An order rolling processing optimization method and system of a point mall based on big data

Abstract

The application provides a big data-based integral mall order rolling processing optimization method and system, which comprises the following steps: for the conflict order list, a dynamic priority adjustment mechanism is adopted, the order processing priority is adjusted based on the user behavior trajectory data including historical exchange preferences, and a resource allocation scheme is obtained; the order sequence is integrated through a queue reordering mechanism, the order sequence is rearranged based on the resource allocation scheme, a final processing sequence is obtained, the high-priority request is preferentially executed, and the queue stability is maintained; if there is a remaining conflict in the final processing sequence, a standby resource set scheduling mechanism is used to extract supplementary resources from a pre-device inventory, a complete execution plan is obtained, and the plan covers all order processing paths; based on the complete execution plan, the order processing progress is monitored, the real-time state is fed back through progress tracking and abnormality detection, iterative adjustment is performed, and a stable running order flow is obtained.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method and system for optimizing the rolling processing of orders in an points mall based on big data. Background Technology

[0002] As a crucial component of e-commerce and user incentive mechanisms, points malls play a key role in enhancing user engagement and platform loyalty. With the rapid development of the digital economy, points malls need to process massive amounts of order data and meet diverse user redemption needs. Especially during peak periods such as promotional activities or holidays, order processing efficiency and the rationality of resource allocation directly impact user experience and platform operational effectiveness. However, traditional order processing methods often struggle to meet the demands of modern points malls for efficient and intelligent processing when faced with complex and ever-changing user behavior and dynamic resource needs.

[0003] Currently, order processing methods in points malls largely rely on static rules or simple priority scheduling, making it difficult to cope with dynamic changes in user redemption behavior. For example, existing systems are typically based on fixed order queues, unable to flexibly adjust resource allocation according to user behavior or seasonal consumption trends. This approach easily leads to uneven resource allocation during peak periods, with some orders delayed while other resources remain idle. Furthermore, existing systems lack real-time monitoring capabilities for order processing, failing to quickly identify the causes of delays and resulting in a degraded user experience. This static and fragmented approach limits the operational efficiency of points malls in complex scenarios.

[0004] One of the core technical challenges in processing orders in a points mall lies in how to dynamically adjust the order queue to adapt to fluctuations in user redemption demand. The complexity of user behavior patterns and consumption patterns makes predicting order demand exceptionally difficult. For example, during large-scale promotional events, users may concentrate on redeeming popular items, leading to intensified competition for specific resources, while existing systems struggle to dynamically allocate processing capacity based on real-time demand. This lack of dynamic adjustment directly results in low order processing efficiency and can even cause system congestion.

[0005] Another key technical challenge is managing conflicts arising from competing order resources. When multiple orders compete for limited processing resources simultaneously, such as for popular items with limited stock, the system often fails to effectively identify and handle such conflicts. Resource competition between orders can lead to delays in processing some orders, or even failure due to insufficient resources. For example, when a user attempts to redeem a limited-edition item, the system may fail to detect competing requests from other orders in time, causing processing order chaos and ultimately resulting in redemption failure or long waiting times for the user. This lack of conflict management further exacerbates the complexity of order processing.

[0006] Dynamically adjusting order queues and effectively managing resource contention in high-concurrency scenarios has become a key issue for improving the operational efficiency of points malls. In actual business operations, such as during promotional periods, points malls may receive thousands of redemption requests simultaneously. However, the inventory and processing capacity of popular items are limited. The system needs to quickly determine which orders can be prioritized and how to avoid processing failures due to resource contention. Therefore, solving the problems of dynamic order queue adjustment and resource conflict management has become crucial for the intelligent and stable operation of points malls. Summary of the Invention

[0007] On the one hand, this invention provides a method for optimizing the rolling processing of orders in an points mall based on big data, mainly including:

[0008] By collecting order inflow data and user behavior trajectory data from the points mall, demand change parameters are calculated. Combined with historical load data and a preset load coefficient, peak load levels are determined, reflecting the peak demand for system processing capacity. Based on these peak load levels, a priority sorting algorithm is used to classify order queues, resulting in order sequences divided into high-priority requests and regular requests. When the proportion of high-priority requests in the order sequence exceeds a preset threshold, the number of requests within the concurrent access window is compared with the real-time inventory quantity multiplied by the inventory safety margin coefficient to identify conflict points due to insufficient inventory, resulting in a conflict order list. For the conflict order list, a preset weighted scoring model is used, combined with historical redemption preference features from the user behavior trajectory data, to adjust order processing priorities and obtain a resource allocation scheme. The order sequences are integrated through a queue reordering mechanism, and the order is rearranged based on the resource allocation scheme to obtain the final processing order. If there are remaining conflicts in the final processing order, supplementary resources are extracted from the preset backup inventory through the backup resource set scheduling mechanism to obtain a complete execution plan covering all order processing paths; based on the complete execution plan, the order processing progress is monitored, and real-time status is fed back through progress tracking and anomaly detection, and iterative adjustments are made to obtain a stable order flow.

[0009] On the other hand, this invention also provides a big data-based points mall order rolling processing optimization system, mainly comprising: a data acquisition and demand calculation module, used to collect points mall order inflow data and user behavior trajectory data, calculate demand change parameters, and determine peak load levels by combining historical load data and preset load coefficients, wherein the level reflects the peak demand of the system's processing capacity; an order classification and queue management module, used to classify order queues based on the peak load level using a priority sorting algorithm, resulting in an order sequence that divides orders into high-priority requests and regular requests; a resource contention detection module, used to identify conflict points of insufficient inventory and obtain a conflict order list when the proportion of high-priority requests in the order sequence exceeds a preset threshold by comparing the number of requests within the concurrent access window with the real-time inventory quantity multiplied by the inventory safety margin coefficient; a dynamic priority adjustment module, used to adjust the order processing priority for the conflict order list using a preset weighted scoring model, combined with historical redemption preference features in the user behavior trajectory data, to obtain a resource allocation scheme; and a queue reordering module, used to integrate the order sequence through a queue reordering mechanism, rearrange the order based on the resource allocation scheme, and obtain the final processing order. A backup resource scheduling module is used to extract supplementary resources from a preset backup inventory through a backup resource set scheduling mechanism if there are remaining conflicts in the final processing order, thereby obtaining a complete execution plan covering all order processing paths. A progress monitoring and iterative adjustment module is used to monitor the order processing progress based on the complete execution plan, and through progress tracking and anomaly detection, provide real-time status feedback, iteratively adjust, and obtain a stable order flow. The technical solution provided by this embodiment of the invention can include the following beneficial effects:

[0010] This invention discloses a method and system for optimizing the rolling processing of orders in a points mall based on big data. Addressing the problems of surged load, resource contention conflicts, and low order processing efficiency during peak periods in points malls, this invention constructs a dynamically optimized order processing flow by integrating user behavior trajectory data and order inflow data. The invention collects browsing time and redemption frequency to calculate demand change parameters to determine peak load levels. It employs a priority ranking algorithm to divide orders into high-priority and regular requests. For situations where the proportion of high-priority requests for frequently redeemed items exceeds a threshold, a resource contention detection module scans inventory and concurrent access status to identify conflicting orders. A dynamic priority adjustment mechanism, combined with historical redemption preferences, optimizes resource allocation. This invention integrates order sequences through a queue reordering mechanism, prioritizing the execution of high-priority requests. Simultaneously, a backup resource scheduling mechanism replenishes inventory, generating a complete execution plan covering all orders. Real-time monitoring and anomaly detection iterative adjustments ensure stable order flow operation. This invention improves order processing efficiency during peak periods in points malls, optimizes resource allocation, reduces conflict rates, and enhances system stability and user experience. Attached Figure Description

[0011] Figure 1 This is a flowchart of an optimization method for the rolling processing of orders in a points mall based on big data, according to the present invention.

