A method and system for managing the security of transaction data in an enterprise e-commerce platform
By evaluating multi-dimensional information and load status of enterprise e-commerce transactions and dynamically adjusting security verification strategies, the problem of erroneous release of high-risk transactions in high-concurrency scenarios has been solved, ensuring fund security and system stability.
Patent Information
- Application Number
- CN202611571038.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-12
- Publication Date
- 2026-11-17
AI Technical Summary
In high-concurrency scenarios of enterprise e-commerce transactions, existing technologies cannot effectively distinguish between high-risk and low-risk transactions, leading to the erroneous release of high-risk transactions when the system is overloaded, resulting in fund settlement errors and economic losses.
By acquiring multi-dimensional information about payment confirmation requests, we assess their risk level and dynamically adjust security verification strategies based on load status. This ensures that high-risk transactions undergo strong verification even under high load, while low-risk transactions are downgraded to avoid erroneous approval.
Under high load conditions, it effectively prevents the accidental release of high-risk transactions, ensures the safety of transaction funds, and improves the security and stability of the system in concurrent scenarios.
Smart Images

Figure CN123118094A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of enterprise security technology, and more specifically, to a method and system for managing the security of transaction data in an enterprise e-commerce platform. Background Technology
[0002] In the security management of transaction data in enterprise e-commerce platforms, tamper-proof verification and integrity verification are typically performed on payment confirmation requests. During peak payment confirmation periods caused by high-concurrency activities such as flash sales, the computing resources of the security verification processing unit can quickly become saturated, with a large number of pending requests accumulating in the buffer queue. When the waiting time exceeds the waiting threshold and triggers the overload protection mechanism, a strategy of global verification strength degradation is adopted. For example, the signature verification length may be shortened indiscriminately, secondary comparison may be skipped, or the additional identity verification process for high-risk transactions may be temporarily disabled. Requests for large-amount payments with high-risk flags mixed in with normal small-amount payments may be allowed through the same low-strength verification strategy as ordinary requests. This leads to the neglect of risk differences between different transactions when processing high-concurrency requests, resulting in the erroneous release of abnormal payment requests with high-risk flags, leading to fund settlement errors and direct economic losses. Summary of the Invention
[0003] The purpose of this application is to provide a method and system for managing the security of transaction data in an enterprise e-commerce platform. The system aims to achieve differentiated security verification by matching load status with risk level, and effectively prevent high-risk transactions from being mistakenly released when the system is overloaded.
[0004] Firstly, this application provides a method for secure management of transaction data in an enterprise e-commerce platform, including: Obtain transaction attribute information, user behavior information, device identification information, and security context information for each payment confirmation request; Based on the transaction attribute information, user behavior information, device identification information, and security context information, assess the risk level of each payment confirmation request; Obtain the load monitoring indicators of the verification unit used for security verification, determine the current load status based on the load monitoring indicators, and call the corresponding verification strategy to perform security verification on the payment confirmation request based on the current load status and the risk level of each payment confirmation request, obtain the security verification result, and generate a verification completion status certificate based on the security verification result. Based on the verification completion status certificate, the corresponding payment confirmation request is released.
[0005] Secondly, this application provides a secure management system for enterprise e-commerce transaction data, including: The information acquisition module is used to acquire transaction attribute information, user behavior information, device identification information, and security context information for each payment confirmation request; The risk assessment module is used to assess the risk level of each payment confirmation request based on the transaction attribute information, user behavior information, device identification information, and security context information. The security verification module is used to obtain the load monitoring indicators of the verification unit used for security verification, determine the current load status based on the load monitoring indicators, call the corresponding verification strategy to perform security verification on the payment confirmation request based on the current load status and the risk level of each payment confirmation request, obtain the security verification result, and generate a verification completion status certificate based on the security verification result. The release control module is used to release the corresponding payment confirmation request based on the verification completion status certificate.
[0006] As can be seen from the above, the enterprise e-commerce transaction data security management method and system provided in this application, by acquiring and comprehensively evaluating multi-dimensional information of payment confirmation requests, and combining load monitoring indicators and transaction risk levels when the system load fluctuates, can achieve differentiated adjustment of strategies, ensuring that high-risk transactions maintain high-intensity verification under high-load scenarios, avoiding the erroneous release of high-risk theft requests due to overload degradation, and effectively protecting the security of transaction funds. Attached Figure Description
[0007] Figure 1 This is a flowchart of a method for secure management of transaction data in an enterprise e-commerce platform, provided in the first specific embodiment of this application; Figure 2 This is a flowchart of a method for managing the security of transaction data in an enterprise e-commerce platform, provided in the second specific embodiment of this application; Figure 3 This is a flowchart of a method for managing the security of transaction data in an enterprise e-commerce platform, provided in the third specific embodiment of this application. Detailed Implementation
[0008] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.
[0009] It should be noted that similar reference numerals and letters in the following figures denote similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0010] Firstly, such as Figure 1 As shown, Figure 1 This is a flowchart of a method for managing the security of transaction data in an enterprise e-commerce platform, provided in the first specific embodiment of this application. This method may include, but is not limited to, steps S110 to S140.
[0011] Step S110: Obtain transaction attribute information, user behavior information, device identification information, and security context information for each payment confirmation request; Step S120: Based on transaction attribute information, user behavior information, device identification information, and security context information, assess the risk level of each payment confirmation request; Step S130: Obtain the load monitoring index of the verification unit used for security verification, determine the current load status based on the load monitoring index, call the corresponding verification strategy to perform security verification on the payment confirmation request based on the current load status and the risk level of each payment confirmation request, obtain the security verification result, and generate a verification completion status certificate based on the security verification result. Step S140: Based on the verification completion status certificate, release control is applied to the corresponding payment confirmation request.
[0012] Understandably, after receiving a payment confirmation request, this application first collects multi-dimensional contextual information about the request, and then classifies the potential risks of the request based on this information. At the same time, it monitors the operating pressure of the verification unit used to perform the verification task in real time to obtain the current load status. Subsequently, it cross-matches the current load status with the risk level of the request and dynamically assigns different strength verification strategies to different requests. After the verification is completed, a certificate is generated. Finally, the release control link decides whether to allow the transaction to enter the final fund clearing stage based on the certificate.
[0013] For example, when a payment confirmation request forwarded by the front-end gateway is received, the required information is extracted by parsing the request message header and body data. Among them, transaction attribute information may include order amount and category of purchased goods; user behavior information may include the duration of the user's stay on the page before submitting payment; device identification information may include the device's operating system version; and security context information may include whether the current network environment is a public wireless LAN.
