Electronic payment transaction method and system

Abstract

The invention relates to the technical field of electronic payment, in particular to an electronic payment transaction method and system. The method comprises the following steps: obtaining historical transaction data of a POS machine, and carrying out abnormal transaction data extraction on the historical transaction data of the POS machine to obtain historical abnormal transaction data of the POS machine; performing standardization processing on historical abnormal transaction data of the POS machine and constructing an abnormal transaction detection model; obtaining transaction request data, and performing first security detection on the transaction request data to obtain data to be subjected to secondary detection; performing abnormal transaction prediction on the data to be detected for the second time through an abnormal transaction detection model to obtain abnormal transaction prediction data; performing second security detection on the to-be-secondarily detected data through the abnormal transaction prediction data to obtain a security detection result; and performing risk assessment on the transaction request data according to the security detection result to obtain risk assessment data, and sending the risk assessment data to the POS machine transaction system. According to the invention, the secure transaction of the offline POS machine is realized based on the data mining technology.

Description

Technical Field

[0001] The present invention relates to the field of payment technology, and in particular to an electronic payment transaction method and system. Background Art

[0002] Traditional POS systems rely on real-time network connections to interact with payment gateways to complete transaction authorization and processing. However, this reliance on network connections may be limited in certain environments, such as areas with unstable networks or no network coverage. Therefore, a solution that can conduct transactions in an offline environment is needed to ensure payment security. Summary of the invention

[0003] Based on this, it is necessary for the present invention to provide an electronic payment transaction method to solve at least one of the above technical problems.

[0004] To achieve the above object, an electronic payment transaction method comprises the following steps:

[0005] Step S1: Acquire historical transaction data of the POS machine, extract abnormal transaction data from the historical transaction data of the POS machine, and thus obtain historical abnormal transaction data of the POS machine;

[0006] Step S2: standardize the historical abnormal transaction data of the POS machine to obtain standardized historical abnormal transaction data, and construct an abnormal transaction detection model for the standardized historical abnormal transaction data based on an integrated learning method;

[0007] Step S3: Acquire transaction request data, and perform a first security check on the transaction request data, thereby obtaining data to be subjected to a second check;

[0008] Step S4: using the abnormal transaction detection model to perform abnormal transaction prediction on the secondary detection data, thereby obtaining abnormal transaction prediction data;

[0009] Step S5: performing a second security test on the data to be secondary tested using the abnormal transaction prediction data, thereby obtaining a security test result;

[0010] Step S6: Perform risk assessment on the transaction request data according to the security detection result, thereby obtaining risk assessment data, and sending it to the POS transaction system to execute the offline classification transaction task.

[0011] The present invention obtains POS machine historical transaction data through the database of the POS system, extracts abnormal transaction data from the POS machine historical transaction data, and thus obtains POS machine historical abnormal transaction data; it can identify potential abnormal behaviors and fraudulent transactions, and improve transaction security and risk management capabilities. At the same time, a sample set of abnormal transactions can be established for subsequent abnormal transaction detection and prediction. By standardizing the historical abnormal transaction data of the POS machine and constructing an abnormal transaction detection model based on an integrated learning method, the accuracy and robustness of abnormal transaction detection can be improved. This helps to reduce the false alarm rate and the missed alarm rate, and provide more effective abnormal transaction monitoring and prediction functions. By acquiring transaction request data through the input device of the POS machine and performing a first security test, the transaction request can be preliminarily screened, obvious security risks are eliminated, the system's recognition rate of abnormal transactions is improved, and unnecessary transaction risks are avoided. By performing abnormal transaction prediction on the secondary detection data through the abnormal transaction detection model, the transaction request can be more accurately abnormally determined, and the system's detection and prediction performance of abnormal transactions can be improved. By using the abnormal transaction prediction data to perform a second security test on the secondary detection data, the abnormal transaction can be further confirmed and refined security screening can be performed. By performing risk assessment on transaction request data based on security detection results, the risk level of transactions can be evaluated and classified so that appropriate risk management measures can be taken. This helps protect the system and users from fraud and security risks.

[0012] Optionally, step S1 specifically includes:

[0013] Step S11: Obtaining historical transaction data of POS machine;

[0014] Step S12: extracting platform transaction data and bank card transaction data from the POS machine historical transaction data, thereby obtaining the POS machine platform historical transaction data and the POS machine bank card historical transaction data;

[0015] Step S13: extracting abnormal transaction data of the POS platform from the historical transaction data of the POS platform, thereby obtaining the abnormal transaction data of the POS platform;

[0016] Step S14: extracting abnormal bank card transaction data from the POS machine bank card historical transaction data, thereby obtaining abnormal bank card transaction data from the POS machine;

[0017] Step S15: Time-sequentially merge the historical abnormal transaction data of the POS platform and the historical abnormal transaction data of the POS bank card to obtain the historical abnormal transaction data of the POS.

[0018] The present invention obtains POS machine historical transaction data through the database of the POS system, which can be used for subsequent data analysis and abnormal transaction detection, and is helpful to discover abnormal transaction behavior and improve the security of the transaction system. By extracting the POS machine historical transaction data, separating the platform transaction data and the bank card transaction data, different types of transactions can be analyzed and monitored more clearly, which is convenient for subsequent abnormal transaction detection and risk management. By extracting abnormal transaction data from the POS machine platform historical transaction data, abnormal behaviors in the platform historical transactions, such as abnormal amounts, frequent transactions, etc., can be identified, and the detection capability of abnormal platform transactions can be improved, and the security of the transaction system can be further strengthened. By extracting abnormal transaction data from the POS machine bank card historical transaction data, abnormal behaviors in bank card transactions can be identified, such as stealing, forging cards, etc., and the detection and prevention measures of abnormal bank card transactions can be improved to protect the interests of cardholders. By performing time series merging on the POS machine platform historical abnormal transaction data and the POS machine bank card historical abnormal transaction data, abnormal transaction data from different sources can be integrated, and a comprehensive abnormal transaction data view is provided, which is convenient for comprehensive analysis, model training and prediction, thereby enhancing the discovery and prevention capabilities of abnormal transactions.

[0019] Optionally, step S13 is specifically:

[0020] Step S131: Statistically analyzing the historical transaction data of the POS platform, thereby obtaining historical abnormal transaction data of the high-frequency platform and historical abnormal transaction data of the low-frequency platform;

[0021] Step S132: Performing potential platform abnormal transaction detection on the historical transaction data of the POS platform, thereby obtaining potential platform historical abnormal transaction data;

[0022] Step S133: Merge the historical abnormal transaction data of the potential platform, the historical abnormal transaction data of the high-frequency platform and the historical abnormal transaction data of the low-frequency platform in time series, so as to obtain the historical abnormal transaction data of the POS platform.

[0023] The present invention performs statistical analysis on the historical transaction data of the POS platform, thereby obtaining high-frequency platform historical abnormal transaction data and low-frequency platform historical abnormal transaction data; through statistical analysis, abnormal transactions with high and low frequencies in the platform historical transaction data can be identified. These abnormal transactions may be the result of fraud, system failure, data error or other abnormal operations. By obtaining abnormal transaction data, the platform can be helped to manage risks. High-frequency abnormal transactions may be loopholes or fraud in the system, which need to be handled and prevented in time. Low-frequency abnormal transactions may be emerging risks or abnormalities in specific circumstances, which also need to be monitored and managed. Potential platform abnormal transaction detection is performed on the historical transaction data of the POS platform, thereby obtaining potential platform historical abnormal transaction data; in addition to obvious abnormal transactions, potential abnormal patterns in the platform historical data can also be identified through the potential abnormal transaction detection method. These abnormal patterns may not be easily discovered by traditional statistical analysis methods, so they can provide additional abnormal identification capabilities. By obtaining potential abnormal transaction data, the platform can discover potential risks and loopholes early and take corresponding early warning and preventive measures to protect the safety of the platform and users. The historical abnormal transaction data of potential platforms, high-frequency platforms, and low-frequency platforms are merged in time series to obtain the historical abnormal transaction data of POS platforms. Through time series merging, different categories of abnormal transaction data are integrated together to provide a comprehensive data set of historical abnormal transactions on the platform. This comprehensive data set can be used for further analysis, modeling, and prediction, which helps to improve the efficiency of data processing and analysis. The platform can more conveniently further process, visualize, and analyze abnormal transaction data in order to formulate corresponding strategies and measures.

