Advertisement budget intelligent allocation method and system based on dynamic value score

Abstract

This invention belongs to the technical field of intelligent advertising budget allocation, and relates to a method and system for intelligent advertising budget allocation based on dynamic value scoring. The method includes: acquiring raw data and local in-transit adjustment data of advertising campaigns, merging them to generate a budget status snapshot; parsing the snapshot to calculate multi-dimensional features such as conversion cost efficiency and budget demand intensity, generating dynamic value scores and budget utilization rate indicators, and strategically stratifying advertising campaigns accordingly; combining scores, utilization rates, and stratification to perform supply and demand matching and construct budget transfer requests; performing physical constraint verification, generating controlled instructions, and updating logs; when high-value groups face budget constraints, retrieving funds from an adaptive redundancy pool to supplement them; summarizing all approved instructions and sending them to the advertising platform, and completing state loop cleanup through asynchronous callback information. This invention solves the problems of inaccurate and unstable budget allocation caused by decision data delays and a single evaluation dimension.

Description

Technical Field

[0001] This invention belongs to the technical field of intelligent advertising budget allocation, and relates to a method and system for intelligent advertising budget allocation based on dynamic value scoring. Background Technology

[0002] In programmatic advertising, advertisers typically need to allocate a limited daily budget across multiple advertising campaigns to achieve predetermined business goals. Campaign performance is dynamic, and there is often a communication delay between data feedback from the advertising platform and the effective implementation of budget adjustment instructions. This delay can affect the accuracy of budget allocation decisions.

[0003] Existing technologies mainly include manual monitoring and adjustments, and rule-based automated scripts. Manual adjustments rely on operations staff analyzing reports based on experience and manually increasing or decreasing the budget. Rule-based automated scripts, on the other hand, trigger automatic budget adjustments based on preset conditions, such as when the conversion cost of an activity exceeds a threshold or the budget is consumed rapidly. These methods improve management efficiency to some extent, as they can execute predefined, simple logic based on a single metric.

[0004] However, the above methods have some limitations when dealing with complex campaign scenarios. First, if the decision-making system relies solely on real-time data returned by the advertising platform, it may not fully consider budget adjustment instructions that have been issued but not yet taken effect. In high-frequency budget allocation operations, this could lead to duplicate deductions from the budget for the same advertising campaign. Second, rule systems based on a single performance metric may not provide a comprehensive perspective when evaluating the value of advertising campaigns. For example, they might classify campaigns that are in the learning phase, temporarily lacking conversions but with high traffic potential, as inefficient. Furthermore, when multiple high-performing campaigns face budget shortages, conventional budget transfer mechanisms lack the flexibility to supplement them with external reserve funds. Summary of the Invention

[0005] In a first aspect, the present invention provides a method for intelligent allocation of advertising budget based on dynamic value scoring, comprising the following steps: S1. Obtain the original performance data of each advertising campaign in the target advertising platform, synchronously extract the in-transit adjustment data recorded locally, and generate a budget status snapshot that eliminates the interference of effective delay through logical merging. S2. Analyze the budget status snapshot, calculate the conversion cost efficiency characteristics, budget demand intensity characteristics, and traffic exposure potential characteristics of each advertising campaign, and generate dynamic value scores and budget utilization rate indicators based on the comprehensive weighting of these characteristics. S3. Based on dynamic value scores and actual conversion costs, stratify the execution strategy of advertising campaigns and determine the level identifiers of advertising campaigns. S4. Combine budget utilization rate indicators, dynamic value scores, and advertising campaign level identifiers to perform supply and demand matching calculations to construct a budget transfer request that includes demand source information, power supply information, and transferable amount. S5. Input the budget transfer request into the delay-aware budget recovery arbitration module, perform physical constraint verification, generate controlled unconfirmed budget adjustment instructions, and update the arbitration log; S6. Monitor the budget utilization rate of high-value groups. When it reaches the preset conditions, call the adaptive redundancy pool to generate an emergency budget instruction and append it to the arbitration log. S7. Summarize all approved budget adjustment instructions and emergency budget instructions and send them to the target advertising platform. Then, perform a closed-loop cleanup of the arbitration log status by listening to the callback information returned by the asynchronous listening interface.

[0006] A further aspect of the present invention, step S1, includes the following steps: Call the target advertising platform's application programming interface to obtain the current budget and current cumulative spending for each advertising campaign; Read all issued but not yet confirmed effective budget adjustment instructions in the local arbitration log, and calculate the increase and decrease in budget in transit for each advertising campaign. Obtain the cumulative historical deviation variable for the corresponding advertising campaign; The current budget value, the increase in budget in transit, and the cumulative historical deviation variable are added together, and the decrease in budget in transit is subtracted to generate a snapshot budget value for each advertising campaign; The snapshot budget value is associated with and encapsulated with the current cumulative spending value to generate a budget status snapshot.

[0007] A further aspect of the present invention, step S2, includes the following steps: Calculate the budget utilization rate index based on the snapshot budget value and the current cumulative spending value, which reflects the intensity of budget demand. Based on the current cumulative spending and number of converted orders, calculate the conversion cost efficiency characteristics, and based on the actual click volume, calculate the smoothed click volume of the traffic exposure potential characteristics. The system checks whether each advertising campaign generates conversion orders. If so, it multiplies the conversion cost efficiency feature, budget utilization rate, and smoothed click volume by a preset first weighting coefficient and sums them to generate a dynamic value score. If not, it multiplies the budget utilization rate and smoothed click volume by a preset second weighting coefficient and sums them to generate a dynamic value score.

[0008] A further aspect of the present invention, step S3, includes the following steps: Read the pre-configured lower and upper limits of the conversion cost threshold; Advertising campaigns with no converted orders or actual conversion costs exceeding the conversion cost threshold are filtered out and assigned a tiered label to the elimination group. Advertising campaigns that generate conversion orders and whose actual conversion cost is not higher than the lower limit of the conversion cost threshold are selected and assigned a high-value group tier label. The remaining advertising campaigns that have generated conversion orders and whose actual conversion costs are between the lower and upper limits of the conversion cost threshold are assigned a low-value group tier label.

[0009] A further aspect of the present invention, step S4, includes the following steps: Ad campaigns whose budget utilization rate exceeds the preset demand-side threshold lower limit are identified as demand-side campaigns and arranged in descending order according to the priority of the ad campaign hierarchy to form a priority supplement sequence. All potential source advertising campaigns are arranged between groups according to the reverse priority from the elimination group to the high-value group, and dynamic value scores are used to sort them in reverse order from low to high within the group to determine the forced extraction priority within the group. Based on the priority order of replenishment sequence and mandatory extraction priority within the group, the transferable amount of the source activity is calculated to generate a budget transfer request that includes demand source information, power supply information, and transferable amount.

[0010] A further aspect of the present invention, step S5, includes the following steps: Obtain the snapshot budget value and current cumulative spending value of the source activity from the budget status snapshot, and subtract the transferable amount to be transferred in the budget transfer application to obtain the actual available remaining space; The built-in hierarchical flow rule library is invoked to conduct a compliance review of this budget transfer operation. The rule library includes rules that prohibit downward bypassing of budget collection. When the available remaining space is greater than the minimum budget retention protection threshold pre-allocated at each level and passes the level flow rule verification, a controlled unconfirmed budget adjustment instruction is generated and written to the arbitration log with the status details of extracting deduction and injecting gain.

[0011] A further aspect of the present invention, step S6, includes the following steps: The snapshot budget values ​​obtained in step S1 for each advertising campaign within the high-value group are combined with the amount of the newly generated unconfirmed budget increase instruction for the advertising campaign in step S5 to calculate and evaluate the expected final budget and expected final utilization rate of each advertising campaign within the high-value group. Advertising campaigns that are expected to have a final usage rate that is still above the preset safety alert line and have the highest historical score are identified as unmet targets. From the independently established adaptive redundancy pool, compensation funds are allocated to unmet objects, and the compensation funds are encapsulated into emergency budget instructions that only contain the addition action and do not specify the source of recovery. The emergency budget instructions are then submitted for timing verification.