[0014] Specifically, during the process of a user entering the checkout page, confirming an order, entering a payment password, and clicking the payment button, the front-end application or web page continuously generates context data related to the payment behavior. Some of this data is collected through front-end page scripts, some through terminal security components, and some is supplemented by the back-end order and account systems. When the front-end gateway receives a payment confirmation request, it first decomposes the message according to predefined field mapping relationships, and then writes the decomposed fields into a unified request context object. Taking transaction attribute information as an example, fields such as order amount, product category, coupon usage, and shipping address region identifier can be directly read from the order details returned by the order service. Taking user behavior information as an example, the front-end page can record the start time when the user enters the payment page and the end time when the user clicks to confirm payment, thus obtaining the page dwell time. It can also further record behavioral trajectory summaries such as the number of password entries, button click intervals, and page switching times. Taking device identification information as an example, terminal security components can collect information such as operating system version, device model, application version, device fingerprint summary, and whether debug mode is enabled. Taking security context information as an example, the network access layer and front-end security components can jointly provide information such as network type, Internet protocol address, coarse-grained geographical location label, whether a proxy network label is hit, and whether the user is in a public Wi-Fi environment.
[0015] For example, the risk level of each payment confirmation request can be assessed by pre-configuring a static rule engine. The static rule engine sets fixed threshold conditions for multiple dimensions. After obtaining information, it is compared with the conditions in the static rule engine one by one. For example, if the order amount in the transaction attribute information is greater than RMB 10,000 and the security context information shows that it is a public wireless LAN, a high-risk alarm is triggered, and the risk level of the payment confirmation request is directly assessed as the highest risk level. If none of the information triggers the preset fixed threshold conditions, it is assessed as the lowest risk level.
[0016] For example, a verification unit refers to an independent computing service node in the backend architecture of an enterprise e-commerce platform, used to perform security checks such as digital signature verification and data tamper-proof comparison on received request data. A verification unit can be a single service process or a service cluster composed of multiple service instances, located after the payment gateway and before the clearing service. It can also be triggered as an independent microservice through an internal call chain. To obtain load monitoring metrics for the verification unit used for security verification, a monitoring agent program set up on the verification unit's host machine can periodically collect single hardware resource consumption data as load monitoring metrics. Alternatively, only the real-time CPU utilization rate of the verification unit can be obtained. When the real-time CPU utilization rate exceeds 85%, the current load state is determined to be overloaded; when the real-time CPU utilization rate is less than or equal to 85%, it is determined to be normal. The monitoring agent program can periodically read CPU usage through the performance statistics interface provided by the operating system and report the collected results to the load judgment module. After receiving multiple consecutive sample values, the load judgment module uses the smoothed result within a short time window as the current judgment basis to avoid misjudgment caused by a single instantaneous spike.
[0017] For example, based on the current load status and the risk level of each payment confirmation request, when the current load status is normal, a complex verification strategy including full checks can be applied to all requests. When the current load status is overloaded, requests can be triaged according to their risk level. For requests assessed as the highest risk level, a complex verification strategy is applied to ensure security; for requests assessed as the medium risk level, a medium-strength verification strategy is applied; and for requests assessed as the lowest risk level, a simplified verification strategy including only basic signature verification is applied. After verification, a boolean value indicating whether the verification passed or failed can be bound to the unique identifier of the request, encapsulating and generating a verification completion status credential.
[0018] Complex verification strategies may include signature verification, key field consistency comparison, order and account binding relationship check, request timeliness check, replay request identification, device environment anomaly check, and pre-risk control label review. Medium-strength verification strategies may retain signature verification, key field consistency comparison, request timeliness check, and some environment checks. Simplified verification strategies shall at least retain signature verification and key field integrity check.
[0019] During the execution of different strategies, each check item can be executed sequentially or in parallel and the results can be summarized. If any critical check item fails, the security verification is deemed to have failed and the failure reason code is recorded. If all mandatory check items pass, the security verification is deemed to have passed. When generating the verification completion status certificate, in addition to whether it has passed and the unique identifier of the request, the verification strategy identifier, risk level, load status, failure reason code, and certificate generation time can also be written. In this way, subsequent release control can make decisions based on whether it has passed or failed, and can also identify the system load conditions and verification strength corresponding to the request, thereby providing a complete basis for audit trails and anomaly reviews.
[0020] For example, the verification completion status credential refers to a structured data tag generated after the payment confirmation request has gone through the security verification process. It is used to record the verification result of the request and the verification process information it has gone through, as the basis for whether to allow subsequent steps. The structured data tag can be stored in the form of key-value pairs, structured message fragments or internal tokens. The verification completion status credential may include a unique request identifier, a verification policy identifier, a verification start time, a verification end time, whether the verification passed, a reason code for failure, a risk level, a load status, and anti-tampering protection fields.
[0021] Based on the verification completion status credential, the backend gateway intercepts all requests preparing to enter the clearing database, parses the verification completion status credential, and if the credential indicates successful verification, allows the request to enter the clearing process. If the credential indicates failed verification, the request is directly blocked and a payment failure message is returned to the frontend. The gateway is located between the security verification process and the fund clearing process. All payment confirmation requests must pass through the gateway before entering the clearing service. The gateway first checks whether the request carries a verification completion status credential. If it does not, it is directly considered as having failed security verification and blocked. If it does, it first verifies the credential. The authenticity and completeness of the certificate are verified, such as the certificate signature, whether the certificate has expired, and whether the unique identifier of the request in the certificate is consistent with the current request. Only when the certificate is authentic and valid and bound to the current request will the verification result fields be parsed. If the verification result is successful and the additional information such as the policy identifier and risk level in the certificate meets the pre-liquidation requirements, the request will be forwarded to the liquidation database or liquidation service. If the verification result is unsuccessful, or the certificate verification fails, the certificate does not match the request, or the certificate has expired, the gateway will allow the request to proceed directly to terminate the subsequent flow of the request and return a unified failure response to the front end.
[0022] In addition, the gateway can require the credentials to include summary information that is bound together with the order identifier, payment transaction identifier, and timestamp, and verify it again when allowing the request to prevent attackers from copying the credentials that have been approved for other requests. This double-verifies the credentials themselves and the binding relationship between the credentials and the requests, thus preventing the credentials from being forged, reused, or swapped.
[0023] Understandably, this application constructs a two-dimensional dynamic scheduling logic by implementing a risk assessment mechanism before performing security verification and by monitoring the load status of the verification unit in real time. When faced with a surge in load due to a large number of concurrent requests, it prioritizes computing resources to high-risk requests based on the pre-assessed risk level to ensure that they undergo strict security verification. At the same time, it implements a strategy downgrade for low-risk requests to quickly release resources. This effectively alleviates the computing pressure on the verification unit while accurately intercepting high-risk large-amount payment fraud requests mixed in during overload, avoiding fund settlement errors caused by blind downgrades, and improving the transaction security of the enterprise e-commerce platform in concurrent scenarios.
[0024] like Figure 2 As shown, Figure 2 This is a flowchart of a method for managing the security of transaction data in an enterprise e-commerce platform provided in the second specific embodiment of this application. Regarding the above step S120, the method may include, but is not limited to, steps S210 to S240.