[0024] Optionally, step S132 is specifically:

[0025] The historical transaction data of the POS platform is calculated using the platform potential anomaly score formula to obtain the platform potential anomaly score data. The platform potential anomaly score formula is as follows:

[0026]

[0027] Where S p is the potential anomaly score of the platform, n is the number of historical platform transaction data, i is the platform transaction sequence number, t is the time, and a i is the amount of the ith platform transaction, b i is the average amount of the ith platform transaction, c is the abnormality sensitivity coefficient, d i is the difference between the time of the ith platform transaction and the average time, e i is the refund rate of the ith platform transaction, f i is the coupon usage rate of the ith platform transaction, g iis the payment method coefficient of the i-th platform transaction;

[0028] The present invention constructs a platform potential anomaly score formula for calculating the historical transaction data of the POS platform. The formula fully considers the factors that affect the platform potential anomaly score S p The number of historical platform transaction data n, platform transaction sequence number i, time t, and the amount of the i-th platform transaction a i , the average amount of the ith platform transaction b i , abnormality sensitivity coefficient c, the difference between the time of the ith platform transaction and the average time d i , the refund rate e of the i-th platform transaction i , the coupon usage rate f of the i-th platform transaction i , the payment method coefficient g of the i-th platform transaction i , forming a functional relationship:

[0029]

[0030] in This part calculates the average of the logarithm of the rate of change of the transaction amount relative to the average amount. It indicates the speed of change of the transaction amount. If the speed of change of the transaction amount is large, it may mean that an abnormal situation has occurred. This part uses a sine function to represent the difference between the transaction time and the average time. When the deviation of the transaction time is large, the value of this part will increase, indicating a potential anomaly. i This section shows the refund rate of transactions. A high refund rate may mean that there is an anomaly, as a high refund rate may be associated with fake transactions, fraud, and other issues. i This section indicates the percentage of transactions that used coupons. If the percentage of transactions that used coupons was unusually high or low, it could indicate unusual behavior or a system bug. i This section represents the coefficient of the payment method in the transaction. Different payment methods may have different risk characteristics, so the coefficient of the payment method can be used to measure the risk of the transaction. The formula takes into account multiple factors such as the rate of change of transaction amount, transaction time difference, refund rate, coupon usage rate, payment method, etc. By combining these factors into a single score, a comprehensive assessment can be provided, which does not rely on a single factor to judge the abnormality of the transaction. The mathematical calculations in the formula can quantify the impact of different factors and incorporate them into the score. This quantitative assessment can provide an objective way to measure the abnormality of the transaction.

[0031] Conduct statistical analysis on the platform's potential anomaly score data to obtain the platform's potential anomaly threshold;

[0032] The platform's potential anomaly score data is classified and calculated according to the platform's potential anomaly threshold, thereby obtaining the potential platform's historical abnormal transaction data.

[0033] The present invention calculates the historical transaction data of the POS machine platform through the platform potential abnormality score formula, thereby obtaining the platform potential abnormality score data, and this score can reflect the abnormal nature and degree of the transaction. Through the platform potential abnormality score, some transactions with high scores can be obtained, and these transactions may involve abnormal behavior or risks on the platform. These high-scoring transactions can provide information on abnormal hotspots, which is helpful for more targeted abnormal transaction detection and management. The platform potential abnormality score data is statistically analyzed to obtain the platform potential abnormality threshold; by statistically analyzing the platform potential abnormality score, a suitable threshold can be determined as a criterion for determining abnormalities. This threshold can be adjusted according to demand and risk tolerance to distinguish abnormal transactions from normal transactions. By setting a threshold to divide abnormal and normal transactions, more refined abnormal identification can be performed. Transaction scores exceeding the threshold are marked as abnormal transactions, which can improve the accuracy of abnormal transactions and reduce false positives. By applying the platform potential abnormality threshold to classify the score data, potential abnormal transactions and normal transactions can be distinguished. In this way, potential platform historical abnormal transaction data can be obtained for further analysis and processing.

[0034] Optionally, step S14 is specifically:

[0035] Step S141: Statistically analyzing the historical transaction data of POS bank cards, thereby obtaining the historical abnormal transaction data of high-frequency bank cards and the historical abnormal transaction data of low-frequency bank cards;

[0036] Step S142: Performing potential bank card abnormal transaction detection on the POS machine bank card historical transaction data, thereby obtaining potential bank card historical abnormal transaction data;

[0037] Step S143: Time-sequentially merge the potential bank card historical abnormal transaction data, the high-frequency bank card historical abnormal transaction data and the low-frequency bank card historical abnormal transaction data to obtain the POS bank card historical abnormal transaction data.

[0038] The present invention can find high-frequency and low-frequency abnormal transaction data by statistically analyzing the historical transaction data of bank cards. This helps to identify potential abnormal transaction behaviors, such as large transactions, frequent transactions, abnormal time, etc. After identifying high-frequency and low-frequency abnormal transaction data, risk management can be better performed. High-frequency abnormal transaction data may involve risks such as fraud and theft, while low-frequency abnormal transaction data may contain atypical transaction patterns. Through statistical analysis, these abnormal transactions can be monitored and managed more accurately. Potential bank card abnormal transaction detection is performed on the historical transaction data of POS bank cards to obtain potential bank card historical abnormal transaction data; through potential bank card abnormal transaction detection, the detection accuracy of abnormal transactions can be further improved. This can help the platform better identify transactions with potential risks and reduce false positives and false negatives. Potential bank card abnormal transaction detection can discover some hidden abnormal transaction patterns, which may not be easy to determine through conventional statistical analysis methods. Therefore, it can help identify more types of abnormal transaction behaviors. Merge the potential bank card historical abnormal transaction data, high-frequency bank card historical abnormal transaction data, and low-frequency bank card historical abnormal transaction data in time series to obtain the POS bank card historical abnormal transaction data. Merging different types of abnormal transaction data can provide a more comprehensive abnormal transaction data set, including high-frequency, low-frequency, and potential abnormal transactions. Such a data set facilitates deeper analysis and model training, thereby improving the accuracy and generalization of abnormal transaction detection. By merging different types of abnormal transaction data in time series, the evolution and trend of transaction behavior can be observed. This helps to discover the periodicity, seasonality, or other time-related laws of abnormal transaction behavior, thereby increasing the understanding of abnormal behavior.

[0039] Optionally, step S142 is specifically:

[0040] The bank card potential anomaly score formula is used to calculate the historical transaction data of the POS bank card to obtain the bank card potential anomaly score. The bank card potential anomaly score formula is as follows:

[0041]

[0042] Where S y is the potential anomaly score of the bank card, l is the number of historical bank card transaction data, k is the sequence number of the bank card transaction, t is the time, and a k is the amount of the kth bank card transaction, b k is the distance between the location of the kth bank card transaction and the common location, c k is the coefficient of difference between the time of the kth bank card transaction and the usual time, d k is the difference coefficient between the frequency of the kth bank card transaction and the common frequency, e kis the difference coefficient between the type of the kth transaction and the common type, f k The merchant credit score for the kth transaction;

[0043] The present invention constructs a bank card potential anomaly score formula for calculating the historical transaction data of POS bank cards. The formula fully considers the factors that affect the bank card potential anomaly score S y The number of historical bank card transaction data l, the sequence number of the bank card transaction k, the time t, and the amount of the kth bank card transaction a k , the distance b between the location of the kth bank card transaction and the common location k , the difference coefficient c between the time of the kth bank card transaction and the usual time k , the difference coefficient d between the frequency of the kth bank card transaction and the common frequency k , the difference coefficient e between the type of the kth transaction and the common type k , the merchant reputation score f of the kth transaction k , forming a functional relationship:

[0044]

[0045] The formula can help banks or financial institutions detect potential abnormal transactions. By considering multiple factors such as transaction amount, location, time, frequency, type and merchant reputation, the formula can comprehensively evaluate the degree of abnormality of each transaction and derive a score for potential abnormal transactions. This helps banks identify possible fraud, theft or other illegal transactions in advance, thereby protecting customers' funds. The formula can be used as a tool to help banks strengthen security and preventive measures. By comprehensively analyzing and evaluating transaction data, banks can identify and monitor potential abnormal transaction patterns and behaviors. This helps banks improve anti-fraud systems and risk management strategies, increase transaction security, and reduce the risk of financial crimes.

[0046] Perform statistical analysis on bank card potential anomaly score data to obtain bank card potential anomaly thresholds;

[0047] The bank card potential abnormality score data is classified and calculated according to the bank card potential abnormality threshold, so as to obtain the potential bank card historical abnormal transaction data.

[0048] The present invention calculates the historical transaction data of a POS bank card by a bank card potential abnormality score formula, and can use the potential abnormality score formula to evaluate and calculate the historical transaction data of the bank card to obtain the potential abnormality score of each transaction. In this way, the bank can quantitatively evaluate each transaction, identify the potential abnormal transactions that may exist, and mark their abnormality. This helps to improve the accuracy and efficiency of automated abnormal transaction detection. The bank card potential abnormality score data is statistically analyzed to obtain the bank card potential abnormality threshold; the bank can obtain statistical indicators such as the distribution, mean, and standard deviation of the abnormal score. These statistical indicators can be used as a basis for determining the bank card potential abnormality threshold. By setting a suitable abnormality threshold, the bank can customize the boundary of abnormal transactions according to specific circumstances to better meet customer needs and reduce the false alarm rate and the missed alarm rate. The bank card potential abnormality score data is classified and calculated according to the bank card potential abnormality threshold, so as to obtain the potential bank card historical abnormal transaction data. By comparing the bank card potential abnormality score data with the bank card potential abnormality threshold, the transaction data can be divided into two categories: potential abnormalities and normal transactions. This helps banks identify and focus on potential abnormal transaction data and provide more accurate and efficient abnormal transaction management. By obtaining potential historical abnormal bank card transaction data, banks can conduct further detailed analysis and investigation to confirm whether they are actual abnormal transactions and take appropriate measures to protect customer assets and reduce financial risks.