[0012] A further aspect of the present invention, step S7, includes the following steps: Summarize all approved but unconfirmed budget adjustment instructions and emergency budget instructions, encapsulate them into a batch communication request message, and send it to the target advertising platform; The asynchronous listening component is used to detect the execution result response of the target advertising platform and to parse and extract the confirmation status ticket containing whether the platform executed successfully and the actual time of processing. For the instruction content corresponding to the confirmation status ticket that has obtained a positive success signature, update its status in the arbitration log to confirmed, thereby releasing the computational occupation at the logical level; In response to unknown state command requests that cause deadlock due to network timeout or communication anomalies, the data tracing conservative locking link is activated, and the cumulative historical deviation variable corresponding to the advertising campaign is positively updated according to the amount attached to the abnormal command request, so as to form a placeholder lock to prevent duplicate distribution or deduction.

[0013] A further aspect of the present invention includes the following method for calculating the transferable amount: obtaining the baseline original budget of the source activity and multiplying it by a preset ratio to obtain the maximum allowable recovery amount; obtaining the current available remaining budget of the source activity, wherein the current available remaining budget is the snapshot budget value minus the current cumulative expenditure value; and taking the target value between the maximum allowable recovery amount and the current available remaining budget as the transferable amount.

[0014] Secondly, the present invention provides an intelligent advertising budget allocation system based on dynamic value scoring, comprising the following modules: The budget status snapshot generation module is used to obtain the original performance data of each advertising campaign in the target advertising platform, synchronously extract the in-transit adjustment data recorded locally, and generate a budget status snapshot that eliminates the interference of effective delay through logical merging. The dynamic value scoring module is used to analyze the budget status snapshot, calculate the conversion cost efficiency characteristics, budget demand intensity characteristics, and traffic exposure potential characteristics of each advertising campaign, and generate dynamic value scores and budget utilization rate indicators based on the comprehensive weighting of these characteristics. The strategy stratification module, based on dynamic value scores and actual conversion costs, implements strategy stratification for advertising campaigns and determines the stratification identifiers for advertising campaigns; The supply and demand matching calculation module is used to combine budget utilization rate indicators, dynamic value scores and advertising campaign level identifiers to perform supply and demand matching calculations in order to construct a budget transfer request that includes demand source information, power supply information and transferable amount. The budget recovery arbitration module is used to input budget transfer requests into the delay-aware budget recovery arbitration module, perform physical constraint verification, generate controlled unconfirmed budget adjustment instructions, and update the arbitration log; The adaptive redundancy pool compensation module is used to monitor the budget utilization rate of high-value groups. When it reaches the preset conditions, it calls the adaptive redundancy pool to generate an emergency budget instruction and appends it to the arbitration log. The execution and status loop module is used to aggregate all approved budget adjustment instructions and emergency budget instructions and send them to the target advertising platform. It also performs status loop cleaning of the arbitration log by listening to the callback information returned by the asynchronous listening interface.

[0015] In summary, the present invention has the following beneficial technical effects: 1. By creating a budget status snapshot, real-time data obtained from the advertising platform is logically merged with locally recorded budget adjustment instructions in progress before budget allocation decisions are made. This technique allows the data used for decision-making to pre-calculate budget changes that have not yet taken effect, thus providing a more accurate view of the future budget status. This helps reduce the risk of data lag caused by platform communication delays, thereby reducing erroneous deductions from advertising campaign budgets and improving the accuracy of budget allocation and the stability of system operation.

[0016] 2. This invention employs a multi-dimensional dynamic value scoring model that comprehensively considers the conversion cost efficiency, budget demand intensity, and traffic exposure potential of advertising campaigns. This model can also adaptively adjust the weights of each evaluation dimension based on whether the advertising campaign generates conversions. Compared to evaluation methods relying on a single indicator, this comprehensive evaluation mechanism can more fully characterize the advertising campaign's potential, allowing the budget to be more rationally allocated to campaigns with higher overall value, thereby helping to improve the overall efficiency of budget utilization.

[0017] 3. This invention incorporates a latency-aware budget recovery arbitration module, providing a control mechanism for concurrent budget allocation. Before approving a budget deduction, this module verifies all issued but unconfirmed deduction requests for the same source activity to ensure that the source's available budget is sufficient to support the current and all en route deduction operations. This mechanism helps avoid excessive budget recovery from source activities in concurrent operation scenarios, ensuring the orderliness and reliability of budget flow in high-frequency scheduling environments.

[0018] 4. By deploying an adaptive redundancy pool and a closed-loop state management mechanism, this invention enhances the system's resource allocation flexibility and fault tolerance. When internal budget flow is insufficient to meet the needs of high-value activities, the system can draw funds from an independent redundancy pool for supplementation. Simultaneously, it confirms the final execution status of instructions by asynchronously listening to platform callbacks and traces and conservatively locks the execution data of instructions with unknown states that encounter network timeouts. This design effectively blocks the double-deduction illusion and concurrent retry storms induced by communication anomalies, maintaining not only the dynamic consistency between internal system data and the platform's actual state but also providing robust self-healing protection for the system in the face of extreme network fluctuations. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. The drawings are used to provide a further understanding of the present invention.

[0020] Figure 1 A flowchart illustrating an embodiment of this application is disclosed.

[0021] Figure 2 Structural schematic diagrams of embodiments of this application are disclosed. Detailed Implementation

[0022] The following is in conjunction with the appendix Figure 1 - Figure 2 A preferred description of the present invention is provided below.

[0023] See attached document Figure 1 This invention proposes an intelligent advertising budget allocation method based on dynamic value scoring, comprising the following steps: S1. Obtain the original performance data of each advertising campaign in the target advertising platform, synchronously extract the in-transit adjustment data recorded locally, and generate a budget status snapshot that eliminates the interference of effective delay through logical merging. S2. Analyze the budget status snapshot, calculate the conversion cost efficiency characteristics, budget demand intensity characteristics, and traffic exposure potential characteristics of each advertising campaign, and generate dynamic value scores and budget utilization rate indicators based on the comprehensive weighting of these characteristics. S3. Based on dynamic value scores and actual conversion costs, stratify the execution strategy of advertising campaigns and determine the level identifiers of advertising campaigns. S4. Combine budget utilization rate indicators, dynamic value scores, and advertising campaign level identifiers to perform supply and demand matching calculations to construct a budget transfer request that includes demand source information, power supply information, and transferable amount. S5. Input the budget transfer request into the delay-aware budget recovery arbitration module, perform physical constraint verification, generate controlled unconfirmed budget adjustment instructions, and update the arbitration log; S6. Monitor the budget utilization rate of high-value groups. When it reaches the preset conditions, call the adaptive redundancy pool to generate an emergency budget instruction and append it to the arbitration log. S7. Summarize all approved budget adjustment instructions and emergency budget instructions and send them to the target advertising platform. Then, perform a closed-loop cleanup of the arbitration log status by listening to the callback information returned by the asynchronous listening interface.

[0024] In one embodiment of the present invention, step S1 includes the following steps: Call the target advertising platform's application programming interface to obtain the current budget and current cumulative spending for each advertising campaign; Read all issued but not yet confirmed effective budget adjustment instructions in the local arbitration log, and calculate the increase and decrease in budget in transit for each advertising campaign. Obtain the cumulative historical deviation variable for the corresponding advertising campaign; The current budget value, the increase in budget in transit, and the cumulative historical deviation variable are added together, and the decrease in budget in transit is subtracted to generate a snapshot budget value for each advertising campaign; The snapshot budget value is associated with and encapsulated with the current cumulative spending value to generate a budget status snapshot.

[0025] Specifically, the budget control server automatically triggers the budget status assessment process within a preset scheduling cycle. At the start of the process, the budget control server first initiates a predefined API call request to the target advertising platform's application programming interface (API) server via its network interface module, using the HTTPS protocol carrying a valid authentication token. It should be noted that the target advertising platform refers to a third-party online service system that provides programmatic advertising services, such as Amazon Ads or Google Ads, which provides an API for external systems to query data and control budgets.