[0025] Step S210: Extract continuous numerical fields from transaction attribute information and user behavior information, calculate the deviation of continuous numerical fields, and obtain the numerical deviation score of each payment confirmation request. Step S220: Extract discrete security feature data from device identification information and security context information. Discrete security feature data includes device fingerprint, Internet Protocol address, and pre-positioned risk control tag. Step S230: Perform risk analysis based on device fingerprint, Internet Protocol address and pre-installed risk control tag to obtain the risk score of each payment confirmation request; Step S240: Determine the risk level of each payment confirmation request based on the numerical deviation score and risk score of each payment confirmation request.
[0026] For example, continuous numerical fields are extracted from transaction attribute information and user behavior information. Continuous numerical fields refer to data that can take values continuously within a certain range, such as transaction amount, page scrolling speed, and the time interval between two clicks. For example, the transaction amount of the current request and the payment operation time can be extracted as continuous numerical fields. Then, the mean and standard deviation of the same field over the past 30 days of the user are retrieved, and a standard score algorithm is used to calculate the degree of deviation of the current field value from the historical baseline. If the current transaction amount is much higher than the historical average and the payment operation time is short, the deviation calculation result outputs a higher numerical deviation score, thereby quantifying the degree of abnormality of the behavior.
[0027] For example, each time a user completes a payment, cancels a payment, or enters the payment page, a summary of the relevant behavior can be written to the historical behavior sample library. The sample library stores at least the user identifier, transaction scenario identifier, transaction amount, payment operation time, page dwell time, click interval summary, and sample time. The mean and standard deviation of similar fields can be extracted from historical samples similar to the current transaction scenario. For example, historical payment records under the same account, the same terminal type, and the same business scenario are preferred to avoid baseline distortion caused by excessive differences between different scenarios. When the current payment confirmation request is received, similar samples of the user within a preset historical time range are first read from the historical behavior sample library. Then, historical baselines for transaction amount and payment operation time are formed respectively. Subsequently, the deviation of the current value from the historical baseline is compared. If the current transaction amount is significantly higher than the user's usual payment amount range, it indicates that the transaction is abnormal in terms of fund size. If the current payment operation time is significantly shorter than the user's usual input, confirmation, and submission time, it indicates that the operation may be scripted, automated, or abnormally skilled. In this way, by summarizing the deviation results of multiple continuous numerical fields, a unified numerical deviation score is formed.
[0028] Furthermore, a minimum sample size requirement can be set. When historical samples are insufficient, user group baseline, device type baseline, or business scenario baseline can be used as an alternative reference to avoid baseline instability when historical samples are too few. Even when facing new users, low-frequency users, or insufficient historical data, usable deviation judgment results can still be obtained.
[0029] For example, discrete security feature data refers to state-type data whose values are independent of each other and have no continuous increasing relationship. Discrete security feature data includes the device fingerprint of the current request, Internet Protocol address, and pre-prepared risk control tags attached by front-end security components. Device fingerprints can be generated by combining multiple terminal attributes to reduce the instability of identification caused by missing or changed single fields. Internet Protocol addresses can also be preprocessed by address reputation database, proxy identification database, regional tag database, etc. before participating in risk analysis, so as to convert the original address into a more suitable tagged result for risk judgment.
[0030] After extracting discrete security feature data from device identification information and security context information, it can be input into a preset risk probability matrix. The device identification information and security context information requested can be discretized and encoded first, and then input into the preset risk probability matrix to search for single feature risks and combined feature risks. If a high-risk combination is hit, the risk score is increased. If only a single weak risk feature is hit, a lower risk increase is output. If multiple medium-risk features appear at the same time, they are superimposed to form a higher risk score. For example, when the Internet Protocol address is identified as a high-risk proxy network segment and the pre-control label shows that it is running as an emulator, the risk probability matrix will output a higher risk score based on the statistical distribution of historical fraud samples.
[0031] For example, a weighted fusion approach can be used to process the numerical deviation score and risk score of each payment confirmation request to determine the risk level of each payment confirmation request. For instance, the numerical deviation score can be assigned a weight of 40%, the risk score a weight of 60%, and a comprehensive risk value can be calculated. Then, the comprehensive risk value can be mapped to a preset level classification table to accurately determine the final risk level of the payment confirmation request.
[0032] Based on this, the system abandons the crude static rule judgment and instead adopts a targeted analysis model based on the inherent attributes of the data. This application calculates the deviation of continuous numerical fields, which can identify abnormal operations that deviate from users' daily habits. It performs risk analysis on discrete security feature data, which can effectively identify objective risk factors in the device environment and network links. The combined analysis makes the risk assessment process include the analysis of dynamic behavior and the review of the static environment, thereby improving the accuracy and anti-interference ability of risk level assessment in complex transaction scenarios.
[0033] In the enterprise e-commerce transaction data security management method provided in some specific embodiments of this application, regarding the above step S130, the method may include, but is not limited to, steps S310 to S320.
[0034] Step S310: Obtain the load monitoring indicators of the verification unit used for security verification. The load monitoring indicators include processor utilization, memory usage, network throughput, the total number of payment confirmation requests in the current buffer queue, the request arrival rate within the historical preset monitoring time, and the average waiting time of payment confirmation requests in the current buffer queue. Step S320: Input the load monitoring indicators into the preset comprehensive load index calculation formula to obtain the current comprehensive load index. Determine the current load status based on the position of the current comprehensive load index in the preset segment judgment interval.
[0035] For example, underlying processor utilization, memory usage, and network throughput can be obtained through the operating system interface. The total number of payment confirmation requests backlogged in the current buffer queue can be counted in real time through the application layer middleware interface. Simultaneously, a historical preset monitoring time is set to the past five seconds to calculate the request arrival rate within those five seconds (i.e., the number of new requests per second), and the average waiting time of all requests in the current buffer queue from entry into the queue to the current moment is tracked. Specifically, processor utilization and memory usage can be read from the operating system performance interface through the host monitoring agent. Network throughput can be provided through the network interface card statistics interface, container network monitoring interface, or service mesh side components. The total number of payment confirmation requests in the current buffer queue can be returned in real time through message queues, middleware buffers, or application internal task queues. The request arrival rate can be obtained by counting requests at the gateway entry point or verification unit entry point and counting the number of new requests within a continuous monitoring window. The average waiting time can be obtained by recording the enqueue time for each request entering the buffer queue, iterating through the requests in the current queue that have not yet been processed at the monitoring moment, calculating the waiting time from enqueue to the current moment, and then taking the average.
[0036] If the verification unit is set up in a cluster, local metrics of each instance can be collected first, and then cluster-level metrics can be generated through the aggregation module, such as average value, peak value or total value, to characterize the real load status of the entire verification service cluster.