[0049] Optionally, step S3 specifically includes:

[0050] Step S31: Obtain transaction request data;

[0051] Step S32: extracting the customer transaction voucher from the transaction request data, thereby obtaining the customer transaction voucher information;

[0052] Step S33: extracting basic information of the customer from the customer transaction voucher information, thereby obtaining basic information of the customer;

[0053] Step S34: extracting customer blacklist from the historical abnormal transaction data of the POS machine by frequency analysis, thereby obtaining customer blacklist data;

[0054] Step S35: classify and calculate the customer's basic information according to the customer blacklist data. If the customer's basic information is classified as a blacklist customer, it is sent to the POS transaction system to stop the current transaction task; if the customer's basic information is classified as a normal customer, data for secondary detection is generated.

[0055] The present invention obtains transaction request data through the input device of the POS machine; obtaining accurate and complete transaction request data is the basis for subsequent processing and analysis, which is helpful to understand the situation and characteristics of the transaction. By extracting the customer transaction voucher information in the transaction request data, key information related to the customer, such as bank card number, transaction amount, transaction time, etc., can be obtained. This helps to associate the transaction with a specific customer and perform subsequent customer information analysis and processing. The customer basic information is extracted from the customer transaction voucher information to obtain the customer basic information; the customer basic information, such as customer name, ID card number, mobile phone number, etc., is extracted from the customer transaction voucher information. Obtaining the customer's basic information is helpful for further customer risk assessment and determination. By applying the frequency analysis method, the historical abnormal transaction data of the POS machine is analyzed, and specific customers with abnormally high frequency of occurrence can be identified, and these customers may have abnormal behavior or risks. Extracting customer blacklist data is helpful to establish a customer risk library for subsequent risk control and management. According to the customer blacklist data, the customer basic information is classified and calculated, and the customer can be divided into two categories: blacklist customers and normal customers. For blacklist customers, measures can be taken immediately to stop the current transaction task to prevent further risks. For normal customers, data for secondary testing is generated to provide support for subsequent refined risk assessment and decision-making.

[0056] Optionally, step S5 specifically includes:

[0057] The abnormal transaction prediction data is calculated by the compliance calculation formula to obtain the predicted transaction compliance data, wherein the compliance calculation formula is specifically as follows:

[0058]

[0059] C is the compliance index coefficient, T is the observation period, v is the total number of abnormal transactions, w is the abnormal transaction sequence number, x w is the ratio of the amount of the wth abnormal transaction to the average transaction amount;

[0060] The present invention constructs a compliance calculation formula for calculating abnormal transaction prediction data. The formula fully considers the observation time period T, the total number of abnormal transactions v, the abnormal transaction sequence w, the ratio of the amount of the wth abnormal transaction to the average transaction amount x, which affects the compliance index coefficient C. w , forming a functional relationship:

[0061]

[0062] in Represents the ratio of the amount of each abnormal transaction to the average transaction amount x wMultiply them one by one. In this way, we can get an overall growth rate, because each ratio greater than 1 means that the abnormal transaction amount is higher than the average level, and each ratio less than 1 means that the abnormal transaction amount is lower than the average level. This part of the calculation takes the vth root of the cumulative multiplication result, where v is the total number of abnormal transactions. This gives the average growth rate of each abnormal transaction. This part indicates that when the total number of abnormal transactions tends to infinity, the average growth rate of each abnormal transaction is calculated to the limit. This can take into account the impact of potential more abnormal transactions on compliance, and then perform logarithmic operations on this part to obtain compliance indicators. Logarithmization can transform exponentially growing data into a form that is easier to compare and analyze. Finally, a derivative is performed, which means that the compliance indicator is differentiated to obtain the rate of change of the compliance indicator over time. This can be used to monitor compliance trends and promptly detect the increase or decrease of abnormal transactions. The formula takes into account factors such as the amount ratio, quantity and time changes of abnormal transactions, and obtains a comprehensive compliance indicator by performing mathematical operations on these factors. This makes it more accurate and comprehensive to evaluate and compare compliance in different time periods or different organizations.

[0063] The secondary inspection data is marked for security based on the predicted transaction compliance data to obtain the security inspection results.

[0064] The present invention uses a compliance calculation formula to calculate abnormal transaction prediction data, aiming to evaluate the compliance degree of transactions. Through calculation, the predicted compliance data of each transaction can be obtained, and the data can be used to determine whether the transaction complies with laws and regulations. This helps to identify potential illegal transactions, reduce compliance risks, and provide a basis for subsequent security marking and screening. According to the compliance data of the predicted transaction, the secondary detection data is marked for security. Through marking, transactions can be divided into two categories: safe and unsafe. The security detection results can be used to identify potential risks, prevent fraud and fraudulent behavior, and protect the security of customers and POS machine transaction systems. In addition, the security detection results can also be used for subsequent risk decision-making and further risk management. The amount ratio in the formula can reflect the degree of deviation of the abnormal transaction amount from the average level. Through mathematical operations such as cumulative multiplication and square root, the formula can comprehensively consider the deviation of multiple abnormal transactions. This enables the formula to help detect and identify abnormal transactions, especially when the number of abnormal transactions is large or the degree of abnormality is inconsistent. The derivative operation in the formula can calculate the rate of change of the compliance index over time. This makes it possible to perform compliance trend analysis and identify the growth or decline trend of the compliance level. This is important for monitoring compliance status, detecting growth or decrease in abnormal transactions, and taking timely measures.

[0065] Optionally, the risk assessment in step S6 is specifically:

[0066] Extract the credit rating of the customer's basic information to obtain the customer's credit rating;

[0067] Extracting transaction time from transaction request data to obtain transaction time information;

[0068] The transaction time information, security test results and customer credit rating are calculated through the risk calculation formula to obtain risk assessment data. The risk calculation formula is as follows:

[0069]

[0070] Where R is the risk coefficient, t is the time, C is the compliance index coefficient, S is the customer credit rating, j is the limit integer, T r is the transaction time, and w is the transaction time weight coefficient.

[0071] The present invention constructs a risk calculation formula for calculating transaction time information, security test results and customer credit rating. The formula fully considers the compliance index coefficient C that affects the risk coefficient R, the customer credit rating S, the transaction time T r , the transaction time weight coefficient w forms a functional relationship:

[0072]

[0073] in This part represents the derivative of the logarithm of the ratio of security and customer credit rating with respect to time. It can reflect the change of security and customer credit rating over time. The positive or negative of the derivative is related to the trend of change, which can express the rate of change of risk. This part represents the j-root of the transaction time, where j approaches infinity. It is used to represent the impact of the increase in transaction time on risk, and is weighted by multiplying it by the weight coefficient w. When the transaction time is longer, the j-root value under the square root approaches 1, indicating that the impact of risk is smaller; on the contrary, when the transaction time is shorter, the j-root value under the square root approaches 0, indicating that the impact of risk is greater. This formula comprehensively considers the security test results, customer credit rating and transaction time information, and incorporates them into the risk assessment model. This can analyze risks more comprehensively and avoid relying solely on a single indicator or factor for assessment. Therefore, the formula helps to provide a more comprehensive and integrated risk assessment result. The formula The part represents the time rate of change of security and credit rating. Such consideration can capture the impact of time on risk and model the changes of security and credit rating over time. This is beneficial for risk assessment scenarios with time sensitivity, such as financial transactions and network security. The formula uses the weight coefficient w to weight the transaction time. This allows the relative importance of transaction time to risk to be adjusted according to specific needs. Different application scenarios may have different levels of concern about the impact of transaction time. By adjusting the weight coefficient w, its contribution to the final risk assessment can be flexibly controlled.

[0074] The present invention extracts the credit rating of the basic information of the customer, thereby obtaining the credit rating of the customer; the extraction of the credit rating of the customer can provide a basis for risk management. The transaction time is extracted from the transaction request data, thereby obtaining the transaction time information; the transaction time information can be used to detect abnormal transactions. If the time of a transaction is significantly deviated from the customer's historical transaction time or regular transaction time, it may indicate the existence of risks or fraud. Timely detection and processing of these abnormal transactions helps to prevent risks and protect the safety of customer funds. The transaction time information, security test results and customer credit rating are calculated through a risk calculation formula to obtain risk assessment data. By using the risk calculation formula, the risk assessment process can be automated to improve the efficiency and accuracy of decision-making. According to the calculated risk assessment data, the system can intelligently determine whether to process the transaction or take other measures, reduce the workload of manual intervention, and reduce misjudgments caused by human factors.