[0012] Figure 2 This is a schematic diagram of the DVP demand change parameter calculation model in an embodiment of the present invention.

[0013] Figure 3 This diagram illustrates the concurrent access window (CAW) and inventory safety margin (ISM) detection mechanism of this invention.

[0014] Figure 4 This is a schematic diagram illustrating the working principle of the dynamic priority adjustment mechanism in an embodiment of the present invention.

[0015] Figure 5 A flowchart of the standby resource set scheduling mechanism according to an embodiment of the present invention.

[0016] Figure 6 This is a schematic diagram of the structure of an order rolling processing optimization system for an points mall based on big data, according to the present invention.

[0017] Figure 7 This is a comparison chart showing the effectiveness of the method of this invention and existing technologies in handling scarce goods during peak periods. Detailed Implementation

[0018] To further understand the content of this invention, a detailed description of the invention is provided in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0019] Example 1

[0020] like Figure 1 This embodiment presents an optimization method for the rolling processing of orders in a points mall based on big data, which may specifically include:

[0021] S101. By collecting order inflow data and user behavior trajectory data from the points mall, including browsing time and redemption frequency, the demand change parameters are calculated to obtain the peak load level, which reflects the peak demand of the system's processing capacity.

[0022] For example, such as Figure 2 As shown, in one implementation, to accurately quantify system pressure over a future period, the present invention first calculates a Demand Variation Parameter (DVP). Specifically, the system collects behavioral trajectory data of all users within a preset time window T (e.g., the past hour) and calculates DVP according to the following formula: .

[0023] in:

[0024] i and j respectively iterate through the browsing events and redemption events within the time window T.

[0025] B i It is the effective browsing duration of the i-th browsing event, which refers to the duration for which the user stays on the product page for more than a preset threshold (e.g., 5 seconds).

[0026] Ej is the high-value identifier for the j-th redemption event, for example, 1 when redeeming popular or high-point items, and 0 otherwise.

[0027] w i wj and wj are time decay weights, for example, w = e^(-λ*Δt), where Δt is the time since the event occurred, and λ is the decay coefficient. Their values ​​typically range from 0.1 to 0.5, and are used to adjust the influence of recent behavior on the prediction result. The larger the value of λ, the higher the weight of recent behavior, giving it greater weight.

[0028] α and β are the business weighting coefficients for browsing and redemption behaviors.

[0029] N_B and N_E are the total number of views and total number of redemptions within the time window T, respectively.

[0030] The DVP value directly reflects the user attention invested in a unit exchange behavior. The higher the value, the stronger the potential exchange demand.

[0031] Subsequently, the system determines the Peak Load Level (PLL) based on the DVP value, calculated using the following formula: .

[0032] in:

[0033] C_base is the system's baseline processing capacity during normal periods, such as processing 100 orders per second.

[0034] K_load is a preset load factor used to adjust the degree of influence of DVP on the load level.

[0035] For example, if the calculated DVP is 0.8 and K_load is 0.5, the predicted peak load level will reach 1.4 times the baseline processing capacity. Based on this, the system pre-allocates resources, such as dynamically increasing server instances or database connection pool size. This quantitative prediction method provides accurate data input for subsequent order processing strategies.

[0036] S102. Based on the peak load level, the order queue is classified using a priority sorting algorithm to obtain a classified order sequence. The sequence divides the orders into high-priority requests and regular requests. The high-priority requests involve frequently redeemed points items.

[0037] Based on the load level assessment, high-frequency redemption identification features are extracted from the order queue using behavioral trajectory fusion. This extraction is achieved by statistically counting redemption frequencies and combining them with user request stratification to obtain a preliminary priority list. This list marks frequently redeemed points items as high-priority requests. For this preliminary priority list, peak demand response data is obtained and combined with a priority sorting mechanism to perform stratified processing on the order queue. The peak demand response data is derived from the load level assessment to determine the categorized order sequence, which distinguishes between high-priority requests and regular requests. From the categorized order sequence, if the proportion of high-priority requests exceeds a preset threshold, the points item classification is adjusted through resource priority allocation. This adjustment uses sequence generation logic to fuse behavioral trajectories, resulting in an optimized sequence response scheme. High-frequency redemption, as referred to in this invention, means a product whose redemption frequency exceeds the average redemption frequency of all products within a preset statistical period (e.g., the past 24 hours).

[0038] For example, in one implementation, the system uses a priority sorting algorithm to classify the order queue based on the previously calculated peak load level.

[0039] Specifically, peak load levels are used as input parameters to adjust the algorithm's thresholds. For example, when the load level exceeds 120% of the normal value, the algorithm prioritizes orders involving frequently redeemed points items. This ensures that the system can efficiently allocate resources during peak periods in the points mall. Furthermore, the implementation of the priority ranking algorithm includes several key steps. First, the system extracts relevant attributes for each order from the order queue, such as the redemption frequency of points items and user behavior data. Frequently redeemed points items refer to those that have been redeemed more times than the average in the past 24 hours.

[0040] For example, popular e-coupons or virtual items. The algorithm calculates a priority score by comparing the redemption frequency of each order with a preset threshold. Specifically, each order is assigned a score based on a frequency-weighted formula.

[0041] For example, frequently purchased items receive a 30% bonus on points, and the queue is then sorted from highest to lowest points. This categorized order sequence places high-priority requests at the front of the queue and regular requests at the back. High-priority requests specifically involve frequently redeemed point-based items, ensuring these orders are processed first during peak load periods to avoid delays. The algorithm works by dynamically adjusting the sorting rules based on load levels.

[0042] For example, in high-load scenarios, the priority weight of frequently purchased items is increased to maintain system responsiveness. This detailed classification logic helps optimize order flow in the points mall, especially during periods of high user activity. When the system detects an increase in load levels during peak periods, it automatically triggers a priority sorting algorithm to classify the real-time order queue. Assuming an order queue contains 10 orders, 3 of which involve frequently redeemed points items, such as mobile phone accessories redeemed with points, will be classified as high-priority requests. The remaining orders, such as low-frequency household goods, will be regular requests. The categorized sequence allows the system to process the high-priority portion first, thereby reducing overall waiting time.