[0026] Through this API call, the server can obtain real-time raw performance data for all active advertising campaigns under the current account. In this embodiment, the raw performance data is real-time status data directly obtained from the target advertising platform without local logic processing. The application programming interface (API) server returns a data set containing multiple data objects. The budget control server parses this data set, extracts a unique campaign identifier, the current budget value reflected on the platform, and the current cumulative spending value accumulated since midnight of the current day for each advertising campaign, and temporarily stores this data in the server's memory data structure.

[0027] Meanwhile, the budget control server accesses a locally deployed persistent database to query the arbitration log. It should be understood that the arbitration log is a local data table used to record the status of budget adjustment instructions. Its key fields include the activity identifier, adjustment type, adjustment amount, and status flag. A status flag of "unconfirmed" indicates that the corresponding adjustment instruction has been arbitrated by this system and sent to the platform API, but a successful execution confirmation message has not yet been received from the platform. This status setting is based on the common practice of asynchronous communication between distributed systems in this field. The arbitration log records all budget adjustment operations that have been submitted to the target advertising platform but have not yet received a successful callback confirmation from the platform.

[0028] Specifically, the server executes a database query command, filters out all log entries with a status field of "unconfirmed," and groups these entries according to the activity identifier. Within each group, the amounts of commands with adjustment types of "increase" and "decrease" are summed to calculate the corresponding in-transit budget increase and decrease for each advertising campaign. Finally, the budget control server iterates through each advertising campaign that has obtained raw performance data from the platform, extracts the cumulative historical deviation variable for that campaign from its local records, performs a mathematical addition operation on its current budget value, the queried in-transit budget increase, and the cumulative historical deviation variable, and then subtracts the in-transit budget decrease to calculate the snapshot budget value after logical merging and correction. In this embodiment, to prevent the system from being polluted by the infinite accumulation of data due to occasional anomalies from external platforms, the system immediately resets the cumulative historical deviation variable of the corresponding advertising campaign in the database to zero after completing the snapshot budget value calculation.

[0029] The calculation process for the snapshot budget value can be represented by the following formula:

[0030] In the formula, It serves as a unique identifier for the advertising campaign and is automatically assigned by the system by reading the platform's data stream. This is a snapshot budget value, representing the logical true budget after eliminating platform latency. It is calculated in real time using this formula. This is the current budget value, the raw budget value obtained directly from the target advertising platform API. This is the increase in the budget in transit, calculated by summing the amounts of instructions in the local arbitration log that are in an unconfirmed state and have an action type of "increase". It is used to compensate for injected funds that have not yet taken effect on the platform. This is a reduction in the budget in transit. It is calculated by summing the amounts of instructions in the local arbitration log that are in an unconfirmed state and whose action type is reduction. It is used to deduct funds that have been transferred out but not yet deducted by the platform to prevent redundant and duplicate deductions that could lead to overselling. The cumulative historical deviation variable corresponding to the advertising campaign is used to take over the suspended amount due to network timeout or other unclear status anomalies in the previous scheduling cycle. It can be generated through the data tracing conservative locking link in step S7 to positively lock the questionable amount and prevent duplicate payment or deduction for the same advertising campaign due to misjudged instruction failure.

[0031] This snapshot budget value, along with the original campaign identifier and the current cumulative spending value, constitutes a record. The collection of records for all advertising campaigns forms a budget status snapshot that eliminates the interference of activation delays. This budget status snapshot is a structured dataset that provides an isolated and consistent data view for all subsequent calculation steps within the current cycle, thus avoiding data inconsistencies caused by platform API activation delays. This snapshot serves as the unique and definitive data basis for all subsequent decision calculations within this scheduling cycle.

[0032] For example, in one execution cycle, the budget control server first initiates an API call to the target advertising platform to obtain the raw performance data of two advertising campaigns, A and B. Assume that campaign A has a current budget of 100.00 yuan and a current cumulative expenditure of 65.00 yuan; campaign B has a current budget of 300.00 yuan and a current cumulative expenditure of 250.00 yuan. Also assume that the cumulative historical deviation variable recorded locally for both campaigns A and B is 0.00 yuan. Next, the local arbitration log is queried, revealing the following adjustments in progress: campaign A has one unconfirmed increase instruction of 30.00 yuan; campaign B has one unconfirmed increase instruction of 50.00 yuan and one unconfirmed decrease instruction of 20.00 yuan. Based on this, the in-progress budget increase for campaign A is calculated to be 30.00 yuan, and the in-progress budget decrease is 0.00 yuan; the in-progress budget increase for campaign B is 50.00 yuan, and the in-progress budget decrease is 20.00 yuan. Subsequently, a logical merge calculation is performed. For Activity A, its snapshot budget value is calculated as 100.00 + 30.00 - 0.00 + 0.00 = 130.00 yuan. For Activity B, its snapshot budget value is calculated as 300.00 + 50.00 - 20.00 + 0.00 = 330.00 yuan. Finally, the system generates a budget status snapshot containing two records: the first record is for Activity A, with a current cumulative expenditure of 65.00 yuan and a snapshot budget value of 130.00 yuan; the second record is for Activity B, with a current cumulative expenditure of 250.00 yuan and a snapshot budget value of 330.00 yuan. This budget status snapshot will serve as input for the next processing step.

[0033] In one embodiment of the present invention, step S2 includes the following steps: Calculate the budget utilization rate index based on the snapshot budget value and the current cumulative spending value, which reflects the intensity of budget demand. Based on the current cumulative spending and number of converted orders, calculate the conversion cost efficiency characteristics, and based on the actual click volume, calculate the smoothed click volume of the traffic exposure potential characteristics. The system checks whether each advertising campaign generates conversion orders. If so, it multiplies the conversion cost efficiency feature, budget utilization rate, and smoothed click volume by a preset first weighting coefficient and sums them to generate a dynamic value score. If not, it multiplies the budget utilization rate and smoothed click volume by a preset second weighting coefficient and sums them to generate a dynamic value score.

[0034] Specifically, after obtaining the budget status snapshot, the calculation module of the budget control server traverses each advertising campaign record in the budget status snapshot, parses out the current cumulative spending value and snapshot budget value contained therein, and simultaneously extracts performance indicators such as the number of converted orders and the actual click volume contained in the record.

[0035] First, the budget utilization rate, representing the intensity of budget demand, is calculated by dividing the current cumulative spending by the snapshot budget. Next, for each advertising campaign, its conversion cost efficiency characteristic is calculated based on its number of converted orders and a target conversion cost floor threshold preset in the system configuration library. Specifically, the actual conversion cost of the advertising campaign is obtained by dividing the current cumulative spending by the number of converted orders. It should be noted that the actual conversion cost is a metric that measures the cost of advertising investment required to acquire a single converted order. Then, the target conversion cost floor threshold is divided by the calculated actual conversion cost to obtain a ratio, which serves as the conversion cost efficiency characteristic. In this embodiment, the preset target conversion cost lower limit threshold is a key business parameter. Its physical dimension is the same as the actual conversion cost: yuan / order. The ratio obtained after performing a division operation between the two is a dimensionless pure proportional constant, ensuring the uniformity of various weighted scoring data dimensions. This threshold is stored in the system configuration table, and its setting is usually based on the analysis of product profit margin. For example, assuming the application scenario is the promotion of highly competitive beauty products, in order to ensure the profitability of the campaign, based on historical data statistics and gross profit analysis, this threshold can be set to the cost of acquiring a single order between 50.00 yuan and 80.00 yuan.