[0075] Optionally, this specification also provides an electronic payment transaction system, including:

[0076] The abnormal transaction extraction module is used to obtain the historical transaction data of the POS machine, extract the abnormal transaction data from the historical transaction data of the POS machine, and thus obtain the historical abnormal transaction data of the POS machine;

[0077] A model building module is used to standardize the historical abnormal transaction data of POS machines, thereby obtaining standardized historical abnormal transaction data, and building an abnormal transaction detection model for the standardized historical abnormal transaction data based on an integrated learning method;

[0078] A first security detection module, used to obtain transaction request data, perform a first security detection on the transaction request data, and thereby obtain data to be subjected to a second detection;

[0079] The abnormal transaction prediction module is used to perform abnormal transaction prediction on the secondary detection data through the abnormal transaction detection model, thereby obtaining abnormal transaction prediction data;

[0080] A second security detection module, used to perform a second security detection on the secondary detection data through abnormal transaction prediction data, so as to obtain a security detection result;

[0081] The risk assessment module is used to perform risk assessment on transaction request data according to the security detection results, thereby obtaining risk assessment data and sending it to the POS transaction system to perform offline classification transaction tasks.

[0082] The electronic payment transaction system of the present invention can implement any electronic payment transaction method of the present invention, and is used to combine the operation and signal transmission medium between various modules to complete the electronic payment transaction method. The internal structure of the system cooperates with each other to achieve security protection for offline POS transactions. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0084] Figure 1 A schematic diagram of the steps of the electronic payment transaction method of the present invention;

[0085] Figure 2 Detailed step flow diagram of step S1 in the present invention;

[0086] Figure 3 Detailed step flow diagram of step S13 in the present invention;

[0087] Figure 4 Detailed step flow diagram of step S14 in the present invention;

[0088] Figure 5 Detailed step flow diagram of step S3 in the present invention;

[0089] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0090] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0091] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0092] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0093] To achieve this, please refer to Figures 1 to 5 The present invention provides an electronic payment transaction method, the method comprising the following steps:

[0094] Step S1: Acquire historical transaction data of the POS machine, extract abnormal transaction data from the historical transaction data of the POS machine, and thus obtain historical abnormal transaction data of the POS machine;

[0095] In this embodiment, the historical transaction data of the POS machine is obtained through the database of the POS system, and the platform abnormal transaction data and the bank card abnormal transaction data are extracted from the historical transaction data of the POS machine, so as to obtain the historical abnormal transaction data of the POS machine.

[0096] Step S2: standardize the historical abnormal transaction data of the POS machine to obtain standardized historical abnormal transaction data, and construct an abnormal transaction detection model for the standardized historical abnormal transaction data based on an integrated learning method;

[0097] In this embodiment, the historical abnormal transaction data is standardized by the Z-score standardization method, and the values ​​of different features are scaled to a unified range to eliminate the dimensional differences between different features. The standardized historical abnormal transaction data is divided into a training set and a test set. Usually, most of the data is used for model training, and a small part is used for model evaluation and verification. The AdaBoost algorithm is selected to build a model for the training set and the test set, so as to obtain an abnormal transaction detection model.

[0098] Step S3: Acquire transaction request data, and perform a first security check on the transaction request data, thereby obtaining data to be subjected to a second check;

[0099] In this embodiment, transaction request data is obtained through the input device of the POS machine, and the customer transaction voucher (such as bank card, QR code) is extracted from the transaction request data to obtain the customer transaction voucher information; the customer basic information is extracted from the customer transaction voucher information to obtain the customer basic information; the customer blacklist is extracted from the historical abnormal transaction data of the POS machine through the frequency analysis method to obtain the customer blacklist data; the customer basic information is classified and calculated according to the customer blacklist data, and if the customer basic information is classified as a blacklist customer, it is sent to the POS machine transaction system to execute the task of stopping the current transaction; if the customer basic information is classified as a normal customer, the data to be tested again is generated.

[0100] Step S4: using the abnormal transaction detection model to perform abnormal transaction prediction on the secondary detection data, thereby obtaining abnormal transaction prediction data;

[0101] In this embodiment, the data to be re-detected is subjected to feature extraction to extract the same features as the abnormal transaction detection model. The extracted feature data is standardized to ensure that it is consistent with the data range of the training model. The trained abnormal transaction detection model is used to predict the data to be re-detected to obtain the prediction results of abnormal transactions.

[0102] Step S5: performing a second security test on the data to be secondary tested using the abnormal transaction prediction data, thereby obtaining a security test result;

[0103] In this embodiment, the abnormal transaction prediction data is calculated by a compliance calculation formula to obtain predicted transaction compliance data, and the secondary detection data is marked for security based on the predicted transaction compliance data to obtain a security detection result.

[0104] Step S6: Perform risk assessment on the transaction request data according to the security detection result, thereby obtaining risk assessment data, and sending it to the POS transaction system to execute the offline classification transaction task.

[0105] In this embodiment, the credit rating of the customer's basic information is extracted to obtain the customer's credit rating; the transaction time is extracted from the transaction request data to obtain the transaction time information; the transaction time information, the security test results and the customer's credit rating are calculated by the risk calculation formula to obtain the risk assessment data, and sent to the POS machine transaction system. The POS machine transaction system classifies the transaction tasks according to the risk assessment data, wherein the classified transaction tasks include suspending the transaction tasks for high-risk (risk coefficient greater than or equal to 0.8) transactions, waiting for online transaction tasks for medium-risk (risk coefficient less than 0.8 and greater than or equal to 0.5) transactions, and offline transaction tasks for low-risk (risk coefficient less than 0.5) transactions.

[0106] The present invention obtains POS machine historical transaction data through the database of the POS system, extracts abnormal transaction data from the POS machine historical transaction data, and thus obtains POS machine historical abnormal transaction data; it can identify potential abnormal behaviors and fraudulent transactions, and improve transaction security and risk management capabilities. At the same time, a sample set of abnormal transactions can be established for subsequent abnormal transaction detection and prediction. By standardizing the historical abnormal transaction data of the POS machine and constructing an abnormal transaction detection model based on an integrated learning method, the accuracy and robustness of abnormal transaction detection can be improved. This helps to reduce the false alarm rate and the missed alarm rate, and provide more effective abnormal transaction monitoring and prediction functions. By acquiring transaction request data through the input device of the POS machine and performing a first security test, the transaction request can be preliminarily screened, obvious security risks are eliminated, the system's recognition rate of abnormal transactions is improved, and unnecessary transaction risks are avoided. By performing abnormal transaction prediction on the secondary detection data through the abnormal transaction detection model, the transaction request can be more accurately abnormally determined, and the system's detection and prediction performance of abnormal transactions can be improved. By using the abnormal transaction prediction data to perform a second security test on the secondary detection data, the abnormal transaction can be further confirmed and refined security screening can be performed. By performing risk assessment on transaction request data based on security detection results, the risk level of transactions can be evaluated and classified so that appropriate risk management measures can be taken. This helps protect the system and users from fraud and security risks.

[0107] Optionally, step S1 specifically includes:

[0108] Step S11: Obtaining historical transaction data of POS machine;

[0109] In this embodiment, the historical transaction data of the POS machine is obtained through the database of the POS system.

[0110] Step S12: extracting platform transaction data and bank card transaction data from the POS machine historical transaction data, thereby obtaining the POS machine platform historical transaction data and the POS machine bank card historical transaction data;

[0111] In this embodiment, transaction data extraction is performed on the POS machine historical transaction data, thereby obtaining the POS machine platform historical transaction data and the POS machine bank card historical transaction data.

[0112] Step S13: extracting abnormal transaction data of the POS platform from the historical transaction data of the POS platform, thereby obtaining the abnormal transaction data of the POS platform;

[0113] In this embodiment, statistical analysis is performed on the historical transaction data of the POS platform and abnormal transactions on the potential platform are detected, so as to obtain the historical abnormal transaction data of the POS platform.

[0114] Step S14: extracting abnormal bank card transaction data from the POS machine bank card historical transaction data, thereby obtaining abnormal bank card transaction data from the POS machine;

[0115] In this embodiment, statistical analysis is performed on the historical transaction data of POS bank cards and potential abnormal bank card transactions are detected, thereby obtaining the historical abnormal bank card transaction data of POS machines.

[0116] Step S15: Time-sequentially merge the historical abnormal transaction data of the POS platform and the historical abnormal transaction data of the POS bank card to obtain the historical abnormal transaction data of the POS.

[0117] In this embodiment, the historical abnormal transaction data of the POS platform and the historical abnormal transaction data of the POS bank card are merged in chronological order to obtain the historical abnormal transaction data of the POS.