[0043] In one possible implementation, this technical solution is extended to member activity scenarios in a points mall. Specifically, when member day activities cause a surge in load levels, the algorithm categorizes the order queue, prioritizing orders for frequently redeemed points items.

[0044] For example, frequently redeemed points packages are identified as high-priority, ensuring these requests are queued and executed early, while regular requests such as one-off points queries are deferred. This categorization enhances the system's robustness, adapting to varying activity intensities within the same domain.

[0045] It should be noted that the priority ranking algorithm emphasizes objective classification criteria in its business process. The determination of high-frequency redemptions is based on historical data statistics. For example, the system queries the database for redemption records of goods, calculates the average frequency, and if the number of redemptions for a certain goods exceeds a threshold, such as 50 times per day, then its related orders receive high priority. After obtaining the categorized order sequence, the system can allocate server resources accordingly to ensure stable operation under peak load conditions.

[0046] S103. When the proportion of high-priority requests in the order sequence exceeds a preset threshold, the conflict point of insufficient inventory is identified by comparing the number of requests in the concurrent access window with the real-time inventory quantity multiplied by the inventory safety margin coefficient, and a conflict order list is obtained.

[0047] For example, such as Figure 3 As shown, in one implementation, when the proportion of high-priority requests exceeds a preset threshold (e.g., 10%), the system activates a resource contention detection module. To accurately identify the real system bottleneck caused by a sudden surge in traffic, rather than misjudging it due to long-term accumulated requests, this module does not simply compare the total number of requests with inventory, but employs a more refined judgment logic. First, the module defines a Concurrent Access Window (CAW) with a duration of Δt, for example, 2 seconds. The module counts in real-time the number of all redemption requests N_req for the same high-priority item within any CAW, reflecting the maximum concurrent pressure the system experiences instantaneously. Second, the module introduces the concept of an Inventory Safety Margin (ISM), whose value is a coefficient S_factor less than 1, for example, 0.9, representing the proportion of inventory reserved to cope with emergencies. When making a conflict judgment, only 90% of the real-time inventory is considered as safely allocated capacity, with the remaining 10% reserved. The conflict trigger condition is defined as: N_req > (Current_Stock * S_factor). Where Current_Stock is the real-time inventory of the product. The system only determines that a resource conflict exists when the instantaneous number of requests within a concurrent window exceeds the safe capacity of the inventory, and adds all N_req orders to the conflict order list.

[0048] For example, a product has a real-time inventory of 100 units and a safety margin coefficient of 0.9. Within a 2-second CAW (Concurrent Access Window), the system detects 120 redemption requests. Since 120 > (100 * 0.9), the conflict condition is met. This detection method based on time windows and safety margins can more accurately identify the real system bottlenecks, avoid misjudgments due to temporary request fluctuations, and improve the accuracy of detection and the stability of the system. Compared to the simple method of comparing the total number of requests with the total inventory over a period of time, this invention, by introducing a concurrent access window (CAW), can more accurately capture the instantaneous traffic surges that cause system bottlenecks, avoiding misjudgments caused by accumulating scattered requests over a long period. At the same time, the introduction of an inventory safety margin (ISM) provides fault tolerance for unexpected delays in the system processing flow or minor lags in inventory data synchronization, improving the accuracy of the entire conflict detection mechanism and the robustness of the system.

[0049] It should be noted that after obtaining the list of conflicting orders, this list contains attributes of the conflicting orders, such as product ID, request time, and priority. The generation of this list is based on the module's judgment output, for example, using a sorting algorithm to place orders with a higher degree of conflict at the front of the list.

[0050] S104. For the list of conflicting orders, a preset weighted scoring model is used, combined with the historical redemption preference features in the user behavior trajectory data, to adjust the order processing priority and obtain the resource allocator.

[0051] The conflicting order list is obtained, historical redemption preferences are extracted from the user behavior trajectory data, and priority weight values ​​are calculated through behavior data fusion to obtain an adjusted priority sequence. For the adjusted priority sequence, a dynamic priority adjustment mechanism based on historical redemption preferences is adopted, incorporating redemption frequency statistics obtained from the historical redemption preferences to determine the order processing order corresponding to the order processing priority. Based on the order processing order and the allocation balancing mechanism extracted from the order processing order, a resource allocation scheme is obtained.

[0052] For example, such as Figure 4 As shown, for the identified list of conflicting orders, a Dynamic Priority Scoring Model (DPSM) is used to recalculate the priority of each order. This model is a linear weighted model, and the formula is as follows: Where P_score is the final priority score, w1 to w4 are weight coefficients that can be trained from historical data or configured by operational strategies, and Σw = 1.

[0053] The various feature factors are defined as follows: Historical redemption preference factor (F_pref): quantifies the user's preference for the currently conflicting product. It is calculated as: F_pref = (number of times the user has redeemed this product category in the past) / (total number of times the user has redeemed in the past). For example, if user A has redeemed a total of 10 times in the past, 3 of which were for transportation cards, then when he redeems a commemorative subway card, his F_pref is 0.3.

[0054] User level factor (F_level): Maps a user's membership level to a normalized numerical value. For example, levels from V1 to V5 are linearly mapped to 0.2 to 1.0.

[0055] Order wait time factor (F_wait): To reflect urgency but avoid infinite growth, a logarithmic function is used for quantification: F_wait = log 10 (T_wait + 1), where T_wait is the number of seconds the order has been waiting since it was submitted.

[0056] Product scarcity factor (F_scarce): Reflects the intensity of current resource competition. It is calculated as: F_scarce = (Total number of requests within the concurrent access window Δt) / (Current remaining inventory of the product).

[0057] Using the DPSM model, the system calculates an accurate and dynamic P_score for each conflicting order. For example, an order from a high-level user (F_level=1.0) with extremely high historical preference (F_pref=0.8), even if submitted slightly later, may have a higher P_score than an order from a low-level user who clicked by chance, thus gaining higher priority in the resource allocation scheme.

[0058] For example, the weight coefficients w1 to w4 in the Dynamic Priority Scoring Model (DPSM) can be determined through offline training. For instance, massive amounts of order data from historical promotional activities and their processing results (such as whether users complained, whether they repurchased, etc.) can be collected as training samples. With the optimization objective of maximizing user satisfaction or minimizing operating costs, machine learning algorithms such as gradient descent can be used to iteratively solve for the optimal combination of weight coefficients. During system operation, these weights can also be dynamically adjusted based on real-time operational strategies.

[0059] Preferably, to enhance the flexibility of the mechanism, a real-time feedback loop can be introduced. If conflicts persist after adjustments, the system will recalculate priorities based on the latest user trajectory data. For example, in one scenario, if a user continues to browse related products after submitting an order, the system will detect this behavior and dynamically increase their preference score to ensure the adaptability of the allocation scheme. In another possible implementation, the resource allocation scheme is generated by applying the adjusted priorities to the actual allocation logic.