[0036] For the traffic exposure potential feature, a logarithmic function is used to smooth the actual click volume value. Specifically, the actual click volume value is increased by a constant of 1 and then the common logarithm is taken to generate a smoothed click volume. It should be understood that smoothing the click volume is a logarithmic transformation of the original click volume, aiming to reduce the significant differences in click volume levels between different advertising campaigns, making the impact of traffic scale on the final score more balanced. Finally, a comprehensive weighted calculation is performed to generate a dynamic value score. This process achieves dynamic adjustment through a conditional branch, checking whether the currently processed advertising campaign has at least one converted order. If it does, the conversion cost efficiency feature, budget utilization rate, and smoothed click volume are multiplied by a set of preset first weighting coefficients and then summed. If not, only the budget utilization rate and smoothed click volume are multiplied by a set of preset second weighting coefficients and then summed.

[0037] Furthermore, the weighting ratio is used to balance the importance of the three feature dimensions, and its setting is based on business priorities. For example, for activities that generate converted orders, the first weighting ratio can be set to... , , This emphasizes the core importance of conversion efficiency. For campaigns with no converted orders, the second weighting ratio can directly use the original weighting. , At this point, due to the lack of a dominant term with a weight of 0.7, its score will naturally be lower than that of activities with conversion, thus implicitly inhibiting activities without conversion. The calculation process of dynamic value score can be represented by the following piecewise function: When the number of converted orders hour:

[0038] When the number of converted orders hour:

[0039] In the formula, It is a core indicator for dynamic value scoring and quantitative assessment of the overall potential of an activity. This is the lower limit threshold for the target conversion cost, set based on the product's gross profit margin and historical conversion data. It is used to measure whether the current actual conversion cost has room for excess profit. This represents the current cumulative spending value, the actual amount spent obtained from the platform. The number of converted orders can be understood as a business achievement metric. This represents the actual number of clicks. The +1 is to prevent errors caused by taking the logarithm of zero clicks, and is a mathematical smoothing process. This is a weighting coefficient, set based on business operation orientation. If the orientation is based on return on investment (ROI), then set... Larger, such as 0.7, , When there are no conversions, remove the main weight to ensure a naturally low score for activities with no conversions and prevent blind bias.

[0040] After the above processing, each advertising campaign is assigned a dynamic value score that quantifies its placement potential and a budget utilization metric that reflects its consumption status. These results will be passed to subsequent steps for strategy stratification.

[0041] For example, following the example from the previous step, the server processes a snapshot of the budget status of activities A and B. In the system configuration, the preset target conversion cost threshold is 30.00 yuan, the first weighting coefficient is (0.7, 0.1, 0.2), and the second weighting coefficient is (0.1, 0.2). First, activity A is processed, with a current cumulative spending of 65.00 yuan, a snapshot budget of 130.00 yuan, and it is assumed that its number of converted orders is 2 and its actual click volume is 150. Its budget utilization rate is calculated as 65.00 / 130.00 = 0.5. Next, the actual conversion cost is calculated as 65.00 / 2 = 32.50 yuan, resulting in a conversion cost efficiency characteristic of 30.00 / 32.50 ≈ 0.923. Then, the smoothed click volume is calculated as log... 10 (150+1)≈2.179. Since Activity A has converted orders, a weighted sum is used with the first weight, calculating its dynamic value score as 0.7×0.923+0.1×0.5+0.2×2.179≈0.646+0.05+0.436=1.132. Next, Activity B is processed, with a current cumulative spending of 250.00 yuan, a snapshot budget of 330.00 yuan, and assumed to have 0 converted orders and 800 actual clicks. Its budget utilization rate is calculated as 250.00 / 330.00≈0.758. Since it has no converted orders, the conversion cost efficiency feature is not calculated. Its smoothed click volume is log 10 (800+1)≈2.904. Using the second weight for weighted summation, the dynamic value score is calculated as 0.1×0.758+0.2×2.904≈0.076+0.581=0.657. Finally, the output of this step is: Activity A's budget utilization rate is 0.5, and its dynamic value score is 1.132; Activity B's budget utilization rate is 0.758, and its dynamic value score is 0.657.

[0042] In one embodiment of the present invention, step S3 includes the following steps: Read the pre-configured lower and upper limits of the conversion cost threshold; Advertising campaigns with no converted orders or actual conversion costs exceeding the conversion cost threshold are filtered out and assigned a tiered label to the elimination group. Advertising campaigns that generate conversion orders and whose actual conversion cost is not higher than the lower limit of the conversion cost threshold are selected and assigned a high-value group tier label. The remaining advertising campaigns that have generated conversion orders and whose actual conversion costs are between the lower and upper limits of the conversion cost threshold are assigned a low-value group tier label.

[0043] Specifically, after obtaining the dynamic value score and budget utilization rate, the strategy layering module of the budget control server reads the pre-configured upper and lower limits of the conversion cost threshold from the system configuration database. These are the lower and upper limits of the conversion cost threshold. It's important to note that the lower limit is typically set based on the business's profit model, ensuring that the conversion cost of high-value advertising campaigns remains within an acceptable profit margin. For example, it can be set to 80% of the historical average conversion cost or calculated backwards from the product's gross profit margin. The upper limit, or target tolerance limit, is the highest conversion cost the system can tolerate. It allows for a buffer when setting this limit. For instance, assuming the application scenario is a new product launch period, allowing for higher trial-and-error costs, it can be set to 120% to 150% of the lower limit.

[0044] The module then iterates through each advertising campaign processed in the previous round of calculations. For each campaign, the system performs a set of ordered logical checks to determine its hierarchy.

[0045] First, the system checks if the number of converted orders for the campaign is zero. If it is zero, the campaign is directly assigned the elimination group's advertising campaign tier identifier, and the tier determination for the campaign ends. If the number of converted orders is greater than zero, the system continues to the next step, comparing the actual conversion cost calculated in step S2 with the upper limit of the conversion cost threshold. If the actual conversion cost is strictly greater than the upper limit of the conversion cost threshold, the campaign is also assigned the elimination group's advertising campaign tier identifier. If neither of the aforementioned elimination conditions is met, the system then compares the actual conversion cost with the lower limit of the conversion cost threshold. If the actual conversion cost is not higher than the lower limit of the conversion cost threshold, the campaign is assigned the high-value group's advertising campaign tier identifier.

[0046] Finally, for all advertising campaigns that do not meet any of the above conditions—that is, campaigns with converted orders and whose actual conversion cost is between the lower and upper limits of the conversion cost threshold—the system assigns them a low-value campaign tier label. This campaign tier label is a classification tag used to distinguish the priority of campaign budget adjustments. It includes three preset tiers: the high-value group represents campaigns with high conversion efficiency that should be given priority for budget replenishment; the low-value group represents campaigns with acceptable performance that are within the observation range; and the elimination group represents campaigns with no conversions or excessively high conversion costs that should be used as a source of budget recovery.

[0047] Once this process is completed, each participating advertising campaign will receive a clear tier identifier, which will serve as the core basis for determining priorities and tolerance in subsequent budget allocation decisions.

[0048] For example, the output data from the aforementioned steps is used to strategically stratify advertising campaigns A and B. The system first loads parameters from the configuration library, assuming the lower limit of the conversion cost threshold used in this execution is 30.00 yuan and the upper limit is 36.00 yuan. Next, the system processes campaign A, whose actual conversion cost calculated in step S2 is 32.50 yuan, and whose number of converted orders is 2. The system first determines that its number of converted orders is not zero, then determines that its actual conversion cost of 32.50 yuan is not greater than the upper limit of 36.00 yuan, and finally determines that its actual conversion cost of 32.50 yuan is higher than the lower limit of 30.00 yuan. Therefore, campaign A does not meet the conditions for the high-value group and the elimination group, and is assigned the advertising campaign tier identifier of the low-value group. Subsequently, the system processes campaign B, whose number of converted orders is 0. According to the tiering logic, the system detects that its number of converted orders is zero, directly meeting the admission conditions for the elimination group. Therefore, campaign B is assigned the advertising campaign tier identifier of the elimination group. Finally, this step completes the stratification, and the output is: Campaign A is categorized as low-value, and Campaign B is categorized as eliminated. These stratifications, along with the dynamic value score, will be used in the next step of supply and demand matching calculation.