[0118] The present invention obtains POS machine historical transaction data through the database of the POS system, which can be used for subsequent data analysis and abnormal transaction detection, and is helpful to discover abnormal transaction behavior and improve the security of the transaction system. By extracting the POS machine historical transaction data, separating the platform transaction data and the bank card transaction data, different types of transactions can be analyzed and monitored more clearly, which is convenient for subsequent abnormal transaction detection and risk management. By extracting abnormal transaction data from the POS machine platform historical transaction data, abnormal behaviors in the platform historical transactions, such as abnormal amounts, frequent transactions, etc., can be identified, and the detection capability of abnormal platform transactions can be improved, and the security of the transaction system can be further strengthened. By extracting abnormal transaction data from the POS machine bank card historical transaction data, abnormal behaviors in bank card transactions can be identified, such as stealing, forging cards, etc., and the detection and prevention measures of abnormal bank card transactions can be improved to protect the interests of cardholders. By performing time series merging on the POS machine platform historical abnormal transaction data and the POS machine bank card historical abnormal transaction data, abnormal transaction data from different sources can be integrated, and a comprehensive abnormal transaction data view is provided, which is convenient for comprehensive analysis, model training and prediction, thereby enhancing the discovery and prevention capabilities of abnormal transactions.

[0119] Optionally, step S13 is specifically:

[0120] Step S131: Statistically analyzing the historical transaction data of the POS platform, thereby obtaining historical abnormal transaction data of the high-frequency platform and historical abnormal transaction data of the low-frequency platform;

[0121] In this embodiment, the historical transaction data of the POS platform is statistically analyzed by frequency analysis method, so as to obtain the historical abnormal transaction data of the high-frequency platform and the historical abnormal transaction data of the low-frequency platform.

[0122] Step S132: Performing potential platform abnormal transaction detection on the historical transaction data of the POS platform, thereby obtaining potential platform historical abnormal transaction data;

[0123] In this embodiment, the historical transaction data of the POS platform is calculated using the platform potential abnormality score formula to obtain the platform potential abnormality score data, and the platform potential abnormality score data is statistically analyzed to obtain the platform potential abnormality threshold; the platform potential abnormality score data is classified and calculated according to the platform potential abnormality threshold to obtain the potential platform historical abnormal transaction data.

[0124] Step S133: Merge the historical abnormal transaction data of the potential platform, the historical abnormal transaction data of the high-frequency platform and the historical abnormal transaction data of the low-frequency platform in time series, so as to obtain the historical abnormal transaction data of the POS platform.

[0125] In this embodiment, the historical abnormal transaction data of the potential platform, the historical abnormal transaction data of the high-frequency platform and the historical abnormal transaction data of the low-frequency platform are merged in chronological order to obtain the historical abnormal transaction data of the POS platform.

[0126] The present invention performs statistical analysis on the historical transaction data of the POS platform, thereby obtaining high-frequency platform historical abnormal transaction data and low-frequency platform historical abnormal transaction data; through statistical analysis, abnormal transactions with high and low frequencies in the platform historical transaction data can be identified. These abnormal transactions may be the result of fraud, system failure, data error or other abnormal operations. By obtaining abnormal transaction data, the platform can be helped to manage risks. High-frequency abnormal transactions may be loopholes or fraud in the system, which need to be handled and prevented in time. Low-frequency abnormal transactions may be emerging risks or abnormalities in specific circumstances, which also need to be monitored and managed. Potential platform abnormal transaction detection is performed on the historical transaction data of the POS platform, thereby obtaining potential platform historical abnormal transaction data; in addition to obvious abnormal transactions, potential abnormal patterns in the platform historical data can also be identified through the potential abnormal transaction detection method. These abnormal patterns may not be easily discovered by traditional statistical analysis methods, so they can provide additional abnormal identification capabilities. By obtaining potential abnormal transaction data, the platform can discover potential risks and loopholes early and take corresponding early warning and preventive measures to protect the safety of the platform and users. The historical abnormal transaction data of potential platforms, high-frequency platforms, and low-frequency platforms are merged in time series to obtain the historical abnormal transaction data of POS platforms. Through time series merging, different categories of abnormal transaction data are integrated together to provide a comprehensive data set of historical abnormal transactions on the platform. This comprehensive data set can be used for further analysis, modeling, and prediction, which helps to improve the efficiency of data processing and analysis. The platform can more conveniently further process, visualize, and analyze abnormal transaction data in order to formulate corresponding strategies and measures.

[0127] Optionally, step S132 is specifically:

[0128] The historical transaction data of the POS platform is calculated using the platform potential anomaly score formula to obtain the platform potential anomaly score data. The platform potential anomaly score formula is as follows:

[0129]

[0130] Where S p is the potential anomaly score of the platform, n is the number of historical platform transaction data, i is the platform transaction sequence number, t is the time, and a i is the amount of the ith platform transaction, b i is the average amount of the ith platform transaction, c is the abnormality sensitivity coefficient, di is the difference between the time of the ith platform transaction and the average time, e i is the refund rate of the ith platform transaction, f i is the coupon usage rate of the ith platform transaction, g i is the payment method coefficient of the i-th platform transaction;

[0131] The present invention constructs a platform potential anomaly score formula for calculating the historical transaction data of the POS platform. The formula fully considers the factors that affect the platform potential anomaly score S p The number of historical platform transaction data n, platform transaction sequence number i, time t, and the amount of the i-th platform transaction a i , the average amount of the ith platform transaction b i , abnormality sensitivity coefficient c, the difference between the time of the ith platform transaction and the average time d i , the refund rate e of the i-th platform transaction i , the coupon usage rate f of the i-th platform transaction i , the payment method coefficient g of the i-th platform transaction i , forming a functional relationship:

[0132]

[0133] in This part calculates the average of the logarithm of the rate of change of the transaction amount relative to the average amount. It indicates the speed of change of the transaction amount. If the speed of change of the transaction amount is large, it may mean that an abnormal situation has occurred. This part uses a sine function to represent the difference between the transaction time and the average time. When the deviation of the transaction time is large, the value of this part will increase, indicating a potential anomaly. i This section shows the refund rate of transactions. A high refund rate may mean that there is an anomaly, as a high refund rate may be associated with fake transactions, fraud, and other issues. i This section indicates the percentage of transactions that used coupons. If the percentage of transactions that used coupons was unusually high or low, it could indicate unusual behavior or a system bug. iThis section represents the coefficient of the payment method in the transaction. Different payment methods may have different risk characteristics, so the coefficient of the payment method can be used to measure the risk of the transaction. The formula takes into account multiple factors such as the rate of change of transaction amount, transaction time difference, refund rate, coupon usage rate, payment method, etc. By combining these factors into a single score, a comprehensive assessment can be provided, which does not rely on a single factor to judge the abnormality of the transaction. The mathematical calculations in the formula can quantify the impact of different factors and incorporate them into the score. This quantitative assessment can provide an objective way to measure the abnormality of the transaction.

[0134] Conduct statistical analysis on the platform's potential anomaly score data to obtain the platform's potential anomaly threshold;

[0135] In this embodiment, the Z-score method is used to perform statistical analysis on the platform potential anomaly score data to obtain the platform potential anomaly threshold.

[0136] The platform's potential anomaly score data is classified and calculated according to the platform's potential anomaly threshold, thereby obtaining the potential platform's historical abnormal transaction data.

[0137] In this embodiment, the platform potential anomaly threshold is used to classify the potential anomaly score data. Data with a platform potential anomaly score exceeding the set threshold will be regarded as abnormal data, while data with a score below the threshold will be regarded as normal data, thereby obtaining potential platform historical abnormal transaction data.

[0138] The present invention calculates the historical transaction data of the POS machine platform through the platform potential abnormality score formula, thereby obtaining the platform potential abnormality score data, and this score can reflect the abnormal nature and degree of the transaction. Through the platform potential abnormality score, some transactions with high scores can be obtained, and these transactions may involve abnormal behavior or risks on the platform. These high-scoring transactions can provide information on abnormal hotspots, which is helpful for more targeted abnormal transaction detection and management. The platform potential abnormality score data is statistically analyzed to obtain the platform potential abnormality threshold; by statistically analyzing the platform potential abnormality score, a suitable threshold can be determined as a criterion for determining abnormalities. This threshold can be adjusted according to demand and risk tolerance to distinguish abnormal transactions from normal transactions. By setting a threshold to divide abnormal and normal transactions, more refined abnormal identification can be performed. Transaction scores exceeding the threshold are marked as abnormal transactions, which can improve the accuracy of abnormal transactions and reduce false positives. By applying the platform potential abnormality threshold to classify the score data, potential abnormal transactions and normal transactions can be distinguished. In this way, potential platform historical abnormal transaction data can be obtained for further analysis and processing.