[0037] The technical approach involves inputting load monitoring indicators into a preset comprehensive load index calculation formula to obtain the current comprehensive load index, and then determining the current load status based on the position of the current comprehensive load index within a preset segmented judgment interval. For example, the preset comprehensive load index calculation formula can adopt a normalized weighted summation algorithm. First, the load monitoring indicators are mapped to a standardized range of 0 to 100. Then, weights are assigned according to business sensitivity to calculate the current comprehensive load index. Subsequently, the preset segmented judgment range is consulted. For example, 0 to 40 is set as the first range, 41 to 80 as the second range, and 81 to 100 as the third range. If the calculated current comprehensive load index is 75, it falls into the second range, and the current load status is determined to be at the corresponding specific overload level.
[0038] In addition, a duration judgment or adjacent interval buffering mechanism can be introduced during state switching. That is, when the comprehensive load index falls stably into the target interval for multiple consecutive sampling periods, the load state is switched to avoid the load state frequently switching back and forth near the critical interval. This can reduce the interference of instantaneous jitter on the verification strategy scheduling and make the strategy switching smoother and more stable.
[0039] Based on this, this application integrates the comprehensive evaluation of underlying hardware resource consumption and upper-layer business processing efficiency, and business indicators such as request arrival rate and average waiting time, which enables the prediction of congestion trends in advance, thus overcoming the lag of single hardware indicators. By converting multi-dimensional indicators into a comprehensive load index and performing interval mapping, interference caused by instantaneous data jitter can be effectively filtered out, ensuring the comprehensiveness, real-time performance, and smoothness of load status assessment.
[0040] like Figure 3 As shown, Figure 3 This is a flowchart of a method for managing the security of transaction data in an enterprise e-commerce platform provided in the third specific embodiment of this application. The current load status includes normal load status, light overload status and heavy overload status. The risk level includes high risk level, medium risk level and low risk level. The verification strategy includes a first strategy and a second strategy. The verification level of the first strategy is higher than that of the second strategy. Regarding the above step S130, the method may include, but is not limited to, steps S410 to S420.
[0041] Step S410: When the current load status is normal load status, each payment confirmation request is loaded into the preset strong verification guarantee queue to call the first strategy to perform security verification on the payment confirmation request and obtain the security verification result. Step S420: When the current load state is either slightly overloaded or heavily overloaded, high-risk payment confirmation requests are loaded into a preset strong verification queue to invoke the highest security level verification strategy to perform security verification on the payment confirmation requests and obtain security verification results. Medium-risk payment confirmation requests are loaded into a preset flexible verification queue to invoke the first or second strategy to perform security verification on the payment confirmation requests according to the changes in the current load state and obtain security verification results. Low-risk payment confirmation requests are loaded into a preset lightweight verification queue to invoke the corresponding second strategy to perform security verification on the payment confirmation requests and obtain security verification results.
[0042] For example, the payment confirmation request can be routed to different queues through a gateway routing module, a request scheduling module, or a message middleware delivery module. After the payment confirmation request obtains a risk level label through the pre-control module, a unique request identifier, a risk level label, a generation time, and a source business identifier are appended to the request context. Subsequently, the routing module reads the current load status cache value and completes the queue delivery according to the pre-established correspondence between load status, risk level, target queue, and verification strategy. For example, when the cache records that the current state is slightly overloaded and the risk level label of the request is medium risk, the routing module sets the target queue field in the request header to a preset elastic verification queue and simultaneously writes a candidate strategy identifier. Subsequently, the consumption processor corresponding to the elastic queue determines whether to actually execute the first strategy or the second strategy based on the latest load status.
[0043] For example, under normal load conditions, high-risk, medium-risk, and low-risk levels all enter the preset strong verification protection queue and execute the first strategy. Under light overload conditions, high-risk levels enter the preset strong verification protection queue and execute the first strategy, medium-risk levels enter the preset elastic verification queue and execute the first strategy first, and low-risk levels enter the preset lightweight verification queue and execute the second strategy. Under heavy overload conditions, high-risk levels enter the preset strong verification protection queue and execute the first strategy, medium-risk levels enter the preset elastic verification queue and switch to executing the second strategy, and low-risk levels enter the preset lightweight verification queue and execute the second strategy.
[0044] Specifically, three independent buffer queues can be pre-allocated in memory: a pre-defined strong verification guarantee queue, a pre-defined elastic verification queue, and a pre-defined lightweight verification queue. Each queue is bound to different computing resource quotas and scheduling priorities. Under normal load conditions, all payment confirmation requests, regardless of risk level, are uniformly sent to the pre-defined strong verification guarantee queue, and the first strategy is enforced to ensure the highest level of security protection within resource limits. When a state of mild or severe overload is detected, high-risk requests are sent to the pre-defined strong verification guarantee queue, continuing to use the highest security level verification strategy. Low-risk requests are sent to the pre-defined lightweight verification queue, executing the second strategy to quickly release resources. Medium-risk requests are sent to the pre-defined elastic verification queue, which has dynamic adaptive capabilities; it may invoke the first strategy during mild overload and downgrade to the second strategy during severe overload.
[0045] For example, the preset strong validation guarantee queue, preset elastic validation queue, and preset lightweight validation queue can be implemented using three logical topics within the same message middleware, or as three independent blocking queues within the same service process. If a distributed configuration is adopted, consumption isolation can also be achieved between different consumer groups. The independence between the preset strong validation guarantee queue, preset elastic validation queue, and preset lightweight validation queue includes at least storage isolation, meaning each queue has independent queuing space to prevent a large influx of low-risk requests from crowding out the queuing positions of high-risk requests. It also includes consumption isolation, meaning each queue can be bound to an independent worker thread pool or an independent container instance to prevent the rapid consumption of lightweight tasks from impacting deep validation tasks. Furthermore, it includes scheduling isolation, meaning a higher wake-up priority and more stable resource reservation quota can be set for the preset strong validation guarantee queue. The resource quota of the thread pool for the preset strong validation guarantee queue remains locked during runtime and does not participate in global compression. The thread pool for the preset elastic validation queue can be appropriately scaled up or down according to the current load, while the thread pool for the preset lightweight validation queue prioritizes high throughput.
[0046] Based on this, this application isolates the originally mixed payment confirmation requests at the physical queue level through multi-level load states and multi-queue collaborative scheduling. When resources are scarce, high-risk requests are locked in a strong verification guarantee queue to ensure that they are not affected by the overall system degradation. Low-risk requests are diverted to a lightweight verification queue to complete basic verification at a lower computational cost. Medium-risk requests act as a buffer, and the verification intensity is flexibly adjusted according to the real-time deterioration or relief of the load. This avoids premature full degradation during mild congestion and also prevents incomplete degradation during severe congestion, achieving a smooth transition between security and performance. Furthermore, the isolation of physical queues prevents high and low-risk requests from blocking each other, ensuring that limited computing resources can be used for deep verification of high-risk requests in concurrent scenarios. This maximizes the overall throughput of the system while ensuring the security of core transactions.