[0049] In one embodiment of the present invention, step S4 includes the following steps: Ad campaigns whose budget utilization rate exceeds the preset demand-side threshold lower limit are identified as demand-side campaigns and arranged in descending order according to the priority of the ad campaign hierarchy to form a priority supplement sequence. All potential source advertising campaigns are arranged between groups according to the reverse priority from the elimination group to the high-value group, and dynamic value scores are used to sort them in reverse order from low to high within the group to determine the forced extraction priority within the group. Based on the priority order of replenishment sequence and mandatory extraction priority within the group, the transferable amount of the source activity is calculated to generate a budget transfer request that includes demand source information, power supply information, and transferable amount.

[0050] Specifically, after assigning campaign tier identifiers to all advertising campaigns, the supply and demand matching calculation module on the budget control server first reads the preset lower limit of the demand side threshold from the system configuration. It should be noted that the lower limit of the demand side threshold is a value used to identify campaigns with tight budgets. Its setting is based on the business operations' sensitivity to the rate of budget depletion. For example, for scenarios aiming for stable 24 / 7 ad delivery, this threshold can be set to 0.8, meaning that supplementary demand is triggered when the budget utilization rate exceeds 80%.

[0051] The module iterates through all participating advertising campaigns, comparing their budget utilization rate calculated in step S2 with this threshold. Any advertising campaign whose budget utilization rate exceeds the lower limit of the demand-side threshold and whose campaign level identifier does not belong to the elimination group is identified as a demand-side campaign and added to the demand-side list. Subsequently, the module performs a sorting operation on this demand-side list, using the campaign level identifier as the primary sorting key to arrange it in ascending or descending order; that is, high-value groups take precedence over low-value groups, and low-value groups take precedence over elimination groups, thus generating a priority replenishment sequence that defines the budget replenishment priority.

[0052] Simultaneously, the module treats all advertising campaigns as potential sources and sorts them independently. This sorting first executes a reverse sequence arrangement between groups based on the campaign's tier identifier, with elimination groups first, low-value groups next, and high-value groups last. Within each tier group, a dynamic value score is used as a secondary sorting key, performing a reverse order from low to high to determine the forced extraction priority within the group. Finally, the system initiates the matching calculation process, which searches for budgets for each demand-side campaign in the order that the priority supplement sequence was generated.

[0053] For each demand-side activity, the system will attempt to extract budget from the source-side activities in turn, according to the mandatory extraction priority within the group. When attempting to extract budget from a source-side activity, the system will calculate a transferable amount. The calculation logic for this amount is as follows: First, obtain the baseline original budget of the source-side activity and multiply it by a predetermined ratio to obtain the maximum allowable recovery amount; then, take the smaller of this maximum allowable recovery amount and the current available remaining budget (snapshot budget value) of the source-side activity minus the currently spent value, and take the smaller of the two as the final transferable amount.

[0054] In this embodiment, the baseline budget refers to the initial budget amount allocated to each advertising campaign at the beginning of the daily budget allocation cycle. It serves as a stable and unchanging reference base to ensure the fairness of the recovery ratio calculation. The predetermined ratio is a safety factor used to limit the strength of a single recovery. Its setting is intended to prevent a drastic impact on the budget of the source campaign. For example, assuming the application scenario is the promotion of a mature product, in order to maintain the stability of the campaign, this ratio can be set to 0.1 to 0.2.

[0055] Specifically, the calculation process for transferable amount can be represented by the following formula:

[0056] In the formula, For demand-side activities, For the source of the activity. For transferable amounts, i.e., the amount determined from How much funding to allocate . The initial budget is set based on the original budget and allocated at midnight of the current day. It remains unchanged despite frequent system transfers, ensuring a stable base. The maximum recovery rate for a single transaction is set at a predetermined ratio and is pre-allocated based on market volatility. It is usually set at 10% to 20% to prevent the source's budget from being completely drained at once, which would cause them to stop investing due to traffic restrictions, thus ensuring the stability of the investment.

[0057] Once a transferable amount greater than zero is calculated, the system immediately constructs a budget transfer request. This request is a structured data object that encapsulates a complete budget scheduling intent and serves as a bridge connecting supply and demand matching calculations with subsequent physical constraint verification. The request contains a data structure including the demand-side activity identifier, the source-side activity identifier, and the transferable amount, and is submitted to the next step for arbitration.

[0058] For example, the data generated in the previous steps is reused, and a new high-value activity C is introduced to demonstrate the demand scenario. The current activity list includes: Activity A, low-value group, dynamic value score 1.132, budget utilization rate 0.5; Activity B, eliminated group, dynamic value score 0.657, budget utilization rate 0.758; Activity C, high-value group, dynamic value score 2.5, budget utilization rate 0.85. The system's configured demand side threshold lower limit is 0.8, and the predetermined ratio is 0.1. In addition, the system retrieves the baseline original budget for each activity from the initialization record, assuming Activity A is 100.00 yuan, Activity B is 300.00 yuan, and Activity C is 100.00 yuan. First, the system identifies the demand side. Since only Activity C's budget utilization rate of 0.85 is greater than 0.8, it prioritizes supplementing the sequence with only Activity C. Next, the system sorts all activities by their source. Based on the rules of hierarchical reverse (elimination group - low-value group - high-value group) and ascending score order within the hierarchy, the resulting source extraction order is: Activity B, Activity A, Activity C. Subsequently, the system searches for budget for Activity C in the priority replenishment sequence, first attempting to extract it from the highest priority source, Activity B. The system calculates the transferable amount from Activity B: its available remaining budget is the snapshot budget value minus the expenditure, i.e., 330.00 - 250.00 = 80.00 yuan; its maximum allowable recovery amount is the base original budget multiplied by a predetermined ratio, i.e., 300.00 × 0.1 = 30.00 yuan. Taking the smaller of the two, the transferable amount is 30.00 yuan. Since the calculated amount is greater than zero, the system immediately constructs a budget transfer request, which requests the transfer of 30.00 yuan of budget from source Activity B to demander Activity C. This budget transfer request will be passed to step S5 for further processing.

[0059] In one embodiment of the present invention, step S5 includes the following steps: Obtain the snapshot budget value and current cumulative spending value of the source activity from the budget status snapshot, and subtract the transferable amount to be transferred in the budget transfer application to obtain the actual available remaining space; The built-in hierarchical flow rule library is invoked to conduct a compliance review of this budget transfer operation. The rule library includes rules that prohibit downward bypassing of budget collection. When the available remaining space is greater than the minimum budget retention protection threshold pre-allocated at each level and passes the level flow rule verification, a controlled unconfirmed budget adjustment instruction is generated and written to the arbitration log with the status details of extracting deduction and injecting gain.

[0060] This step is handled by the delay-aware budget recovery arbitration module, which is a software logic unit whose core function is to resolve concurrency conflicts that may be caused by platform activation delays by verifying physical and logical constraints.

[0061] Upon receiving the budget transfer request generated in the previous step, the budget recovery arbitration module immediately initiates a verification process to perform physical constraint verification. First, based on the source activity identifier recorded in the budget transfer request, the module extracts the record corresponding to the source activity from the budget status snapshot generated in step S1, and calculates the snapshot remaining budget for that activity by subtracting the current accumulated expenditure value from its snapshot budget value. Subsequently, the module subtracts the transferable amount to be transferred in this budget transfer request from the snapshot remaining budget to obtain the actual available remaining space. Since the snapshot data has already deducted all outstanding and unconfirmed reduction instructions in step S1, the problem of concurrent duplicate deductions is directly prevented at the physical level. Specifically, the calculation process of the available remaining space can be expressed by the following formula:

[0062] The conditions for determining release are:

[0063] In the formula, Available remaining space represents the source activity under extreme concurrency conditions. A genuine guaranteed cash return. Reserve a protection threshold for the minimum budget. Business fallback parameters. For the elimination group, this can be fixed to a minimum value, such as 1.0 yuan; for the observation group, it can be dynamically set to a low level to ensure the activity continues uninterrupted regardless of deductions.