[0139] Optionally, step S14 is specifically:

[0140] Step S141: Statistically analyzing the historical transaction data of POS bank cards, thereby obtaining the historical abnormal transaction data of high-frequency bank cards and the historical abnormal transaction data of low-frequency bank cards;

[0141] In this embodiment, the frequency analysis method is used to perform statistical analysis on the historical transaction data of the POS bank card, so as to obtain the high-frequency bank card historical abnormal transaction data and the low-frequency bank card historical abnormal transaction data.

[0142] Step S142: Performing potential bank card abnormal transaction detection on the POS machine bank card historical transaction data, thereby obtaining potential bank card historical abnormal transaction data;

[0143] In this embodiment, the bank card potential anomaly score formula is used to calculate the historical transaction data of the POS bank card to obtain the bank card potential anomaly score, and the bank card potential anomaly score data is statistically analyzed to obtain the bank card potential anomaly threshold; the bank card potential anomaly score data is classified and calculated according to the bank card potential anomaly threshold to obtain potential bank card historical abnormal transaction data.

[0144] Step S143: Time-sequentially merge the potential bank card historical abnormal transaction data, the high-frequency bank card historical abnormal transaction data and the low-frequency bank card historical abnormal transaction data to obtain the POS bank card historical abnormal transaction data.

[0145] In this embodiment, potential bank card historical abnormal transaction data, high-frequency bank card historical abnormal transaction data and low-frequency bank card historical abnormal transaction data are merged in chronological order to obtain POS bank card historical abnormal transaction data.

[0146] The present invention can find high-frequency and low-frequency abnormal transaction data by statistically analyzing the historical transaction data of bank cards. This helps to identify potential abnormal transaction behaviors, such as large transactions, frequent transactions, abnormal time, etc. After identifying high-frequency and low-frequency abnormal transaction data, risk management can be better performed. High-frequency abnormal transaction data may involve risks such as fraud and theft, while low-frequency abnormal transaction data may contain atypical transaction patterns. Through statistical analysis, these abnormal transactions can be monitored and managed more accurately. Potential bank card abnormal transaction detection is performed on the historical transaction data of POS bank cards to obtain potential bank card historical abnormal transaction data; through potential bank card abnormal transaction detection, the detection accuracy of abnormal transactions can be further improved. This can help the platform better identify transactions with potential risks and reduce false positives and false negatives. Potential bank card abnormal transaction detection can discover some hidden abnormal transaction patterns, which may not be easy to determine through conventional statistical analysis methods. Therefore, it can help identify more types of abnormal transaction behaviors. Merge the potential bank card historical abnormal transaction data, high-frequency bank card historical abnormal transaction data, and low-frequency bank card historical abnormal transaction data in time series to obtain the POS bank card historical abnormal transaction data. Merging different types of abnormal transaction data can provide a more comprehensive abnormal transaction data set, including high-frequency, low-frequency, and potential abnormal transactions. Such a data set facilitates deeper analysis and model training, thereby improving the accuracy and generalization of abnormal transaction detection. By merging different types of abnormal transaction data in time series, the evolution and trend of transaction behavior can be observed. This helps to discover the periodicity, seasonality, or other time-related laws of abnormal transaction behavior, thereby increasing the understanding of abnormal behavior.

[0147] Optionally, step S142 is specifically:

[0148] The bank card potential anomaly score formula is used to calculate the historical transaction data of the POS bank card to obtain the bank card potential anomaly score. The bank card potential anomaly score formula is as follows:

[0149]

[0150] Where S y is the potential anomaly score of the bank card, l is the number of historical bank card transaction data, k is the sequence number of the bank card transaction, t is the time, and a k is the amount of the kth bank card transaction, b k is the distance between the location of the kth bank card transaction and the common location, c k is the coefficient of difference between the time of the kth bank card transaction and the usual time, d k is the difference coefficient between the frequency of the kth bank card transaction and the common frequency, e kis the difference coefficient between the type of the kth transaction and the common type, f k The merchant credit score for the kth transaction;

[0151] The present invention constructs a bank card potential anomaly score formula for calculating the historical transaction data of POS bank cards. The formula fully considers the factors that affect the bank card potential anomaly score S y The number of historical bank card transaction data l, the sequence number of the bank card transaction k, the time t, and the amount of the kth bank card transaction a k , the distance b between the location of the kth bank card transaction and the common location k , the difference coefficient c between the time of the kth bank card transaction and the usual time k , the difference coefficient d between the frequency of the kth bank card transaction and the common frequency k , the difference coefficient e between the type of the kth transaction and the common type k , the merchant reputation score f of the kth transaction k , forming a functional relationship:

[0152]

[0153] The formula can help banks or financial institutions detect potential abnormal transactions. By considering multiple factors such as transaction amount, location, time, frequency, type and merchant reputation, the formula can comprehensively evaluate the degree of abnormality of each transaction and derive a score for potential abnormal transactions. This helps banks identify possible fraud, theft or other illegal transactions in advance, thereby protecting customers' funds. The formula can be used as a tool to help banks strengthen security and preventive measures. By comprehensively analyzing and evaluating transaction data, banks can identify and monitor potential abnormal transaction patterns and behaviors. This helps banks improve anti-fraud systems and risk management strategies, increase transaction security, and reduce the risk of financial crimes.

[0154] Perform statistical analysis on bank card potential anomaly score data to obtain bank card potential anomaly thresholds;

[0155] In this embodiment, the bank card potential abnormality score data is statistically analyzed by frequency analysis method to obtain the bank card potential abnormality threshold.

[0156] The bank card potential abnormality score data is classified and calculated according to the bank card potential abnormality threshold, so as to obtain the potential bank card historical abnormal transaction data.

[0157] In this embodiment, the bank card potential abnormality threshold is used to classify the bank card potential abnormality score data. Data with a bank card potential abnormality score exceeding the set threshold will be regarded as abnormal data, while data with a score below the threshold will be regarded as normal data, thereby obtaining potential bank card historical abnormal transaction data.

[0158] The present invention calculates the historical transaction data of a POS bank card by a bank card potential abnormality score formula, and can use the potential abnormality score formula to evaluate and calculate the historical transaction data of the bank card to obtain the potential abnormality score of each transaction. In this way, the bank can quantitatively evaluate each transaction, identify the potential abnormal transactions that may exist, and mark their abnormality. This helps to improve the accuracy and efficiency of automated abnormal transaction detection. The bank card potential abnormality score data is statistically analyzed to obtain the bank card potential abnormality threshold; the bank can obtain statistical indicators such as the distribution, mean, and standard deviation of the abnormal score. These statistical indicators can be used as a basis for determining the bank card potential abnormality threshold. By setting a suitable abnormality threshold, the bank can customize the boundary of abnormal transactions according to specific circumstances to better meet customer needs and reduce the false alarm rate and the missed alarm rate. The bank card potential abnormality score data is classified and calculated according to the bank card potential abnormality threshold, so as to obtain the potential bank card historical abnormal transaction data. By comparing the bank card potential abnormality score data with the bank card potential abnormality threshold, the transaction data can be divided into two categories: potential abnormalities and normal transactions. This helps banks identify and focus on potential abnormal transaction data and provide more accurate and efficient abnormal transaction management. By obtaining potential historical abnormal bank card transaction data, banks can conduct further detailed analysis and investigation to confirm whether they are actual abnormal transactions and take appropriate measures to protect customer assets and reduce financial risks.

[0159] Optionally, step S3 specifically includes:

[0160] Step S31: Obtain transaction request data;

[0161] In this embodiment, the transaction request data is obtained through the input device of the POS machine.

[0162] Step S32: extracting the customer transaction voucher from the transaction request data, thereby obtaining the customer transaction voucher information;

[0163] In this embodiment, the customer transaction voucher is extracted from the transaction request data by parsing the transaction message method, thereby obtaining the customer transaction voucher information.

[0164] Step S33: extracting basic information of the customer from the customer transaction voucher information, thereby obtaining basic information of the customer;

[0165] In this embodiment, a text matching method is used to extract the basic information of the customer from the customer transaction voucher information, thereby obtaining the basic information of the customer.

[0166] Step S34: extracting customer blacklist from the historical abnormal transaction data of the POS machine by frequency analysis, thereby obtaining customer blacklist data;

[0167] In this embodiment, the frequency analysis method is used to analyze the historical abnormal transaction data of the POS machine. Frequency analysis can count the number of transactions, transaction amount and other indicators of each customer, set the blacklist threshold according to the number of abnormal transactions of each customer, and define the scope of the blacklist according to the blacklist threshold. For example, abnormal transactions exceeding a certain number of times are considered to be blacklist customers.

[0168] Step S35: classify and calculate the customer's basic information according to the customer blacklist data. If the customer's basic information is classified as a blacklist customer, it is sent to the POS transaction system to stop the current transaction task; if the customer's basic information is classified as a normal customer, data for secondary detection is generated.