[0060] Specifically, for high-priority orders, the system directly allocates resources; for low-priority orders, they are either placed in a waiting queue or alternative resource suggestions are provided. In this way, the solution not only resolves conflicts but also improves user satisfaction, such as ensuring loyal users receive priority access to resources during limited-time flash sales.

[0061] S105. The order sequence is integrated through a queue reordering mechanism, and the order is rearranged based on the resource allocation scheme to obtain the final processing order. The high-priority requests are executed first, while maintaining queue stability.

[0062] The order sequence is obtained, priority indicators are extracted from the resource plan, and the bubble sort algorithm is used to compare, exchange, and integrate the sequence pairwise to obtain a preliminary reordered sequence. For the preliminary reordered sequence, the queue stability attribute extracted from the preliminary reordered sequence is incorporated. If a high-priority request exceeds a preset threshold, the execution position of the high-priority request is adjusted to determine the intermediate processing order. Based on the intermediate processing order and the sequence integration attribute extracted from the intermediate processing order, low-priority requests are rearranged to obtain the final processing order.

[0063] For example, in one implementation, the order sequence is integrated through a queue reordering mechanism. First, it's necessary to understand the concept of a queue reordering mechanism. This mechanism is a priority-based order arrangement method used on e-commerce points redemption platforms to handle order flows when resources are limited.

[0064] Specifically, it takes the existing order sequence as input and adjusts the sequence by combining it with a previously obtained resource allocation scheme. The resource allocation scheme includes priority information for each order, such as a quantified score based on users' historical redemption preferences. The mechanism works by using a sorting algorithm, such as a stable sort variant, to rearrange the orders without changing the relative positions of orders with the same priority, thus maintaining queue stability. Stability here means ensuring that orders with equal priority remain in their original order after reordering, avoiding unnecessary reversals that could cause processing delays. In points redemption activities, this mechanism helps to efficiently allocate limited resources, such as gift inventory. Furthermore, the process of rearranging the order based on the resource allocation scheme includes several key steps. The system first extracts the priority values ​​from the resource allocation scheme, for example, marking high-priority orders as requiring earlier processing. Then, a queue reordering mechanism, such as a stable version of insertion sort or merge sort, is applied to integrate the order sequence. For example, in a points redemption scenario, if multiple user orders are competing for the same batch of coupons, the system will move user orders with higher historical redemption preferences to the front of the queue based on the adjusted priorities in the scheme.

[0065] It's important to note that reordering is not a complete restructuring, but rather an incremental adjustment, moving only conflicting parts to reduce computational overhead and maintain the overall stability of the queue. This approach achieves ordered resource allocation in the business logic, avoiding the unfairness caused by random processing.

[0066] Preferably, after obtaining the final processing order, the system prioritizes executing the high-priority requests. In specific implementations, the final processing order is an ordered list, with high-priority orders at the top. During execution, the system starts processing from the head of the queue, for example, directly allocating resources to high-priority orders, such as electronic gift redemption codes. Simultaneously, queue stability is maintained through additional constraints, such as using timestamps as secondary keys to ensure that orders of the same priority are arranged in order of submission time.

[0067] It's important to note that stability is maintained by avoiding cyclic reordering, meaning that sorting is triggered only once when resources are available. This mechanism improves allocation efficiency in the business and ensures that loyal users benefit first.

[0068] In one embodiment, when a user redeems video membership benefits, the system arranges the order sequence based on a scheme, prioritizing high-priority requests while maintaining the queue's integrity through a stable sorting algorithm. The entire process, from integrating the sequences to obtaining the final order, demonstrates the mechanism's versatility and provides a reliable processing path when resources are limited.

[0069] Specifically, maintaining queue stability involves using a priority queue structure combined with stable sorting. Priority queues allow for efficient insertion of high-priority orders, while stable sorting ensures that the original order is preserved.

[0070] For example, during limited-time promotions on a points redemption platform, when the system detects a new order, it only adjusts the sequence locally, avoiding the system burden caused by a global reordering. This implementation supports the smooth operation of the business and ensures that order processing is not interrupted by frequent changes.

[0071] S106. If there are remaining conflicts in the final processing order, supplementary resources are extracted from the preset backup inventory through the backup resource set scheduling mechanism to obtain a complete execution plan, which covers all order processing paths.

[0072] For the final processing order, the remaining conflict locations are identified, and supplementary resources are extracted from the preset reserve inventory. A conflict resolution path is determined by comparing the matching degree between the conflict locations and the inventory resources. Based on the conflict resolution path, resource aggregation management is used to allocate resources to the conflict locations and adjust the execution order, resulting in an intermediate plan with path coverage assurance. According to the intermediate plan, the order processing path is integrated for reserve resource scheduling. If the remaining conflicts exceed a preset threshold, resources are extracted again from the preset reserve inventory to obtain the complete execution plan.

[0073] For example, such as Figure 5 As shown, in one implementation, if there are remaining conflicts in the final processing order, supplementary resources are extracted from the preset reserve inventory through a backup resource set scheduling mechanism to obtain a complete execution plan. Remaining conflicts refer to situations where, after the initial reordering, some orders still cannot be immediately allocated resources, such as in a points redemption activity where high-priority orders have been processed but low-priority orders are stuck due to insufficient inventory. The backup resource set scheduling mechanism is a resource replenishment strategy used to dynamically manage backup inventory and ensure smooth order processing.

[0074] Specifically, the mechanism first detects remaining conflicts by scanning unallocated orders in the final processing order to identify resource gaps. Furthermore, the principle behind the backup resource pool scheduling mechanism is to maintain a pre-set backup inventory pool containing additional prepared resources, such as backup coupons or virtual goods inventory. These resources do not participate in the initial allocation but are used as supplementary resources. On e-commerce points redemption platforms, this mechanism uses scheduling algorithms to match conflicting orders with available resources, for example, allocating resources based on order priority and resource type.

[0075] like Figure 5As shown, the backup resource scheduling mechanism can be configured with various strategies. In addition to coupon compensation, it can also include: 1) a similar product recommendation strategy, which recommends products with similar functions or values ​​from the backup inventory; 2) a points compensation strategy, which returns part of the user's points and provides additional points rewards; 3) a waiting queue strategy, which prompts users to enter the waiting queue and informs them of the estimated restocking time for products that are not limited in quantity but temporarily lack stock.

[0076] For example, in a points-for-limited-time gift redemption campaign, if user orders still conflict due to inventory depletion in the final processing order, the system activates a backup resource allocation mechanism. The process of extracting supplementary resources from the preset backup inventory involves several steps: First, the mechanism queries the inventory pool for matching resource types, such as e-gift codes; then, it calculates the required replenishment amount, determining the extraction scale based on the number of conflicting orders; finally, it allocates these resources to the corresponding order paths. A complete execution plan is thus generated, which is a detailed list of processing paths covering all orders, including processed and high-priority portions, as well as the remaining portions resolved through resource replenishment.

[0077] For example, the plan specifies that order A is obtained from the primary resource and order B is replenished from the spare inventory to ensure that nothing is missed.