[0047] In the enterprise e-commerce transaction data security management method provided in some specific embodiments of this application, before step S130 above, which calls the corresponding verification strategy to perform security verification on the payment confirmation request based on the current load status and the risk level of each payment confirmation request, the method may also include, but is not limited to, step S510.
[0048] Step S510: Set mandatory verification conditions for the preset strong verification guarantee queue so that when performing security verification in the preset strong verification guarantee queue, all verification items of the first strategy are forcibly completed and the adjustment or modification of the first strategy is restricted.
[0049] For example, the mandatory verification conditions may include a list of mandatory verification items for the first strategy, the execution order of each verification item, a no-skip flag, a no-downgrade flag, an absolute timeout cap handling rule, a failure handling action, and a strategy version lock flag. The list of mandatory verification items is used to clarify which verification items must be executed completely. The execution order is used to prevent confusion in the order of execution due to component reordering at runtime. The no-skip flag and the no-downgrade flag are used to constrain the execution engine not to bypass the item. The absolute timeout cap handling rule is used to avoid infinite waiting caused by individual dependencies not responding for a long time. The failure handling action is used to stipulate that when a mandatory item cannot obtain a valid result, the verification failure should be output directly. The strategy version lock flag is used to prevent temporary switching to a simplified version during overload.
[0050] For example, by enforcing verification conditions, an immutable strong consistency execution attribute is set for the pre-defined strong verification guarantee queue during the queue initialization phase. When a payment confirmation request enters the queue, the bound first policy obtains the highest level of execution exemption, unaffected by any global degradation instructions or the circuit breaker mechanism of the underlying components. The immutability can be achieved through configuration signature verification, version locking, read-only loading, and runtime change interception. For instance, during queue initialization, the queue management module reads the first policy contract file issued by the configuration center and first verifies the file's signature information, version number, and release timestamp. When the signature is valid and the version is in an allowed state, the policy is marked as the enforced version. After loading, the execution engine writes the list of verification items corresponding to that version into a read-only memory area; during runtime, ordinary management instructions can only read and cannot modify it.
[0051] Strong consistency execution means that for requests in the pre-defined strong verification guarantee queue, the execution result must be consistent with the set of items defined in the contract file. Any missing, omitted, or substituted execution is considered non-compliant. When the global scheduler issues a degradation notification due to scheduling pressure, the consumer end of the pre-defined strong verification guarantee queue will recognize a unique exemption flag, thus refusing to reduce the first strategy. When the underlying component returns a degradation response or a default result, the execution engine will not recognize it as a valid success result, but will continue to wait for a clear response according to the contract rules or determine failure according to the absolute timeout limit.
[0052] Based on this, this application establishes a security sandbox for the verification process of high-risk requests by setting mandatory verification conditions at the queue level. Within the security sandbox, the integrity and immutability of the first strategy are guaranteed, thus preventing implicit degradation of high-risk requests during the verification process.
[0053] In the enterprise e-commerce transaction data security management method provided in some specific embodiments of this application, the verification items of the first strategy include full-field hash comparison, full-length digital signature verification, exchange rate snapshot recalculation and intermediate bank compliance whitelist comparison, and the second strategy includes abnormal upgrade and downgrade compensation rules. Abnormal upgrade / downgrade compensation rules include: If a field format distortion or logical contradiction is detected during the security verification process, the security verification will be interrupted and the risk level of the corresponding payment confirmation request will be adjusted to a high-risk level.
[0054] For example, the original message fields required for full-field hash comparison can be obtained from the payment confirmation request body, and the baseline hash value can be obtained from the transaction digest record issued during the upstream payment initiation stage. During execution, the verification engine reassembles the message according to a fixed field order, recalculates the digest, and performs a consistency comparison to identify minor tampering such as field insertion, deletion, or reordering. The signature value, public key identifier, and signature coverage information required for full-length digital signature verification can be obtained from the request header and signature management service. During execution, the signature content covering the entire data packet is verified to prevent attackers from retaining the legitimacy of key fields while tampering with non-core fields. The exchange rate snapshot recalculation requires... The transaction time, original currency, target currency, and message amount can be obtained from the transaction context. Historical exchange rate snapshots can be retrieved from the exchange rate snapshot service or audit database. During execution, the amount conversion result is recalculated based on the snapshot corresponding to the actual time of the transaction to determine whether there is exchange rate manipulation in cross-border or multi-currency transactions. The clearing path, intermediary bank identifier, and payee information required for the intermediary bank compliance whitelist comparison can be obtained from the payment link message. The whitelist data can be regularly distributed through the compliance management system. During execution, each financial institution involved in the payment path is compared to see if it belongs to the permitted list, thereby intercepting funds flowing to restricted institutions, abnormal channels, or institutions with high sanction risk.
[0055] For example, during the transaction initiation phase, an original message trace is generated for each transaction, and the unique transaction identifier, key timestamp, upstream signature information, and clearing path summary are written to the audit storage. Subsequently, when a high-risk payment confirmation request enters the first strategy, the verification engine first reads the trace of that transaction. When performing a full-field hash comparison, the verification engine concatenates the currently received payment confirmation request according to the predefined field order and compares it with the corresponding original summary in the trace. If a change is found in a memo field, payee account alias field, or intermediate line code field, even if it does not appear to affect the basic business process, it will be identified due to the inconsistency in the summary. When performing full-length digital signature verification, the verification engine uses the signature identifier in the request. The system retrieves the corresponding public key from the key management service and verifies the signature data covering the entire request packet. If an attacker only retains the correct amount and order number but tampers with the device fingerprint extension field or the payment path extension field, the situation of partial legality but overall illegality will be identified. When performing exchange rate snapshot recalculation, the system retrieves the exchange rate snapshot corresponding to the time of transaction initiation to avoid misjudgment due to time misalignment. If the converted amount in the message is found to be inconsistent with the snapshot derivation result after recalculation, it is determined that there is a risk of amount distortion. When performing intermediary bank compliance whitelist comparison, the system reads the intermediary bank identifiers involved in the payment path and matches them item by item with the whitelist version issued through the compliance management system. If the path is found to pass through an unauthorized institution, a risk alarm is output.
[0056] For example, field format distortion refers to a field having a value on the surface, but the encoding form, character composition, length structure, or type representation is inconsistent with the predetermined business specifications. For example, the amount field should be in standard numerical format but contains control characters; the order number field should be encoded according to fixed rules but contains abnormal escape segments; the currency field should come from standard currency enumeration but contains unregistered codes. Logical contradiction refers to multiple fields that may be legal when viewed individually, but when combined, they do not conform to business common sense or transaction rules. For example, there is a clear conflict between the country of receipt, payment currency, merchant's business scope, payment account region, and payment path.