[0064] At the same time, the arbitration module calls the built-in hierarchical flow rule library to conduct a compliance review of this transfer operation. Specifically, it verifies whether this budget transfer violates rules such as prohibiting the transfer of budgets from high-level activities to low-level activities, which is a core policy constraint to ensure that the budget can only flow unidirectionally from low-value levels to high-value levels, and whether it triggers the low-score-no-replenishment-high-score anti-inversion mechanism between activities at the same level to prevent the excessive transfer of budgets from low-value activities to high-value activities.

[0065] Finally, the module performs the final decision. The decision is satisfied when the calculated available remaining space is strictly greater than the minimum budget retention protection threshold pre-configured for each level. ,in A minimum budget protection threshold is maintained. This threshold serves as a safety net for the business, designed to ensure that any activity can maintain a minimum level of operational capacity even after the budget is recovered. For example, a fixed, lower value such as 1.00 yuan can be set for the phase-out group, while the current spending amount can be set for the low-value and high-value groups, ensuring that the recovered budget does not lead to an abnormal situation where the spent amount exceeds the budget. Furthermore, the budget transfer application is deemed approved only after all the aforementioned tiered flow rule verifications have been successfully completed.

[0066] The server then generates two controlled, unconfirmed budget adjustment instructions, one for the source and one for the demand side. Using atomic operations, it rewrites these two status details, representing the extraction of deductions and the injection of gains respectively, into the arbitration log. This immediately occupies the budget at the logical level, effectively preventing other budget recovery procedures that may be executed concurrently from repeatedly seizing the funds.

[0067] For example, the delay-aware budget recovery arbitration module receives a budget transfer request from step S4, requesting a transfer of 30.00 yuan from activity B to activity C. The module first queries the arbitration log for the source activity B. By subtracting its current accumulated expenditure of 250.00 yuan from its snapshot budget value of 330.00 yuan, it obtains 80.00 yuan. Since the snapshot has pre-emptively eliminated concurrent deduction interference, its actual available remaining space can be directly calculated, which is the snapshot remaining budget of 80.00 yuan minus the transferable amount of 30.00 yuan, resulting in 50.00 yuan. Simultaneously, the module performs rule verification. This transfer is from activity B in the elimination group to activity C in the high-value group, complying with the rule prohibiting downward bypassing of budget recovery, and is approved. Finally, the module compares the available remaining space of 50.00 yuan with the minimum budget retention protection threshold set for the elimination group, assuming this threshold is 1.00 yuan. Since 50.00 yuan is strictly greater than 1.00 yuan, the final decision is to approve the transfer. The server therefore generated two unconfirmed budget adjustment instructions: one for Activity B, reducing budget by 30.00 yuan, and the other for Activity C, increasing budget by 30.00 yuan. The details of these two instructions were immediately written to the arbitration log as unconfirmed, completing the arbitration and locking of this budget recovery.

[0068] In one embodiment of the present invention, step S6 includes the following steps: The snapshot budget values ​​obtained in step S1 for each advertising campaign within the high-value group are combined with the amount of the newly generated unconfirmed budget increase instruction for the advertising campaign in step S5 to calculate and evaluate the expected final budget and expected final utilization rate of each advertising campaign within the high-value group. Advertising campaigns that are expected to have a final usage rate that is still above the preset safety alert line and have the highest historical score are identified as unmet targets. From the independently established adaptive redundancy pool, compensation funds are allocated to unmet objects, and the compensation funds are encapsulated into emergency budget instructions that only contain the addition action and do not specify the source of recovery. The emergency budget instructions are then submitted for timing verification.

[0069] Specifically, after the internal budget flow is adjudicated, the budget control server performs a final smoothing compensation. First, based on the recently updated arbitration log, the expected status of each advertising campaign within the high-value group is reassessed. Specifically, for each high-value group campaign, the server sums the snapshot budget value obtained in step S1 with the amount of the newly generated unconfirmed budget increase instruction for that campaign in the internal supply and demand matching in step S5, thus calculating the expected final budget. Then, the current cumulative spending value of the campaign is divided by this expected final budget to obtain the expected final utilization rate, which reflects the budget consumption after the normal flow ends. It should be understood that the expected final utilization rate is a logical estimate of the budget consumption ratio of the advertising campaign after all approved but not yet effective budget adjustments have been implemented. Next, the expected final usage rate is compared with the safety alert line preset in the system configuration. This is a value higher than the demand-side threshold in step S4, for example, it can be set to 0.85-0.95. The setting is based on the fact that it serves as the last line of defense to trigger emergency intervention, ensuring that the highest priority advertising campaigns are not interrupted due to budget exhaustion. All advertising campaigns with expected final usage rates still above the safety alert line and with the highest historical score are selected and identified as unmet targets.

[0070] For selected unmet targets, the server will access a separately established reserve budget, known as the adaptive redundancy pool. This is a pre-allocated reserve fund, independent of the regular advertising campaign budget, typically a fixed percentage of the daily total budget, such as 10%, from which a compensation fund will be allocated. The amount of the compensation fund is precisely calculated to ensure that, after being injected, the expected final utilization rate of the unmet target will precisely fall back to the preset target steady-state value. The calculation process for the compensation fund can be expressed by the following formula:

[0071] The compensation for segmentation comes from an isolated adaptive redundancy pool; it is injected only and does not deduct from other activities. The target steady-state value is usually set as a healthy consumption level, such as 0.7% or 70%. The basis for this setting is that once funds are injected, the apparent consumption rate of this high-potential activity will immediately drop to 70%, exiting the demand side's endangered queue and leaving a sufficient consumption buffer. The amount injected into the advertising campaign in step S5 is the internal transfer amount that the demand side has just received.

[0072] Finally, the system encapsulates this compensation payment into an emergency budget instruction that only includes the addition action and does not specify the source of the recovery. This is a special type of budget adjustment instruction, whose source field is marked as the system redundancy pool to distinguish it from the regular flow between activities. It is then submitted again to the latency-aware budget recovery arbitration module for timing verification. Once the verification and payment are approved, the status details of this emergency budget instruction are also written to the arbitration log, thus completing the system's disaster recovery and mitigation work.

[0073] For example, the system continues processing high-value activity C from the previous steps and introduces new state data. Assume activity C's current cumulative cost is 95.00 yuan, its snapshot budget in S1 is 100.00 yuan, and after arbitration in S5, it receives an unconfirmed increase instruction of 10.00 yuan from activity B in the arbitration log. First, the expected final budget for activity C is calculated as 100.00 yuan + 10.00 yuan = 110.00 yuan. Next, its expected final utilization rate is calculated as 95.00 yuan / 110.00 yuan ≈ 0.864. The system's configured safety alert threshold is 0.85. Since 0.864 is greater than 0.85, and activity C is currently the only high-value activity awaiting compensation, it is identified as an unmet need. Subsequently, the system calculates compensation for it from the adaptive redundancy pool, with a target steady-state utilization rate of 0.7. The required compensation amount is (95.00 / 0.7) - 110.00 yuan ≈ 25.71 yuan. Finally, the system generates an emergency budget instruction to increase the budget of activity C by 25.71 yuan. After the instruction is verified by the arbitration module, its details are written to the arbitration log, awaiting unified distribution in the next stage.

[0074] In one embodiment of the present invention, step S7 includes the following steps: Summarize all approved but unconfirmed budget adjustment instructions and emergency budget instructions, encapsulate them into a batch communication request message, and send it to the target advertising platform; The asynchronous listening component is used to detect the execution result response of the target advertising platform and to parse and extract the confirmation status ticket containing whether the platform executed successfully and the actual time of processing. For the instruction content corresponding to the confirmation status ticket that has obtained a positive success signature, update its status in the arbitration log to confirmed, thereby releasing the computational occupation at the logical level; In response to unknown state command requests that cause deadlock due to network timeout or communication anomalies, the data tracing conservative locking link is activated, and the cumulative historical deviation variable corresponding to the advertising campaign is positively updated according to the amount attached to the abnormal command request, so as to form a placeholder lock to prevent duplicate distribution or deduction.