[0169] In this embodiment, the basic information of the customer is matched and compared with the customer blacklist data to determine the classification of the customer. If the customer is classified as a blacklist customer, a notification needs to be sent to the POS machine transaction system to request to stop the current transaction task. If the customer is classified as a normal customer, data for secondary detection is generated to facilitate subsequent abnormal transaction detection or risk assessment.

[0170] The present invention obtains transaction request data through the input device of the POS machine; obtaining accurate and complete transaction request data is the basis for subsequent processing and analysis, which is helpful to understand the situation and characteristics of the transaction. By extracting the customer transaction voucher information in the transaction request data, key information related to the customer, such as bank card number, transaction amount, transaction time, etc., can be obtained. This helps to associate the transaction with a specific customer and perform subsequent customer information analysis and processing. The customer basic information is extracted from the customer transaction voucher information to obtain the customer basic information; the customer basic information, such as customer name, ID card number, mobile phone number, etc., is extracted from the customer transaction voucher information. Obtaining the customer's basic information is helpful for further customer risk assessment and determination. By applying the frequency analysis method, the historical abnormal transaction data of the POS machine is analyzed, and specific customers with abnormally high frequency of occurrence can be identified, and these customers may have abnormal behavior or risks. Extracting customer blacklist data is helpful to establish a customer risk library for subsequent risk control and management. According to the customer blacklist data, the customer basic information is classified and calculated, and the customer can be divided into two categories: blacklist customers and normal customers. For blacklist customers, measures can be taken immediately to stop the current transaction task to prevent further risks. For normal customers, data for secondary testing is generated to provide support for subsequent refined risk assessment and decision-making.

[0171] Optionally, step S5 specifically includes:

[0172] The abnormal transaction prediction data is calculated by the compliance calculation formula to obtain the predicted transaction compliance data, wherein the compliance calculation formula is specifically as follows:

[0173]

[0174] C is the compliance index coefficient, T is the observation period, v is the total number of abnormal transactions, w is the abnormal transaction sequence number, x w is the ratio of the amount of the wth abnormal transaction to the average transaction amount;

[0175] The present invention constructs a compliance calculation formula for calculating abnormal transaction prediction data. The formula fully considers the observation time period T, the total number of abnormal transactions v, the abnormal transaction sequence w, the ratio of the amount of the wth abnormal transaction to the average transaction amount x, which affects the compliance index coefficient C. w , forming a functional relationship:

[0176]

[0177] in Represents the ratio of the amount of each abnormal transaction to the average transaction amount x wMultiply them one by one. In this way, we can get an overall growth rate, because each ratio greater than 1 means that the abnormal transaction amount is higher than the average level, and each ratio less than 1 means that the abnormal transaction amount is lower than the average level. This part of the calculation takes the vth root of the cumulative multiplication result, where v is the total number of abnormal transactions. This gives the average growth rate of each abnormal transaction.

[0178] This part indicates that when the total number of abnormal transactions tends to infinity, the average growth rate of each abnormal transaction is calculated to the limit. This can take into account the impact of potential more abnormal transactions on compliance, and then perform logarithmic operations on this part to obtain compliance indicators. Logarithmization can transform exponentially growing data into a form that is easier to compare and analyze. Finally, a derivative is performed, which means that the compliance indicator is differentiated to obtain the rate of change of the compliance indicator over time. This can be used to monitor compliance trends and promptly detect the increase or decrease of abnormal transactions. The formula takes into account factors such as the amount ratio, quantity and time changes of abnormal transactions, and obtains a comprehensive compliance indicator by performing mathematical operations on these factors. This makes it more accurate and comprehensive to evaluate and compare compliance in different time periods or different organizations.

[0179] The secondary inspection data is marked for security based on the predicted transaction compliance data to obtain the security inspection results.

[0180] In this embodiment, the data to be re-tested is marked for security based on the predicted transaction compliance data, and the data to be re-tested is divided into two categories: security and risk, to form a security test result.

[0181] The present invention uses a compliance calculation formula to calculate abnormal transaction prediction data, aiming to evaluate the compliance degree of transactions. Through calculation, the predicted compliance data of each transaction can be obtained, and the data can be used to determine whether the transaction complies with laws and regulations. This helps to identify potential illegal transactions, reduce compliance risks, and provide a basis for subsequent security marking and screening. According to the compliance data of the predicted transaction, the secondary detection data is marked for security. Through marking, transactions can be divided into two categories: safe and unsafe. The security detection results can be used to identify potential risks, prevent fraud and fraudulent behavior, and protect the security of customers and POS machine transaction systems. In addition, the security detection results can also be used for subsequent risk decision-making and further risk management. The amount ratio in the formula can reflect the degree of deviation of the abnormal transaction amount from the average level. Through mathematical operations such as cumulative multiplication and square root, the formula can comprehensively consider the deviation of multiple abnormal transactions. This enables the formula to help detect and identify abnormal transactions, especially when the number of abnormal transactions is large or the degree of abnormality is inconsistent. The derivative operation in the formula can calculate the rate of change of the compliance index over time. This makes it possible to perform compliance trend analysis and identify the growth or decline trend of the compliance level. This is important for monitoring compliance status, detecting growth or decrease in abnormal transactions, and taking timely measures.

[0182] Optionally, the risk assessment in step S6 is specifically:

[0183] Extract the credit rating of the customer's basic information to obtain the customer's credit rating;

[0184] In this embodiment, data extraction is performed on the basic information of the customer to obtain the customer's credit rating.

[0185] Extracting transaction time from transaction request data to obtain transaction time information;

[0186] In this embodiment, the timestamp of the transaction request data is extracted, and the timestamp is parsed or converted to obtain specific transaction time information, such as year, month, date, hour, minute, etc., so as to obtain the transaction time information.

[0187] The transaction time information, security test results and customer credit rating are calculated through the risk calculation formula to obtain risk assessment data. The risk calculation formula is as follows:

[0188]

[0189] Where R is the risk coefficient, t is the time, C is the compliance index coefficient, S is the customer credit rating, j is the limit integer, T r is the transaction time, and w is the transaction time weight coefficient.

[0190] The present invention constructs a risk calculation formula for calculating transaction time information, security test results and customer credit rating. The formula fully considers the compliance index coefficient C that affects the risk coefficient R, the customer credit rating S, the transaction time T r , the transaction time weight coefficient w forms a functional relationship:

[0191]

[0192] in This part represents the derivative of the logarithm of the ratio of security and customer credit rating with respect to time. It can reflect the change of security and customer credit rating over time. The positive or negative of the derivative is related to the trend of change, which can express the rate of change of risk. This part represents the j-root of the transaction time, where j approaches infinity. It is used to represent the impact of the increase in transaction time on risk, and is weighted by multiplying it by the weight coefficient w. When the transaction time is longer, the j-root value under the square root approaches 1, indicating that the impact of risk is smaller; on the contrary, when the transaction time is shorter, the j-root value under the square root approaches 0, indicating that the impact of risk is greater. This formula comprehensively considers the security test results, customer credit rating and transaction time information, and incorporates them into the risk assessment model. This can analyze risks more comprehensively and avoid relying solely on a single indicator or factor for assessment. Therefore, the formula helps to provide a more comprehensive and integrated risk assessment result. The formula The part represents the time rate of change of security and credit rating. Such consideration can capture the impact of time on risk and model the changes of security and credit rating over time. This is beneficial for risk assessment scenarios with time sensitivity, such as financial transactions and network security. The formula uses the weight coefficient w to weight the transaction time. This allows the relative importance of transaction time to risk to be adjusted according to specific needs. Different application scenarios may have different levels of concern about the impact of transaction time. By adjusting the weight coefficient w, its contribution to the final risk assessment can be flexibly controlled.

[0193] The present invention extracts the credit rating of the basic information of the customer, thereby obtaining the credit rating of the customer; the extraction of the credit rating of the customer can provide a basis for risk management. The transaction time is extracted from the transaction request data, thereby obtaining the transaction time information; the transaction time information can be used to detect abnormal transactions. If the time of a transaction is significantly deviated from the customer's historical transaction time or regular transaction time, it may indicate the existence of risks or fraud. Timely detection and processing of these abnormal transactions helps to prevent risks and protect the safety of customer funds. The transaction time information, security test results and customer credit rating are calculated through a risk calculation formula to obtain risk assessment data. By using the risk calculation formula, the risk assessment process can be automated to improve the efficiency and accuracy of decision-making. According to the calculated risk assessment data, the system can intelligently determine whether to process the transaction or take other measures, reduce the workload of manual intervention, and reduce misjudgments caused by human factors.