[0057] For example, when a field format distortion or logical contradiction is detected during the security verification process, interrupting the security verification and adjusting the risk level of the corresponding payment confirmation request to a high-risk level can be understood as an abnormal upgrade and downgrade compensation rule. That is, a sentinel logic is embedded in the lightweight verification strategy. The sentinel logic can be set in the message parser and basic rule checker of the second strategy. The message parser is used to quickly scan the validity of field names, field types, character encodings, length ranges, and enumeration values in the request body. The basic rule checker is used to perform lightweight judgment on the matching relationship between business fields. For example, when the checker finds that multiple fields have a combination relationship that is difficult to establish, it is identified as a logical contradiction.
[0058] For example, when a payment confirmation request initially assessed as low-risk enters the preset lightweight verification queue and begins executing the second strategy, the verification engine may find that although the payment amount field has a correct signature, the data format has been distorted from the expected floating-point type to a string containing special characters, i.e., field format distortion. Or it may find that the delivery address is located in China, but the payment currency is specified as a rare overseas digital currency, i.e., logical contradiction. At this point, the abnormal upgrade and downgrade compensation rule is triggered, and the verification engine interrupts the currently ongoing lightweight verification process, forcibly overwriting the risk level label of the request from low risk to high risk.
[0059] Based on this, this application configures targeted defense projects for the first strategy to ensure the review of high-risk requests. Meanwhile, in the second strategy, a dynamic error correction mechanism based on runtime anomaly awareness is used. When a spoofed request attempts to exploit the lenient environment of the second strategy, format distortions or logical contradictions in the message will trigger compensation rules, causing the risk level to increase and the request to re-enter the strong verification queue. This achieves secondary correction during the verification execution phase, ensuring the rapid passage of the vast majority of genuine low-risk requests while effectively identifying attacks and improving the overall security of security management.
[0060] In the enterprise e-commerce transaction data security management method provided in some specific embodiments of this application, regarding the above step S130, the method may also include, but is not limited to, step S710.
[0061] Step S710: Logically associate the security verification result, risk level, verification strategy, and first strategy usage information or second strategy usage information corresponding to the payment confirmation request to generate a verification completion status certificate. The information used by the first strategy includes the full process data of the verification project of the first strategy, and the information used by the second strategy includes the full process data of triggering the upgrade and downgrade compensation rules during the security verification process.
[0062] For example, logical associations can be implemented using a unified data structure template, which includes at least a unique request identifier, a queue identifier, a risk level, a strategy identifier, a start verification time, an end verification time, a verification result, a result reason code, a strategy execution details area, and a compensation event area. If the request executes the first strategy, the first strategy usage information is written in the strategy execution details area. If the request executes the second strategy and triggers the abnormal upgrade / downgrade compensation rule, the second strategy usage information is written in the compensation event area.
[0063] The entire process records necessary information. For example, for the information used by the first strategy, it can record the start time, end time, input data source identifier, verification result, failure reason, whether the mandatory verification condition was hit, and the return status of related dependent services for each verification item. For the information used by the second strategy, it can record the specific fields that trigger the compensation rule, the exception category, the exception discovery stage, the risk level adjustment action, the re-routing action, and the context state when interrupted.
[0064] For example, the full-process data includes not only the final results of each inspection, but also intermediate variables and time consumption during the inspection process. Intermediate variables can be procedural information that has a direct explanatory effect on the result determination but should not be simplified into a single conclusion. For example, in the first strategy usage information, the message version identifier, digest source identifier, and comparison conclusion used in the full-field hash comparison can be recorded. In the full-length digital signature verification, the signature key identifier, signature coverage identifier, and verification return status can be recorded. In the exchange rate snapshot recalculation, the exchange rate snapshot version called, transaction time slice identifier, and conversion difference description can be recorded. In the intermediary bank compliance whitelist comparison, the intermediary bank code being compared, the whitelist version that is matched, and the reason for failure can be recorded.
[0065] For example, a layered record structure can be adopted. The basic information layer records the unique request identifier, the de-identified value of the user identifier, the order identifier, the risk level, the current load status, and the final verification result. The policy information layer records the actual verification policy executed, the target queue entered, the policy version number, and the execution instance identifier. The process information layer records the execution trajectory of each verification step, key events, reasons for exceptions, and time intervals. The security encapsulation layer records the generation time, encryption flag, and integrity verification flag.
[0066] Logical association refers to the assembly of information at each layer around the same unique identifier of the request, and the time sequence of the information can represent the actual processing process. For example, if the request first executes the second strategy in the preset lightweight verification queue, and then upgrades to the preset strong verification guarantee queue to execute the first strategy due to the hit compensation rule, the certificate will record the events in the order of the second strategy triggering the abnormal upgrade and downgrade compensation rule, the risk level being adjusted to a high risk level, re-routing to the preset strong verification guarantee queue, and the first strategy starting to execute.
[0067] Based on this, this application constructs a holographic verification completion status certificate by logically linking the verification results, risk level, strategy type, and detailed data of the entire strategy execution process. This provides rich data support for subsequent release control, security audit, and risk control model iteration. The release control module can execute more complex release logic based on detailed data. Security auditors can reconstruct the interception process of abnormal transactions and locate attack methods by analyzing the certificate. The risk control algorithm team can use the full-process data to optimize the front-end risk assessment model, thereby forming a self-evolving security closed-loop ecosystem.
[0068] In the enterprise e-commerce transaction data security management method provided in some specific embodiments of this application, the verification completion status certificate includes a payment confirmation request that represents a high-risk level being security verified in a strong verification guarantee queue, and the verification result representing a high-risk verification suspension status certificate that fails the verification; regarding the above step S140, the method may also include, but is not limited to, steps S810 and S820.
[0069] Step S810: If the verification completion status certificate is a high-risk verification suspended status certificate, input the verification completion status certificate into the preset release control matrix to reject the release of the corresponding payment confirmation request, transfer the corresponding payment confirmation request to the waiting queue, and re-perform security verification after the current load status is changed to the normal load status. Step S820: If the verification completion status certificate is not a high-risk verification suspended status certificate, input the verification completion status certificate into the preset release control matrix, so as to release the corresponding payment confirmation request according to the association logic of multiple data represented by the verification completion status certificate.
[0070] For example, during the certificate generation stage, the certificate type can be semantically specialized. When the initial risk level or upgraded risk level of a payment confirmation request is high risk and it is assigned to the preset strong verification guarantee queue to execute the first strategy, but fails any of the following: full-field hash comparison, full-length digital signature verification, exchange rate snapshot recalculation, or intermediate bank compliance whitelist comparison, or when the timeout judgment in the mandatory verification condition is triggered due to the timeout of the underlying dependency, a high-risk verification suspended status certificate can be generated, including a clear suspended identifier, original request message digest, failed item code, occurrence timestamp, and current system load status marker.