[0075] Specifically, after all budget transfer applications and emergency compensation requests have been arbitrated and recorded in the arbitration log, the execution and status closure module of the budget control server completes the final physical implementation and status closure. This module first performs a query on the arbitration log to extract all unconfirmed budget adjustment instructions and emergency budget instructions marked as unconfirmed. These instructions are then aggregated, each with a unique sequence identifier, and uniformly encapsulated into a batch communication request message containing an array of instructions. In this embodiment, the batch communication request message is a data structure that combines multiple independent API requests into a single network transmission, typically in JSON or XML format, designed to improve communication efficiency and reduce network overhead. Subsequently, the server uses its network communication component to deliver this message as a POST request using the HTTPS protocol to the pre-configured external target advertising platform application programming interface server. This operation triggers a real budget change on the advertising platform.

[0076] Meanwhile, an asynchronous listening component deployed on the server with a clock cycle property begins continuous operation. This is an independent software process or thread running in the background, responsible for handling asynchronous callback events from external systems, ensuring that the main decision-making process is not blocked by waiting for API responses. This component monitors the execution results of the budget change operation in real time by maintaining a long connection to the application programming interface (API) server-side event callback endpoint or by frequently polling the status query interface. Once a response is received from the platform, the component parses out the sequence identifier, the platform's success or failure status flag, and the timestamp of the actual time the platform received and processed the instruction, together forming a confirmation status ticket. It should be noted that the confirmation status ticket is the structured feedback from the advertising platform API regarding the execution result of the budget adjustment instruction, and is a key credential for achieving eventual consistency between the system state and the platform state.

[0077] For confirmation tickets containing a positive success signature, the server's status management unit locates the corresponding instruction record in the arbitration log based on its sequence identifier and updates its status from unconfirmed to confirmed. This operation effectively releases the computational burden on the budget at the logical level. Conversely, for unknown abnormal instruction requests that cannot obtain clear feedback due to network timeouts or communication link disconnections, if the system simply clears its in-transit flag, the possibility that the advertising platform has actually deducted funds but failed to successfully revert to the previous action could induce repeated dangerous actions in the system, such as duplicate compensation injections or duplicate over-purchases for the same object. Therefore, this system adopts a pessimistic locking principle to activate the data traceability conservative locking link. After clearing the unconfirmed flag of the original unknown abnormal instruction, this link updates the cumulative historical deviation variable corresponding to the advertising campaign in a positive retention manner based on the original adjustment amount and adjustment type attached to it, forming a suspended placeholder quota for the next cycle calculation. Specifically, the update process of the cumulative historical deviation variable can be represented by the following formula:

[0078] In the formula, This represents the monetary amount associated with failed execution of abnormal instructions, indicated by a direction symbol: a positive number for failed increases and a negative number for failed decreases. The rationale is as follows: Suppose the system injects 25 yuan into a high-value activity via S5, but encounters a network timeout error after sending it to the platform. Whether the platform actually deposited this 25 yuan is currently unknown. If locking and clearing the in-transit flag are not implemented, the system will retrieve the incompletely updated old value (e.g., 100 yuan) during the next data retrieval from S1, mistakenly believing no addition occurred, and will initiate another 25 yuan recharge in the current cycle. If the previous instruction actually timed out without reporting a reason, the total budget will be incorrectly overstated to 150 yuan, resulting in significant waste. After triggering the locking mechanism, the formula is executed. In the next snapshot merge, the formula will automatically add the 25 yuan held by the deviation to the 100 yuan retrieved via API, forcibly restoring the system's safe expected value of 125 yuan, thereby stabilizing the current situation and temporarily suppressing the repeated issuance of new instructions. (Cumulative historical deviation variable) These are individual state variables maintained in the server's memory.

[0079] For example, after summarizing the arbitration logs, the system finds two pending instructions: Instruction #001, deducting 30.00 yuan from Activity B and increasing 30.00 yuan from Activity C; Instruction #002, urgently increasing 25.71 yuan from Activity C. The system packages these two instructions into a JSON request message and sends it to the target platform. Subsequently, the asynchronous listening component begins to work. First, it receives a callback for Instruction #001, and the parsed confirmation status ticket shows successful execution. The server therefore updates the record status of Instruction #001 to confirmed in the arbitration log. Later, the listening component detects a communication timeout error for Instruction #002. Since it cannot be confirmed whether the platform has actually executed the compensation, the system activates data tracing and conservatively locks the link to prevent a duplicate recharge storm. Assuming the current cumulative historical deviation variable is 0.00 yuan, and the currently unclear Instruction #002 is a budget increase operation, its suspended amount... The value is +25.71 yuan. The system calculates the new cumulative historical deviation based on the locking formula: Yuan. This conservative placeholder value of +25.71 yuan will be recorded by the system. At the start of the next budget decision cycle, regardless of whether the budget for activity C returned by the platform interface has actually been updated due to delays, the system will use this deviation value as a safety cushion in the calculation of its budget status snapshot, thereby forcing the system to believe that the activity is temporarily at a safe level within the tolerance limit, preventing possible secondary or even multiple over-retry payment behaviors. Moreover, as described in step S1, this deviation value can be reset to zero after only one cycle of protection verification, effectively curbing repeated state oscillations caused by communication anomalies, and ensuring the absolute high stability of the closed-loop budget allocation system constructed by local execution, in-transit state, and anomaly compensation.

[0080] See appendix Figure 2 This invention also proposes an intelligent advertising budget allocation system based on dynamic value scoring, comprising the following modules: The budget status snapshot generation module is used to obtain the original performance data of each advertising campaign in the target advertising platform, synchronously extract the in-transit adjustment data recorded locally, and generate a budget status snapshot that eliminates the interference of effective delay through logical merging. The dynamic value scoring module is used to analyze the budget status snapshot, calculate the conversion cost efficiency characteristics, budget demand intensity characteristics, and traffic exposure potential characteristics of each advertising campaign, and generate dynamic value scores and budget utilization rate indicators based on the comprehensive weighting of these characteristics. The strategy stratification module, based on dynamic value scores and actual conversion costs, implements strategy stratification for advertising campaigns and determines the stratification identifiers for advertising campaigns; The supply and demand matching calculation module is used to combine budget utilization rate indicators, dynamic value scores and advertising campaign level identifiers to perform supply and demand matching calculations in order to construct a budget transfer request that includes demand source information, power supply information and transferable amount. The budget recovery arbitration module is used to input budget transfer requests into the delay-aware budget recovery arbitration module, perform physical constraint verification, generate controlled unconfirmed budget adjustment instructions, and update the arbitration log; The adaptive redundancy pool compensation module is used to monitor the budget utilization rate of high-value groups. When it reaches the preset conditions, it calls the adaptive redundancy pool to generate an emergency budget instruction and appends it to the arbitration log. The execution and status loop module is used to aggregate all approved budget adjustment instructions and emergency budget instructions and send them to the target advertising platform. It also performs status loop cleaning of the arbitration log by listening to the callback information returned by the asynchronous listening interface.

[0081] Each of the modules can be implemented in whole or in part through software, hardware, or a combination thereof. It supports hardware embedded in or independent of the processor in the computer device, and also supports software stored in the memory of the computer device, so that the processor can call and execute the operations corresponding to each of the above modules.

[0082] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for intelligent allocation of advertising budget based on dynamic value scoring, characterized in that, Includes the following steps: S1. Obtain the original performance data of each advertising campaign in the target advertising platform, synchronously extract the in-transit adjustment data recorded locally, and generate a budget status snapshot that eliminates the interference of effective delay through logical merging. S2. Analyze the budget status snapshot, calculate the conversion cost efficiency characteristics, budget demand intensity characteristics, and traffic exposure potential characteristics of each advertising campaign, and generate dynamic value scores and budget utilization rate indicators based on the comprehensive weighting of these characteristics. S3. Based on dynamic value scores and actual conversion costs, stratify the execution strategy of advertising campaigns and determine the level identifiers of advertising campaigns. S4. Combine budget utilization rate indicators, dynamic value scores, and advertising campaign level identifiers to perform supply and demand matching calculations to construct a budget transfer request that includes demand source information, power supply information, and transferable amount. S5. Input the budget transfer request into the delay-aware budget recovery arbitration module, perform physical constraint verification, generate controlled unconfirmed budget adjustment instructions, and update the arbitration log; S6. Monitor the budget utilization rate of high-value groups. When it reaches the preset conditions, call the adaptive redundancy pool to generate an emergency budget instruction and append it to the arbitration log. S7. Summarize all approved budget adjustment instructions and emergency budget instructions and send them to the target advertising platform. Then, perform a closed-loop cleanup of the arbitration log status by listening to the callback information returned by the asynchronous listening interface.