[0078] Preferably, in another embodiment, in the scenario of redeeming virtual benefits with membership points, the remaining conflict may stem from multiple users simultaneously applying for the same video membership duration. A backup resource aggregation scheduling mechanism extracts supplementary resources, such as additional membership activation codes, from a preset backup inventory. Specifically, the mechanism assesses the severity of the conflict, filters orders requiring immediate replenishment based on priority thresholds, and then extracts resources from the inventory pool to generate an execution plan. This plan covers all order processing paths; for example, it allocates supplementary resources to medium-priority orders and records paths such as "Order C awaits replenishment before execution." This approach achieves comprehensive resource coverage.

[0079] Specifically, the backup resource aggregation scheduling mechanism uses matching rules to optimize the process when extracting supplementary resources. For example, the mechanism compares order demand with inventory attributes to ensure compatibility, such as matching points values. After the complete execution plan is generated, the coverage path includes the primary path and backup paths, and the system executes order processing sequentially according to these paths.

[0080] In one possible implementation, during a limited-time points redemption event, when the mechanism handles remaining conflicts, it extracts resources from a pre-set reserve inventory, with the plan detailing the resource source and execution sequence for each order.

[0081] For example, in the scenario of redeeming book vouchers with points on an e-commerce platform, if there are remaining conflicts in the final processing order, a backup resource allocation mechanism intervenes. The process of extracting supplementary resources from pre-set backup inventory emphasizes efficiency; for example, the mechanism monitors inventory levels in real time and immediately transfers resources after conflict detection. The resulting complete execution plan covers all order paths, for example, providing backup voucher allocation paths for low-priority orders to ensure fair processing.

[0082] It should be noted that this mechanism supports order management during periods of resource scarcity. By drawing supplementary resources from pre-set reserve inventory, the system generates a comprehensive execution plan, preventing some orders from experiencing long waiting times.

[0083] S107. Based on the complete execution plan, monitor the order processing progress, and through progress tracking and anomaly detection, provide real-time status feedback, iteratively adjust, and obtain a stable order flow.

[0084] Order progress data is obtained from the execution plan. This data is compared with progress tracking data, and if the deviation exceeds a preset threshold, it is identified as an abnormal event, resulting in an anomaly detection result. Real-time status is fed back based on the anomaly detection result. This result is combined with iterative parameter adjustments, and the deviation is corrected through multiple iterations to obtain optimized order flow parameters. The optimized order flow parameters are used to monitor and analyze processing delays. Delay indicators are obtained from the delay analysis, and if the delay indicator is below a preset threshold, the order flow is considered to be operating stably.

[0085] For example, in one implementation, an overall framework for order processing is first established based on a complete execution plan. This execution plan includes all stages from order receipt, inventory check, payment confirmation to logistics and delivery, and the expected time and resource allocation for each step are defined through system modules.

[0086] For example, in e-commerce platforms, a complete execution plan can be preset to a 24-hour timeframe from order placement to shipment, ensuring a benchmark for subsequent monitoring. Furthermore, monitoring order processing progress involves real-time data collection.

[0087] Specifically, a database system is used to record the status changes of each order, such as order ID, current stage, and timestamp. Inventory update data is obtained from the warehouse management system through integrated sensors or API interfaces, thereby tracking whether orders are progressing as planned. This monitoring mechanism can cover various scenarios; for example, during peak order surges, the system automatically adjusts the sampling frequency to maintain accuracy.

[0088] For example, for a retail order, the tracking module compares the actual progress with the execution plan. If the payment confirmation process exceeds a preset 5 minutes, it is marked as a potential delay. This tracking is not limited to single orders but can also aggregate and analyze batch order flows to identify overall bottlenecks. Anomaly detection is a crucial step and its process needs to be explained in detail. Specifically, anomaly detection is based on a rule engine and threshold judgment. First, anomaly indicators are defined, such as a progress deviation rate exceeding 10% or stagnation caused by insufficient inventory. The system compares the current status with the execution plan in real time.

[0089] For example, if the actual delivery time exceeds the expected time by 20%, a detection logic is triggered: the deviation value is calculated as (actual time - expected time) / expected time; if the deviation value is greater than a threshold, it is classified as an anomaly, such as "inventory anomaly" or "payment anomaly." In supply chain order scenarios, for bulk purchase orders, the detection process further analyzes historical data to identify patterns such as seasonal delays. In this way, anomaly detection ensures early detection of problems and supports subsequent feedback.

[0090] For example, when an anomaly occurs, the system immediately reports, "Order 12345 is delayed in the logistics stage, with an estimated impact on delivery time of 2 hours." This feedback is accessible on mobile devices, facilitating real-time response. Iterative adjustments are made based on the feedback results, creating a continuous optimization cycle.

[0091] In one embodiment, the system collects feedback data multiple times to formulate an adjustment strategy: First, it analyzes the root cause of the anomaly, such as determining supplier delays through log review; then, it updates the execution plan, for example, by shortening inventory check time or adding backup suppliers. This iterative process can be repeated multiple times until the order flow stabilizes, for example, after the initial adjustment, the monitoring deviation rate decreases from 15% to 5%. In e-commerce promotional scenarios, for high-frequency orders, adjustments can introduce automated rules, such as dynamically allocating resources to balance the load. This iteration ensures that the order flow gradually transitions from an unstable state to efficient operation. The system can also employ online learning or periodic model retraining. For example, daily, the system fine-tunes the weight coefficients w1-w4 in the Dynamic Priority Scoring Model (DPSM) using a gradient descent algorithm based on the previous day's actual processing data (such as delays in each stage, user complaint rates, etc.), making the scoring results closer to actual operational goals.

[0092] Example 2

[0093] like Figure 6As shown, this invention provides a big data-based optimization system for rolling order processing in a points mall, mainly comprising: a data acquisition and demand calculation module, used to collect order inflow data and user behavior trajectory data from the points mall, including browsing time and redemption frequency, calculate demand change parameters, and obtain peak load levels, which reflect the peak demand of the system's processing capacity; an order classification and queue management module, used to classify order queues based on the peak load levels using a priority sorting algorithm, obtaining a classified order sequence, which divides orders into high-priority requests and regular requests, with high-priority requests involving frequently redeemed points items; and a resource contention detection module, used to scan resource availability if the proportion of high-priority requests in the order sequence exceeds a preset threshold, the module determines conflict points due to insufficient inventory based on available inventory and concurrent access, and obtains a list of conflicting orders. The dynamic priority adjustment module is used to adjust the order processing priority based on the user behavior trajectory data, including historical redemption preferences, to obtain a resource allocation scheme for the conflicting order list. The queue reordering module integrates the order sequence through a queue reordering mechanism, rearranges the order based on the resource allocation scheme, and obtains the final processing order, prioritizing the execution of high-priority requests while maintaining queue stability. The backup resource scheduling module extracts supplementary resources from a preset backup inventory through a backup resource set scheduling mechanism if there are remaining conflicts in the final processing order, obtaining a complete execution plan that covers all order processing paths. The progress monitoring and iterative adjustment module monitors the order processing progress based on the complete execution plan, provides real-time status feedback through progress tracking and anomaly detection, and iteratively adjusts to obtain a stable order flow.