[0071] For example, the preset release control matrix is a multi-dimensional decision table loaded by the policy rule engine. The input dimensions can ignore the voucher type, risk level, verification strategy type, failure reason code in the whole process data, and the current system load status. After the release control module intercepts the payment confirmation request carrying the voucher, it first parses the voucher. When the release control matrix matches the current voucher as a high-risk verification suspended voucher, the release control matrix hits the suspended interception branch, blocking the request from entering the clearing database or downstream deduction service, and returns an asynchronous waiting response to the front end. At the same time, the complete context of the payment confirmation request is pushed into the persistent waiting queue. During the waiting period, the release control module or the background scheduling task continuously subscribes to the system load status published by the running monitoring module. When the comprehensive load index is detected to have dropped and changed from a light or heavy overload state to a normal load state, it indicates that the computing resources have been fully restored. The scheduling task wakes up the suspended request from the waiting queue and resubmits it to the preset strong verification guarantee queue, and re-executes all verification items in the first strategy to obtain a deterministic final verification conclusion.
[0072] For example, when a credential is not a high-risk, pending verification credential, such as a low-risk request that passes the second strategy in the lightweight verification queue, a medium-risk request that passes verification in the elastic queue, or a regular failed credential that is clearly determined to be an illegal forged message, the release control matrix combines and analyzes multiple data related to the internal logic of the credential. For example, for credentials with low risk level, no abnormalities in the second strategy and data throughout the process, and that have passed verification, the matrix directly outputs a release decision, and the request is successfully transferred to the fund clearing stage. For credentials with low risk level and those that trigger the abnormal upgrade and downgrade compensation rule but pass the first strategy after upgrade, the matrix adds an audit trail label before releasing the credential. For credentials that are confirmed to be blacklisted or have a damaged digital signature and no pending mark, the matrix outputs a final blocking instruction, permanently terminating the transaction and releasing the relevant resources.
[0073] Based on this, this application implements elastic fault tolerance based on the linkage between status credentials and load recovery in the release decision stage. By marking high-risk requests that fail verification due to resource shortages or timeouts during overload periods as suspended, and performing secondary verification after the load recovers and stabilizes, it ensures that high-risk requests are not released without deep verification, avoids rejecting potentially legitimate normal transactions during high-pressure periods, improves the transaction success rate and business continuity in concurrent scenarios such as flash sales, and achieves a balance between high-level security protection and user transaction experience.
[0074] In the enterprise e-commerce transaction data security management method provided in some specific embodiments of this application, the abnormal upgrade / downgrade compensation rules also include: When a field format distortion or logical contradiction is detected during the security verification process, the source IP network segment, device fingerprint prefix, and distortion error code type of the corresponding payment confirmation request are extracted and combined to obtain the abnormal feature key. Within a preset time window, the frequency of triggering abnormal feature keys is counted in real time, and the abnormal arrival rate is calculated based on the triggering frequency. If the abnormal arrival rate is lower than the preset escape threshold, the security verification is interrupted and the risk level of the corresponding payment confirmation request is adjusted to the high risk level. If the abnormal arrival rate is higher than the preset escape threshold, the security verification is interrupted and the corresponding payment confirmation request is removed.
[0075] For example, when a field format distortion or logical contradiction is detected during the security verification process, the system extracts and combines the source IP network segment, device fingerprint prefix, and distortion error code type of the corresponding payment confirmation request to obtain the abnormal feature key. This technique adds feature extraction and dimensionality reduction aggregation logic to the sentinel interception point of the second strategy.
[0076] When the message parser or basic rule checker in the second strategy identifies field format distortion or logical contradiction, it can extract the source Internet Protocol address segment of the current request network layer in real time, such as the source IP segment extracted by aggregating Class C address / 24 subnet mask, and the device fingerprint prefix of the terminal layer, such as the first sixteen bits of the device fingerprint hash value, to identify the same batch of simulators or script clusters, as well as the distortion error code type thrown by the rule engine. Then, the source IP segment, device fingerprint prefix and distortion error code type are concatenated or hashed according to a predetermined separator to generate a unique abnormal feature key.
[0077] For example, a sliding time window counter can be maintained in memory or a high-speed distributed cache. The preset time window can be set to a relatively short sliding observation period. Using the exception feature key as an index, the cumulative trigger frequency of the exception feature key in the current time window is recorded in real time. The exception arrival rate per unit time is calculated by dividing the trigger frequency by the time window length, which is the number of exception requests per second. The preset escape threshold is set according to the maximum concurrent load that the preset strong verification guarantee queue can bear.
[0078] If the calculated abnormal arrival rate is lower than the preset escape threshold, it indicates that the current distortion is a sporadic, occasional, low-frequency anomaly, which may be due to input deviations of real users, occasional network packet loss and packet splicing errors, or single-point probe behavior of individual black market operators. It is determined that the request is worth further in-depth review and will not cause congestion to the strong verification queue. Therefore, the current lightweight verification is interrupted, the request risk level is adjusted to a high-risk level, and the request is re-routed to the preset strong verification protection queue to execute the first strategy.
[0079] If the calculated abnormal arrival rate is higher than or equal to the preset escape threshold, it indicates that the system is currently under high-frequency coordinated attack from similar device clusters in the same network segment, with homogeneous distorted packets. The attacker intends to use the upgrade and downgrade rules to fill the strong verification queue. At this time, the request is determined to be malicious attack traffic that consumes system resources. The circuit breaker action is executed, the current security verification is immediately interrupted, and the corresponding payment confirmation request is discarded or removed. That is, the request context is directly destroyed, an interception error code is returned to the gateway, or the request is added to a temporary blacklist, and entry into the preset strong verification protection queue is prohibited.
[0080] Based on this, this application can accurately distinguish between sporadic potential risk transactions and clustered resource-consuming attacks by extracting composite dimension abnormal feature keys and combining them with sliding window to calculate abnormal arrival rate. While ensuring that single suspicious transactions can be successfully upgraded to accept in-depth review, it avoids high-frequency attack traffic implemented by compensation rules from entering the strong verification resource pool, effectively preventing the pre-set strong verification protection queue from congestion and collapse, and ensuring the high availability of security verification units from the architectural level.
[0081] Secondly, this application provides a secure management system for enterprise e-commerce transaction data, including: The information acquisition module is used to acquire transaction attribute information, user behavior information, device identification information, and security context information for each payment confirmation request; The risk assessment module is used to assess the risk level of each payment confirmation request based on transaction attribute information, user behavior information, device identification information, and security context information. The security verification module is used to obtain the load monitoring indicators of the verification unit used for security verification, determine the current load status based on the load monitoring indicators, and call the corresponding verification strategy to perform security verification on the payment confirmation request based on the current load status and the risk level of each payment confirmation request, obtain the security verification result, and generate a verification completion status certificate based on the security verification result. The release control module is used to release the corresponding payment confirmation request based on the verification completion status certificate.
[0082] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.