2. The intelligent advertising budget allocation method based on dynamic value scoring according to claim 1, characterized in that, Step S1 includes the following steps: Call the target advertising platform's application programming interface to obtain the current budget and current cumulative spending for each advertising campaign; Read all issued but not yet confirmed effective budget adjustment instructions in the local arbitration log, and calculate the increase and decrease in budget in transit for each advertising campaign. Obtain the cumulative historical deviation variable for the corresponding advertising campaign; The current budget value, the increase in budget in transit, and the cumulative historical deviation variable are added together, and the decrease in budget in transit is subtracted to generate a snapshot budget value for each advertising campaign; The snapshot budget value is associated with and encapsulated with the current cumulative spending value to generate a budget status snapshot.

3. The intelligent advertising budget allocation method based on dynamic value scoring according to claim 1, characterized in that, Step S2 includes the following steps: Calculate the budget utilization rate index based on the snapshot budget value and the current cumulative spending value, which reflects the intensity of budget demand. Based on the current cumulative spending and number of converted orders, calculate the conversion cost efficiency characteristics, and based on the actual click volume, calculate the smoothed click volume of the traffic exposure potential characteristics. The system checks whether each advertising campaign generates conversion orders. If so, it multiplies the conversion cost efficiency feature, budget utilization rate, and smoothed click volume by a preset first weighting coefficient and sums them to generate a dynamic value score. If not, it multiplies the budget utilization rate and smoothed click volume by a preset second weighting coefficient and sums them to generate a dynamic value score.

4. The intelligent advertising budget allocation method based on dynamic value scoring according to claim 1, characterized in that, Step S3 includes the following steps: Read the pre-configured lower and upper limits of the conversion cost threshold; Advertising campaigns with no converted orders or actual conversion costs exceeding the conversion cost threshold are filtered out and assigned a tiered label to the elimination group. Advertising campaigns that generate conversion orders and whose actual conversion cost is not higher than the lower limit of the conversion cost threshold are selected and assigned a high-value group tier label. The remaining advertising campaigns that have generated conversion orders and whose actual conversion costs are between the lower and upper limits of the conversion cost threshold are assigned a low-value group tier label.

5. The intelligent advertising budget allocation method based on dynamic value scoring according to claim 1, characterized in that, Step S4 includes the following steps: Ad campaigns whose budget utilization rate exceeds the preset demand-side threshold lower limit are identified as demand-side campaigns and arranged in descending order according to the priority of the ad campaign hierarchy to form a priority supplement sequence. All potential source advertising campaigns are arranged between groups according to the reverse priority from the elimination group to the high-value group, and dynamic value scores are used to sort them in reverse order from low to high within the group to determine the forced extraction priority within the group. Based on the priority order of replenishment sequence and mandatory extraction priority within the group, the transferable amount of the source activity is calculated to generate a budget transfer request that includes demand source information, power supply information, and transferable amount.

6. The intelligent advertising budget allocation method based on dynamic value scoring according to claim 1, characterized in that, Step S5 includes the following steps: Obtain the snapshot budget value and current cumulative spending value of the source activity from the budget status snapshot, and subtract the transferable amount to be transferred in the budget transfer application to obtain the actual available remaining space; The built-in hierarchical flow rule library is invoked to conduct a compliance review of this budget transfer operation. The rule library includes rules that prohibit downward bypassing of budget collection. When the available remaining space is greater than the minimum budget retention protection threshold pre-allocated at each level and passes the level flow rule verification, a controlled unconfirmed budget adjustment instruction is generated and written to the arbitration log with the status details of extracting deduction and injecting gain.

7. The intelligent advertising budget allocation method based on dynamic value scoring according to claim 1, characterized in that, Step S6 includes the following steps: The snapshot budget values ​​obtained in step S1 for each advertising campaign within the high-value group are combined with the amount of the newly generated unconfirmed budget increase instruction for the advertising campaign in step S5 to calculate and evaluate the expected final budget and expected final utilization rate of each advertising campaign within the high-value group. Advertising campaigns that are expected to have a final usage rate that is still above the preset safety alert line and have the highest historical score are identified as unmet targets. From the independently established adaptive redundancy pool, compensation funds are allocated for unmet objects, and the compensation funds are encapsulated into emergency budget instructions that only contain the addition action and do not specify the source of recovery. The emergency budget instructions are then submitted for timing verification.

8. The intelligent advertising budget allocation method based on dynamic value scoring according to claim 1, characterized in that, Step S7 includes the following steps: Summarize all approved but unconfirmed budget adjustment instructions and emergency budget instructions, encapsulate them into a batch communication request message, and send it to the target advertising platform; The asynchronous listening component is used to detect the execution result response of the target advertising platform and to parse and extract the confirmation status ticket containing whether the platform executed successfully and the actual time of processing. For the instruction content corresponding to the confirmation status ticket that has obtained a positive success signature, update its status in the arbitration log to confirmed, thereby releasing the computational occupation at the logical level; In response to unknown state command requests that cause deadlock due to network timeout or communication anomalies, the data tracing conservative locking link is activated, and the cumulative historical deviation variable corresponding to the advertising campaign is positively updated according to the amount attached to the abnormal command request, so as to form a placeholder lock to prevent duplicate distribution or deduction.

9. The intelligent advertising budget allocation method based on dynamic value scoring according to claim 5, characterized in that, The calculation method for the transferable amount includes: obtaining the baseline original budget of the source activity and multiplying it by a preset ratio to obtain the maximum allowable recovery amount; obtaining the current available remaining budget of the source activity, which is the snapshot budget value minus the current cumulative expenditure value; and taking the target value between the maximum allowable recovery amount and the current available remaining budget as the transferable amount.

10. An intelligent advertising budget allocation system based on dynamic value scoring, characterized in that: Includes the following modules: The budget status snapshot generation module is used to obtain the original performance data of each advertising campaign in the target advertising platform, synchronously extract the in-transit adjustment data recorded locally, and generate a budget status snapshot that eliminates the interference of effective delay through logical merging. The dynamic value scoring module is used to analyze the budget status snapshot, calculate the conversion cost efficiency characteristics, budget demand intensity characteristics, and traffic exposure potential characteristics of each advertising campaign, and generate dynamic value scores and budget utilization rate indicators based on the comprehensive weighting of these characteristics. The strategy stratification module, based on dynamic value scores and actual conversion costs, implements strategy stratification for advertising campaigns and determines the stratification identifiers for advertising campaigns; The supply and demand matching calculation module is used to combine budget utilization rate indicators, dynamic value scores and advertising campaign level identifiers to perform supply and demand matching calculations in order to construct a budget transfer request that includes demand source information, power supply information and transferable amount. The budget recovery arbitration module is used to input budget transfer requests into the delay-aware budget recovery arbitration module, perform physical constraint verification, generate controlled unconfirmed budget adjustment instructions, and update the arbitration log; The adaptive redundancy pool compensation module is used to monitor the budget utilization rate of high-value groups. When it reaches the preset conditions, it calls the adaptive redundancy pool to generate an emergency budget instruction and appends it to the arbitration log. The execution and status loop module is used to aggregate all approved budget adjustment instructions and emergency budget instructions and send them to the target advertising platform. It also performs status loop cleaning of the arbitration log by listening to the callback information returned by the asynchronous listening interface.

Images