[0094] Example 3

[0095] This embodiment, using the specific scenario of the "Annual Grand Ceremony" large-scale promotional event in the points mall, explains in detail the application process and technical advantages of the present invention.

[0096] During its "Annual Grand Ceremony," a citizen card app's points mall launched two blockbuster products: a limited-edition "City Commemorative Metro Card" (a physical product with only 500 units in stock, high value, and complex processing) and a large supply of "5 Yuan Phone Credit Vouchers" (virtual products, low value, and can be processed instantly). The event started promptly at 8 PM.

[0097] The process of applying the method of this invention is as follows: (1) S101: Load prediction and demand calculation based on the DVP model. One week before the start of the event, the system has begun to collect user behavior trajectory data. It was found that a large number of users repeatedly browsed the "City Commemorative Metro Card" page, with an average browsing time of 120 seconds, far exceeding the 15 seconds for ordinary goods. At the same time, these users generally had a high historical redemption frequency, which was marked by the system as a high-value redemption event. One hour before the start of the event, the system calculated based on the data of the most recent hour using the Demand Variation Parameter (DVP) model: data such as Bi (effective browsing time) and Ej (high-value redemption identifier) ​​were collected. The time decay coefficient λ=0.2, business weight α=0.6, β=0.4 were set. After calculation, DVP = 4.0 was obtained, indicating that the potential redemption demand was extremely strong.

[0098] Subsequently, the system uses the Peak Load Level (PLL) model for prediction: system baseline processing capacity C_base = 1000 orders / second; preset load factor K_load = 1.0; calculated PLL = C_base * (1 + DVP * K_load) = 1000 * (1 + 4.0 * 1.0) = 5000 orders / second.

[0099] Predictions indicate that concurrent requests at the start of the event will reach five times the normal peak, placing the system under extremely high load. Based on this, the system pre-generated a resource allocation plan, dynamically expanding the number of server instances processing orders to the predetermined level.

[0100] (2) S102 & S103: Order classification and resource conflict detection based on CAW and ISM. At 8 p.m., 5,000 orders flooded in instantly.

[0101] S102 Order Classification: Based on the priority sorting algorithm, 800 orders for "City Commemorative Metro Cards" were initially marked as "high-priority requests" because they were identified as "high-frequency redemption" items during the pre-sale period; the remaining 4,200 orders for phone bill vouchers, etc., were marked as "regular requests".

[0102] S103 Resource Conflict Detection: The proportion of high-priority requests (800) is 16% (800 / 5000), which exceeds the system's preset resource tension threshold of 10%, and the resource contention detection module is triggered.

[0103] The resource contention detection module defines a concurrent access window (CAW) Δt = 2 seconds and introduces an inventory safety margin (ISM) S_factor = 0.9.

[0104] Within the first 2 seconds of CAW after the event started, the resource contention detection module counted N_req = 800 requests for the "Commemorative Metro Card".

[0105] The resource contention detection module uses the conflict trigger condition for judgment: N_req>(Current_Stock * S_factor). Substituting the value: 800>(500 * 0.9), that is, 800>450.

[0106] The conflict condition was met. The system determined that there was a serious inventory conflict and generated a "conflicting order list" containing these 800 orders.

[0107] (3) S104: Dynamic Priority Adjustment Based on the DPSM Model. The system uses a Dynamic Priority Scoring Model (DPSM) to precisely reorder these 800 conflicting orders. The model weights are obtained from historical data training and are set as follows: w1 (preference) = 0.4, w2 (level) = 0.3, w3 (waiting) = 0.1, w4 (scarcity) = 0.2. Two typical users are used as examples:

[0108] User A: V5 member, with a very high historical preference for redeeming transportation cards.

[0109] User B: V2 member, first attempt to redeem this type of item.

[0110] The system calculates the feature factors for the orders of the two users:

[0111] Historical exchange preference factor (F_pref):

[0112] User A: F_pref = 0.85 (has redeemed a transportation card 5 times, accounting for 85% of its total redemptions);

[0113] User B: F_pref = 0.05 (This type of product accounts for 5% of historical preferences);

[0114] User level factor (F_level):

[0115] User A (V5): F_level = 1.0;

[0116] User B (V2): F_level = 0.4;

[0117] Order wait time factor (F_wait): Assume user B's request arrives 1 second earlier.

[0118] User A (waiting 2 seconds): F_wait=log 10 (2+1)≈ 0.48;

[0119] User B (waiting 3 seconds): F_wait=log 10 (3+1)≈ 0.60;

[0120] Product scarcity factor (F_scarce):

[0121] The value is the same for both orders: F_scarce = 800 / 500 = 1.6;

[0122] Final priority score P_score calculation:

[0123] P_score_A= 0.40.85 + 0.31.0 + 0.10.48 + 0.21.6 = 0.34 + 0.3 + 0.048 +0.32 =1.008;

[0124] P_score_B= 0.40.05 + 0.30.4 + 0.10.60 + 0.21.6 = 0.02 + 0.12 + 0.06 +0.32 =0.520;

[0125] The calculation results showed that although user B's order was submitted slightly earlier, user A, with high loyalty and high preference, received a significantly higher priority score. Based on this, the system generated an accurate priority sequence reflecting the users' true redemption intentions for all 800 conflicting orders and formulated a resource allocation plan.

[0126] (4) S105 & S106: Stable Queue Reordering and Backup Resource Scheduling; S105 Queue Reordering: The queue reordering module uses a stable sorting algorithm (such as merge sort) to integrate the entire order queue based on the new priority score calculated by DPSM. The final processing order is determined as follows: the 500 highest-scoring "Commemorative Metro Card" orders; 4200 "Phone Recharge Voucher" orders (since they are regular requests, they are placed later overall, and can be sorted internally by submission time); and the remaining 300 "Commemorative Metro Card" orders. This sorting ensures that requests from high-value, high-willing users are prioritized, while not affecting virtual orders that can be processed quickly.

[0127] S106 Backup Resource Scheduling: For the 300 orders that are ranked lower and fail to obtain the "commemorative metro card," the system automatically triggers the backup resource aggregation scheduling mechanism. From the preset backup inventory, a "10% off metro card voucher" is automatically matched and pushed to these users as compensation. This forms a complete execution plan covering all paths, ensuring that no user's request is directly rejected.

[0128] (5) S107: Monitoring and closed-loop iteration; the system monitors the order processing progress in real time. It was found that due to a brief delay in the logistics system interface, the average outbound time of physical card orders was 15 seconds slower than planned.

[0129] Anomaly detection: The system marks this event as a "moderate delay anomaly".