Claims
1. A method for secure management of transaction data in an enterprise e-commerce platform, characterized in that, include: Obtain transaction attribute information, user behavior information, device identification information, and security context information for each payment confirmation request; Based on the transaction attribute information, user behavior information, device identification information, and security context information, assess the risk level of each payment confirmation request; Obtain the load monitoring indicators of the verification unit used for security verification, determine the current load status based on the load monitoring indicators, and call the corresponding verification strategy to perform security verification on the payment confirmation request based on the current load status and the risk level of each payment confirmation request, obtain the security verification result, and generate a verification completion status certificate based on the security verification result. Based on the verification completion status certificate, the corresponding payment confirmation request is released.
2. The enterprise e-commerce transaction data security management method according to claim 1, characterized in that, The assessment of the risk level of each payment confirmation request based on the transaction attribute information, user behavior information, device identification information, and security context information includes: Extract continuous numerical fields from the transaction attribute information and user behavior information, calculate the deviation of the continuous numerical fields, and obtain the numerical deviation score of each payment confirmation request. Extract discrete security feature data from the device identification information and security context information. The discrete security feature data includes device fingerprint, Internet Protocol address, and pre-positioned risk control tag. Risk analysis is performed based on the device fingerprint, Internet Protocol address, and pre-installed risk control tag to obtain a risk score for each payment confirmation request; The risk level of each payment confirmation request is determined based on the numerical deviation score and risk score of each payment confirmation request.
3. The enterprise e-commerce transaction data security management method according to claim 1, characterized in that, The step of acquiring the load monitoring indicators of the verification unit used for security verification, and determining the current load status based on the load monitoring indicators, includes: Obtain the load monitoring metrics of the verification unit used for security verification. The load monitoring metrics include processor utilization, memory usage, network throughput, the total number of payment confirmation requests in the current buffer queue, the request arrival rate within a historical preset monitoring period, and the average waiting time of payment confirmation requests in the current buffer queue. The load monitoring index is input into the preset comprehensive load index calculation formula to obtain the current comprehensive load index. The current load status is determined based on the position of the current comprehensive load index in the preset segmented judgment interval.
4. The enterprise e-commerce transaction data security management method according to claim 1, characterized in that, The current load status includes normal load status, light overload status and heavy overload status; the risk level includes high risk level, medium risk level and low risk level; the verification strategy includes a first strategy and a second strategy, wherein the verification level of the first strategy is higher than the verification level of the second strategy. Based on the current load status and the risk level of each payment confirmation request, the corresponding verification strategy is invoked to perform security verification on the payment confirmation requests, and the security verification results are obtained, including: When the current load status is normal load status, each payment confirmation request is loaded into a preset strong verification guarantee queue to call the first strategy to perform security verification on the payment confirmation request and obtain the security verification result; When the current load state is either slightly overloaded or heavily overloaded, high-risk payment confirmation requests are loaded into a preset strong verification queue to invoke the highest security level verification strategy to perform security verification on the payment confirmation requests and obtain a security verification result. Medium-risk payment confirmation requests are loaded into a preset flexible verification queue to invoke either the first strategy or the second strategy to perform security verification on the payment confirmation requests according to changes in the current load state and obtain a security verification result. Low-risk payment confirmation requests are loaded into a preset lightweight verification queue to invoke the corresponding second strategy to perform security verification on the payment confirmation requests and obtain a security verification result.
5. The enterprise e-commerce transaction data security management method according to claim 4, characterized in that, Before invoking the corresponding verification strategy to perform security verification on the payment confirmation requests based on the current load status and the risk level of each payment confirmation request, the enterprise e-commerce transaction data security management method further includes: Forced verification conditions are set for the preset strong verification guarantee queue so that all verification items in the first strategy are forcibly completed when security verification is performed in the preset strong verification guarantee queue, and the adjustment or modification of the first strategy is restricted.
6. The enterprise e-commerce transaction data security management method according to claim 5, characterized in that, The verification items for the first strategy include full-field hash comparison, full-length digital signature verification, exchange rate snapshot recalculation, and comparison with the compliance whitelist of intermediary banks; The second strategy includes anomaly upgrade / downgrade compensation rules, which include: If a field format distortion or logical contradiction is detected during the security verification process, the security verification will be interrupted and the risk level of the corresponding payment confirmation request will be adjusted to a high-risk level.
7. The enterprise e-commerce transaction data security management method according to claim 6, characterized in that, The step of generating a verification completion status credential based on the security verification result includes: Logically associate the security verification result, risk level, verification strategy, and first strategy usage information or second strategy usage information corresponding to the payment confirmation request to generate a verification completion status certificate. The information used by the first strategy includes the full process data of the verification items of the first strategy, and the information used by the second strategy includes the full process data of triggering the upgrade / downgrade compensation rules during the security verification process.
8. The enterprise e-commerce transaction data security management method according to claim 7, characterized in that, The verification completion status certificate includes a high-risk verification suspension status certificate that indicates that the payment confirmation request has been securely verified in the strong verification guarantee queue and the verification result indicates that the verification failed. The step of releasing the corresponding payment confirmation request based on the verification completion status certificate includes: If the verification completion status certificate is a high-risk verification suspended status certificate, the verification completion status certificate is input into the preset release control matrix to reject the release of the corresponding payment confirmation request, and the corresponding payment confirmation request is transferred to the waiting queue and re-verified after the current load status is changed to the normal load status. If the verification completion status certificate is not a high-risk verification suspension status certificate, the verification completion status certificate is input into a preset release control matrix to release the corresponding payment confirmation request based on the association logic of multiple data represented by the verification completion status certificate.
9. The enterprise e-commerce transaction data security management method according to claim 6, characterized in that, When a field format distortion or logical contradiction is detected during the security verification process, the security verification is interrupted and the risk level of the corresponding payment confirmation request is adjusted to a high-risk level, including: When a field format distortion or logical contradiction is detected during the security verification process, the source IP network segment, device fingerprint prefix, and distortion error code type of the corresponding payment confirmation request are extracted and combined to obtain the abnormal feature key. Within a preset time window, the trigger frequency of abnormal feature keys is counted in real time, and the abnormal arrival rate is calculated based on the trigger frequency. If the abnormal arrival rate is lower than a preset escape threshold, the security verification is interrupted and the risk level of the corresponding payment confirmation request is adjusted to a high risk level. If the abnormal arrival rate is higher than the preset escape threshold, the security verification is interrupted and the corresponding payment confirmation request is removed.
10. A secure management system for enterprise e-commerce transaction data, characterized in that, include: The information acquisition module is used to acquire transaction attribute information, user behavior information, device identification information, and security context information for each payment confirmation request; The risk assessment module is used to assess the risk level of each payment confirmation request based on the transaction attribute information, user behavior information, device identification information, and security context information. The security verification module is used to obtain the load monitoring indicators of the verification unit used for security verification, determine the current load status based on the load monitoring indicators, call the corresponding verification strategy to perform security verification on the payment confirmation request based on the current load status and the risk level of each payment confirmation request, obtain the security verification result, and generate a verification completion status certificate based on the security verification result. The release control module is used to release the corresponding payment confirmation request based on the verification completion status certificate.
