Determine whether the number of the first matching tags is the same as the number of rows or columns of the first similarity matrix. If so, take the transaction account pair corresponding to the first matching tags as the account matching pair.

8. The method according to claim 1, characterized in that, Based on the transaction data, a first transaction matrix and a second transaction matrix corresponding to the two payment systems are generated, including: Convert the transaction amounts in the transaction data corresponding to the two payment systems into the same currency; According to the preset time window, select transaction data from the transformed transaction data that meets the time window and whose transaction amount is greater than the amount threshold; Based on the selected transaction data, the first transaction matrix and the second transaction matrix are generated.

9. The method according to claim 8, characterized in that, When the number of transaction accounts selected by the two payment systems is different, the step of generating the first transaction matrix and the second transaction matrix based on the selected transaction data includes: For payment systems with a small number of transaction accounts, virtual accounts and corresponding virtual transaction information are configured. The transaction amount in the virtual transaction information is less than the amount threshold, and the number of virtual accounts is the difference between the number of transaction accounts selected by the two payment systems. The first transaction matrix and the second transaction matrix are generated based on the virtual account, the virtual transaction information, and the selected transaction data.

10. A risk account identification device, characterized in that, include: The module comprises a data acquisition module, a matrix calculation module, and an account recognition module; among which, The data acquisition module is configured to acquire transaction data corresponding to the two payment systems respectively, wherein the transaction data includes multiple transaction accounts and transaction information of each transaction account; The matrix calculation module is configured to: generate a first transaction matrix and a second transaction matrix corresponding to the two payment systems respectively, based on the transaction data; wherein, the rows of the first transaction matrix and the second transaction matrix represent the transaction accounts of their respective payment systems, and the columns of the first transaction matrix and the second transaction matrix represent the transaction information of their respective payment systems; and calculate a first similarity matrix corresponding to the two payment systems based on the first transaction matrix and the second transaction matrix. The account identification module is configured to: determine account matching pairs from the two payment systems based on the first similarity matrix and the matching algorithm; and determine risky accounts based on the account matching pairs.

11. An electronic device for identifying risky accounts, characterized in that, include: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the risk account identification method as described in any one of claims 1-9.

12. A computer-readable medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the risk account identification method as described in any one of claims 1-9.

Images