[0130] Feedback and Iteration: Abnormal data is fed back to the iterative adjustment module. The system fine-tunes the weights in the DPSM model using online learning algorithms (such as gradient descent), slightly reducing the weight of physical goods requiring complex logistics in subsequent scoring. Simultaneously, the load prediction model PLL learns that the actual processing capacity during that period is slightly lower than expected, providing more accurate predictions for similar events in the future. This closed-loop iteration enables the system to adaptively optimize and maintain high efficiency and stability in the long term.

[0131] Compared to the traditional first-come, first-served (FIFO) mechanism, the improvement of this invention lies in the following: Under the FIFO mechanism, the first 500 users who click to redeem (who may be users who clicked by scripts or by chance) will receive a commemorative card. A large number of highly loyal and willing real users may miss out due to network latency of a fraction of a second, resulting in a poor user experience. In contrast, this invention, through intelligent prediction and dynamic adjustment, accurately allocates limited resources to the user group most likely to generate high value and high satisfaction, and appeases other users through a backup resource mechanism, thereby improving system processing efficiency, resource allocation fairness, and user satisfaction.

[0132] Figure 7 This is a comparison chart showing the effectiveness of the method of this invention and existing technologies in handling scarce goods during peak periods. The chart includes three sub-charts, which visually demonstrate the superior technical effects of the present invention from three dimensions: order processing time, resource utilization, and system stability.

[0133] The top-left sub-chart compares order processing times. The horizontal axis of the bar chart represents different system load levels, and the vertical axis represents the average order processing time (in seconds). As can be seen from the chart, under low load, the processing time of the proposed method is similar to that of the traditional FIFO method; however, as the load level increases, the processing time of the traditional FIFO method increases exponentially, reaching 8.5 seconds at 1.0 times the load and a staggering 25.0 seconds at 2.0 times the load. In contrast, the processing time of the proposed method increases very gradually, maintaining a level of 12.8 seconds even at 2.0 times the load. This demonstrates that the proposed method, through intelligent scheduling and sorting, can effectively avoid processing bottlenecks and shorten user waiting time in high-concurrency scenarios.

[0134] The top right sub-chart is a resource utilization comparison chart. The horizontal axis of this bar chart also represents the system load level, and the vertical axis represents the resource utilization rate (%). It can be seen that at all load levels, the resource utilization rate of the method of this invention is significantly higher than that of the traditional FIFO method. Especially in the 1.0 to 2.0 times load range, the resource utilization rate of the method of this invention can be maintained at a high level of 75%-82%, while the utilization rate of the traditional FIFO method has dropped to 45% due to processing congestion and resource mismatch. This indicates that the present invention can utilize system resources such as servers and inventory more efficiently.

[0135] The bottom right sub-chart compares user satisfaction levels, using a bar chart to contrast the satisfaction of different user groups when handling scarce goods during peak periods using the method of this invention versus existing technologies. Under varying system loads, the user satisfaction rate of the method in this application remained at a high level, increasing from 68% to 85% under 1.0 times the peak load. This demonstrates that the method of this invention significantly improves the satisfaction of highly loyal users and the overall user base.

[0136] The bottom right sub-chart shows the system stability index. The horizontal axis of this line graph represents the change from normal load to peak load (5000 orders / second), and the vertical axis represents the system failure rate (%). Three regions are defined in the graph: stable operating region (<10%), warning region (10-20%), and danger region (>20%). The failure rate curve of the traditional FIFO method rises sharply with increasing load, entering the warning region at peak load and reaching as high as 35% at peak load, entering the danger region, where the system is on the verge of collapse. In contrast, the failure rate curve of the method in this invention is very flat; even under peak load, the failure rate can be controlled at 18%, always remaining in the stable operating region. This invention improves the stability and robustness of the system through forward-looking load prediction and dynamic resource conflict management.

[0137] In summary, compared with the prior art, this application has the following beneficial effects:

[0138] 1. This invention goes beyond simply processing orders after they occur. Instead, it achieves quantitative prediction of future system load by constructing a demand change parameter model based on deep user behavior. By linking user browsing and other behaviors with backend processing capacity, and combining this with a dynamic trigger condition—the percentage of high-priority requests exceeding a threshold—a closed-loop mechanism from demand prediction to conflict warning is built. This enables the system to shift from passively accepting requests to proactively managing risks, effectively preventing system crashes or transaction failures due to resource constraints. This is something that existing technologies, such as simple order sorting or inventory checks, cannot achieve.

[0139] 2. In the face of resource conflicts, this invention does not adopt a simple first-come, first-served or fixed membership level priority strategy. Instead, it uses a multi-dimensional dynamic priority scoring model to dynamically integrate and quantify multiple dimensions such as users' historical preferences, real-time willingness (waiting time), identity value, and the intensity of resource competition. This allows limited and scarce resources to be accurately allocated to requests that best reflect commercial value and user loyalty, achieving a shift from equality of opportunity to equality of value, and greatly improving user satisfaction and operational efficiency.

[0140] 3. This invention not only resolves identified conflicts but also provides compensation paths for unfulfilled requests through a backup resource set scheduling mechanism, ensuring the integrity of the business process. More importantly, through the progress monitoring and iterative adjustment module, the system uses each processing result (such as delays or anomalies) as feedback to continuously optimize the front-end load prediction model and the mid-end priority scoring weights. This closed-loop adaptive capability throughout the entire process enables the system to continuously learn and evolve, maintaining high efficiency and stable operation in various complex scenarios over the long term. The above descriptions are merely embodiments of this invention and do not limit the patent scope of this invention. Any equivalent structural or process transformations made using the content of this specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this invention.

Claims

1. A method for optimizing the rolling processing of orders in a points mall based on big data, characterized in that, The method includes: By collecting order inflow data and user behavior trajectory data from the points mall, demand variation parameters are calculated. Combined with historical load data and preset load coefficients, peak load levels are determined, reflecting the peak demand on the system's processing capacity. Specifically, this includes: By collecting order inflow data and user behavior trajectory data from the points mall, including browsing time records and redemption frequency statistics, the average duration value is extracted from the browsing time records and combined with the frequency count in the redemption frequency statistics to calculate the demand change parameter. The demand change parameter is obtained by dividing the average duration value by the frequency count. Based on the demand change parameters, the peak load level is determined by comparing the demand change parameters with historical load data. For the peak load level, if the demand change parameter exceeds a preset threshold, the order volume statistics in the order inflow data are combined with the peak load level to obtain a resource pre-allocation plan. This plan optimizes the response to peak demand by weighting the peak load level with the order volume statistics. Based on the peak load level, a priority sorting algorithm is used to classify the order queue, resulting in an order sequence that divides orders into high-priority requests and regular requests. When the proportion of high-priority requests in the order sequence exceeds a preset threshold, the number of requests within the concurrent access window is compared with the real-time inventory level multiplied by the inventory safety margin coefficient to identify conflict points related to insufficient inventory, resulting in a conflict order list, specifically including: When the proportion of high-priority requests exceeds a preset threshold, a Concurrent Access Window (CAW) is defined with a time length of Δt. The number of all redemption requests for the same high-priority product within any given CAW is counted in real time. An inventory safety margin is also defined. The ISM (Independent Streaming Management System) is a coefficient S_factor less than 1, representing the proportion of inventory reserved to cope with emergencies. The trigger condition for a conflict is defined as follows: a resource conflict is determined only when the instantaneous request volume within a concurrent window exceeds the safe capacity of the inventory, and all orders with the corresponding requests are added to the conflict order list. For the conflict order list, a preset weighted scoring model is used, combined with historical redemption preference features from the user behavior trajectory data, to adjust the order processing priority and obtain a resource allocation scheme. The order sequence is integrated through a queue reordering mechanism, and the order is rearranged based on the resource allocation scheme to obtain the final processing order. If there are remaining conflicts in the final processing order, supplementary resources are extracted from the preset backup inventory through a backup resource set scheduling mechanism to obtain a complete execution plan covering all order processing paths. Based on the complete execution plan, the order processing progress is monitored, and real-time status is fed back through progress tracking and anomaly detection. Iterative adjustments are made to obtain a stable order flow.