[0194] Optionally, this specification also provides an electronic payment transaction system, including:

[0195] The abnormal transaction extraction module is used to obtain the historical transaction data of the POS machine, extract the abnormal transaction data from the historical transaction data of the POS machine, and thus obtain the historical abnormal transaction data of the POS machine;

[0196] A model building module is used to standardize the historical abnormal transaction data of POS machines, thereby obtaining standardized historical abnormal transaction data, and building an abnormal transaction detection model for the standardized historical abnormal transaction data based on an integrated learning method;

[0197] A first security detection module, used to obtain transaction request data, perform a first security detection on the transaction request data, and thereby obtain data to be subjected to a second detection;

[0198] The abnormal transaction prediction module is used to perform abnormal transaction prediction on the secondary detection data through the abnormal transaction detection model, thereby obtaining abnormal transaction prediction data;

[0199] A second security detection module, used to perform a second security detection on the secondary detection data through abnormal transaction prediction data, so as to obtain a security detection result;

[0200] The risk assessment module is used to perform risk assessment on transaction request data according to the security detection results, thereby obtaining risk assessment data and sending it to the POS transaction system to perform offline classification transaction tasks.

[0201] The electronic payment transaction system of the present invention can implement any electronic payment transaction method of the present invention, and is used to combine the operation and signal transmission medium between various modules to complete the electronic payment transaction method. The internal structure of the system cooperates with each other to achieve security protection for offline POS transactions.

[0202] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0203] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. An electronic payment transaction method, characterized in that: The following steps are involved: Step S1: Acquire historical transaction data of the POS machine, extract abnormal transaction data from the historical transaction data of the POS machine, and thus obtain historical abnormal transaction data of the POS machine; Step S2: standardize the historical abnormal transaction data of the POS machine to obtain standardized historical abnormal transaction data, and construct an abnormal transaction detection model for the standardized historical abnormal transaction data based on an integrated learning method; Step S3: Acquire transaction request data, and perform a first security check on the transaction request data, thereby obtaining data to be subjected to a second check; Step S4: using the abnormal transaction detection model to perform abnormal transaction prediction on the secondary detection data, thereby obtaining abnormal transaction prediction data; Step S5: performing a second security test on the data to be secondary tested using the abnormal transaction prediction data, thereby obtaining a security test result; Step S6: Perform risk assessment on the transaction request data according to the security detection result, thereby obtaining risk assessment data, and sending it to the POS transaction system to execute the offline classification transaction task.

2. The method according to claim 1, characterized in that Step S1 is specifically as follows: Step S11: Obtaining historical transaction data of POS machine; Step S12: extracting platform transaction data and bank card transaction data from the POS machine historical transaction data, thereby obtaining the POS machine platform historical transaction data and the POS machine bank card historical transaction data; Step S13: extracting abnormal transaction data of the POS platform from the historical transaction data of the POS platform, thereby obtaining the abnormal transaction data of the POS platform; Step S14: extracting abnormal bank card transaction data from the POS machine bank card historical transaction data, thereby obtaining abnormal bank card transaction data from the POS machine; Step S15: Time-sequentially merge the historical abnormal transaction data of the POS platform and the historical abnormal transaction data of the POS bank card to obtain the historical abnormal transaction data of the POS.

3. The method according to claim 2, characterized in that Step S13 is specifically as follows: Step S131: Statistically analyzing the historical transaction data of the POS platform, thereby obtaining historical abnormal transaction data of the high-frequency platform and historical abnormal transaction data of the low-frequency platform; Step S132: Performing potential platform abnormal transaction detection on the historical transaction data of the POS platform, thereby obtaining potential platform historical abnormal transaction data; Step S133: Merge the historical abnormal transaction data of the potential platform, the historical abnormal transaction data of the high-frequency platform and the historical abnormal transaction data of the low-frequency platform in time series, so as to obtain the historical abnormal transaction data of the POS platform.

4. The method according to claim 3, characterized in that Step S132 is specifically as follows: The historical transaction data of the POS platform is calculated using the platform potential anomaly score formula to obtain the platform potential anomaly score data. The platform potential anomaly score formula is as follows: Where S p is the potential anomaly score of the platform, n is the number of historical platform transaction data, i is the platform transaction sequence number, t is the time, and a i is the amount of the ith platform transaction, b i is the average amount of the ith platform transaction, c is the abnormality sensitivity coefficient, d i is the difference between the time of the ith platform transaction and the average time, e i is the refund rate of the ith platform transaction, f i is the coupon usage rate of the ith platform transaction, g i is the payment method coefficient of the i-th platform transaction; Conduct statistical analysis on the platform's potential anomaly score data to obtain the platform's potential anomaly threshold; The platform's potential anomaly score data is classified and calculated according to the platform's potential anomaly threshold, thereby obtaining the potential platform's historical abnormal transaction data.

5. The method according to claim 2, characterized in that: Step S14 is specifically as follows: Step S141: Statistically analyzing the historical transaction data of POS bank cards, thereby obtaining the historical abnormal transaction data of high-frequency bank cards and the historical abnormal transaction data of low-frequency bank cards; Step S142: Performing potential bank card abnormal transaction detection on the POS machine bank card historical transaction data, thereby obtaining potential bank card historical abnormal transaction data; Step S143: Time-sequentially merge the potential bank card historical abnormal transaction data, the high-frequency bank card historical abnormal transaction data and the low-frequency bank card historical abnormal transaction data to obtain the POS bank card historical abnormal transaction data.

6. The method according to claim 5, characterized in that Step S142 is specifically as follows: The bank card potential anomaly score formula is used to calculate the historical transaction data of the POS bank card to obtain the bank card potential anomaly score. The bank card potential anomaly score formula is as follows: Where S y is the potential anomaly score of the bank card, l is the number of historical bank card transaction data, k is the sequence number of the bank card transaction, t is the time, and a k is the amount of the kth bank card transaction, b k is the distance between the location of the kth bank card transaction and the common location, c k is the coefficient of difference between the time of the kth bank card transaction and the usual time, d k is the difference coefficient between the frequency of the kth bank card transaction and the common frequency, e k is the difference coefficient between the type of the kth transaction and the common type, f k The merchant credit score for the kth transaction; Perform statistical analysis on bank card potential anomaly score data to obtain bank card potential anomaly thresholds; The bank card potential abnormality score data is classified and calculated according to the bank card potential abnormality threshold, so as to obtain the potential bank card historical abnormal transaction data.

7. The method according to claim 1, characterized in that Step S3 is specifically as follows: Step S31: Obtain transaction request data; Step S32: extracting the customer transaction voucher from the transaction request data, thereby obtaining the customer transaction voucher information; Step S33: extracting basic information of the customer from the customer transaction voucher information, thereby obtaining basic information of the customer; Step S34: extracting customer blacklist from the historical abnormal transaction data of the POS machine by frequency analysis, thereby obtaining customer blacklist data; Step S35: classify and calculate the customer's basic information according to the customer blacklist data. If the customer's basic information is classified as a blacklist customer, it is sent to the POS transaction system to stop the current transaction task; if the customer's basic information is classified as a normal customer, data for secondary detection is generated.

8. The method according to claim 7, characterized in that Step S5 is specifically as follows: The abnormal transaction prediction data is calculated by the compliance calculation formula to obtain the predicted transaction compliance data, wherein the compliance calculation formula is specifically as follows: C is the compliance index coefficient, T is the observation period, v is the total number of abnormal transactions, w is the abnormal transaction number, x w is the ratio of the amount of the wth abnormal transaction to the average transaction amount; The secondary inspection data is marked for security based on the predicted transaction compliance data to obtain the security inspection results.

9. The method according to claim 8, characterized in that The risk assessment in step S6 is specifically as follows: Extract the credit rating of the customer's basic information to obtain the customer's credit rating; Extracting transaction time from transaction request data to obtain transaction time information; The transaction time information, security test results and customer credit rating are calculated through the risk calculation formula to obtain risk assessment data. The risk calculation formula is as follows: Where R is the risk factor, t is the time, C is the compliance index coefficient, S is the customer credit rating, j is the limit integer, T r is the transaction time, and w is the transaction time weight coefficient.

10. An electronic payment transaction system, characterized in that: include: The abnormal transaction extraction module is used to obtain the historical transaction data of the POS machine, extract the abnormal transaction data from the historical transaction data of the POS machine, and thus obtain the historical abnormal transaction data of the POS machine; A model building module is used to standardize the historical abnormal transaction data of POS machines, thereby obtaining standardized historical abnormal transaction data, and building an abnormal transaction detection model for the standardized historical abnormal transaction data based on an integrated learning method; A first security detection module, used to obtain transaction request data, perform a first security detection on the transaction request data, and thereby obtain data to be subjected to a second detection; The abnormal transaction prediction module is used to perform abnormal transaction prediction on the secondary detection data through the abnormal transaction detection model, thereby obtaining abnormal transaction prediction data; A second security detection module, used to perform a second security detection on the secondary detection data through abnormal transaction prediction data, so as to obtain a security detection result; The risk assessment module is used to perform risk assessment on transaction request data according to the security detection results, thereby obtaining risk assessment data and sending it to the POS transaction system to perform offline classification transaction tasks.

Images