2. The method for optimizing the rolling processing of orders in a points mall based on big data, as described in claim 1, is characterized in that... The calculation of the demand change parameter includes: based on the time decay weight, performing weighted statistics on the effective browsing time and high-value redemption events within a preset time window to obtain a quantitative value reflecting the user's attention investment.

3. The method for optimizing the rolling processing of orders in a points mall based on big data, as described in claim 1, is characterized in that... Based on the peak load level, a priority sorting algorithm is used to classify the order queue, resulting in a classified order sequence. This order sequence divides orders into high-priority requests and regular requests. The high-priority requests involve frequently redeemed points-based items, including: Based on the peak load level assessment, high-frequency redemption identification features are extracted from the order queue using behavioral trajectory fusion. The extraction of high-frequency redemption identification features is obtained by statistically counting redemption frequencies and combining user request hierarchies to obtain a preliminary priority list. The preliminary priority list marks frequently redeemed points items as high-priority requests. Based on the preliminary priority list, peak demand response data is obtained and combined with the priority sorting mechanism to perform layered processing on the order queue. The peak demand response data is derived from the peak load level assessment to determine the classified order sequence, which distinguishes between high-priority requests and regular requests. From the categorized order sequence, if the proportion of high-priority requests exceeds a preset threshold, the division of points-based products is adjusted through priority resource allocation. The adjustment of points-based product division adopts sequence generation logic to fuse behavioral trajectories and obtains an optimized sequence response scheme.

4. The method for optimizing the rolling processing of orders in a points mall based on big data, as described in claim 1, is characterized in that... The weighted scoring model also incorporates at least one of the following features: user level, order waiting time, or product scarcity.

5. The method for optimizing the rolling processing of orders in a points mall based on big data, as described in claim 1, is characterized in that... The order sequence is integrated through a queue reordering mechanism, and the order is rearranged based on the resource allocation scheme to obtain the final processing order, including: The order sequence is obtained, priority indicators are extracted from the resource allocation scheme, and the bubble sort algorithm is used to compare, exchange and integrate the order sequence pairwise to obtain a preliminary reordered sequence. For the initial reordering sequence, the queue stability attributes extracted from the initial reordering sequence are integrated to determine if high-priority requests exceed a preset threshold. In this case, the execution position of the high-priority requests is adjusted to determine the intermediate processing order. By combining the intermediate processing order with the sequence integration attributes extracted from the intermediate processing order, low-priority requests are rearranged to obtain the final processing order.

6. The method for optimizing the rolling processing of orders in a points mall based on big data, as described in claim 1, is characterized in that... If there are remaining conflicts in the final processing order, supplementary resources are extracted from the preset backup inventory through the backup resource set scheduling mechanism to obtain a complete execution plan covering all order processing paths, including: For the final processing order, the remaining conflict locations are obtained, supplementary resources are extracted from the preset reserve inventory, and the conflict resolution path is determined by comparing the matching degree between the conflict locations and the inventory resources. Using the conflict resolution path, resource aggregation management is employed to allocate resources to conflict locations and adjust the execution order, resulting in an intermediate plan under path coverage assurance. Based on the intermediate plan, the order processing path is integrated to schedule backup resources. If the remaining conflicts exceed a preset threshold, resources are extracted again from the preset backup inventory to obtain a complete execution plan.

7. A big data-based points mall order rolling processing optimization system, characterized in that, The system includes a data acquisition and demand calculation module, used to collect order inflow data from the points mall and user behavior trajectory data, calculate demand change parameters, and determine peak load levels by combining historical load data and preset load coefficients. The peak load levels reflect the peak demand of the system's processing capacity, specifically including: By collecting order inflow data and user behavior trajectory data from the points mall, including browsing time records and redemption frequency statistics, the average duration value is extracted from the browsing time records and combined with the frequency count in the redemption frequency statistics to calculate the demand change parameter. The demand change parameter is obtained by dividing the average duration value by the frequency count. Based on the demand change parameters, the peak load level is determined by comparing the demand change parameters with historical load data. For the peak load level, if the demand change parameter exceeds a preset threshold, the order volume statistics in the order inflow data are integrated with the peak load level to obtain a resource pre-allocation plan. This plan optimizes the response to peak demand by weighting the peak load level with the order volume statistics. An order classification and queue management module is used to classify order queues based on the peak load level using a priority sorting algorithm, resulting in an order sequence that divides orders into high-priority requests and regular requests. A resource contention detection module is used to identify conflict points of insufficient inventory when the proportion of high-priority requests in the order sequence exceeds a preset threshold. This is achieved by comparing the number of requests within the concurrent access window with the real-time inventory quantity multiplied by the inventory safety margin coefficient, resulting in a conflict order list, specifically including: When the proportion of high-priority requests exceeds a preset threshold, a Concurrent Access Window (CAW) with a duration of Δt is defined to count the number of redemption requests for the same high-priority product within any CAW in real time. An Inventory Safety Margin (ISM) is defined, with a value of a coefficient S_factor less than 1, representing the proportion of inventory reserved to cope with unforeseen circumstances. The conflict trigger condition is defined as a resource conflict only occurring when the instantaneous number of requests within a concurrent window exceeds the inventory's safety capacity, and all orders with the corresponding requests are added to the conflict order list. A dynamic priority adjustment module uses a preset weighted scoring model, combined with historical redemption preference features from user behavior trajectory data, to adjust the order processing priority for the conflict order list, resulting in a resource allocation scheme. A queue reordering module integrates the order sequence through a queue reordering mechanism, rearranging the order based on the resource allocation scheme to obtain the final processing order. A backup resource scheduling module, if there are remaining conflicts in the final processing order, extracts supplementary resources from a preset backup inventory through a backup resource set scheduling mechanism to obtain a complete execution plan covering all order processing paths. The progress monitoring and iterative adjustment module is used to monitor the order processing progress based on the complete execution plan, and to provide real-time status feedback through progress tracking and anomaly detection, and to iteratively adjust to obtain a stable order flow.

Images