Outdoor advertisement intelligent putting optimization method and system based on artificial intelligence

Abstract

The invention discloses an intelligent outdoor advertisement putting optimization method and system based on artificial intelligence, and relates to the technical field of putting optimization, and the method comprises the steps: defining an outdoor advertisement level, collecting advertisement basic data, defining four adjacent observation points of a current working point, and carrying out the linear interpolation according to a price and load two-axis linear interpolation proportion, obtaining a selection function, outputting selection probabilities of different levels, enabling the advertisement conversion frequency to be maximum as an advertisement target function, defining the total revenue of the platform, extracting a time snapshot record for each level by adopting a snapshot evaluation method, distributing the advertiser budget, and updating the residual budget. According to the method, the adjusted unified execution price conforming to the actual economic constraint is calculated through linear interpolation flexible capture and in combination with a budget allocation rule, advertisement display allocation is optimized, and scheme optimization is ensured not only from the angle of conversion rate in combination with causal effect verification and utilization rate deviation screening.

Description

Technical Field

[0001] This invention relates to the field of advertising optimization technology, and in particular to an intelligent advertising optimization method and system based on artificial intelligence. Background Technology

[0002] Outdoor advertising is an important marketing tool with a wide reach and broad audience coverage in public places. Traditionally, outdoor advertisers typically select advertising locations based on fixed advertising areas and estimate advertising costs manually to maximize advertising coverage or influence. With the development of big data-related technologies, advertisers are beginning to try to use big data to analyze and calculate data in order to customize more profitable advertising placement plans.

[0003] However, existing outdoor advertising optimization methods often overlook the dynamic control capabilities of ad levels and the global optimization function across regions, resulting in poor flexibility in resource allocation. Secondly, the causal relationship between advertisers' choices and conversion efficiency is often oversimplified or ignored. Most existing methods are based on static regression models or fixed rules, failing to fully integrate real-time data on changes in ad prices, resource load, and user behavior preferences, making it difficult to obtain accurate conversion efficiency prediction models. Furthermore, in terms of advertisers' resource utilization, existing technologies lack dynamic measurement and verification standards, leading to instability in the operation of ad resources under high or low load conditions. It is impossible to set the optimal target utilization rate based on real historical data to guide the optimization of ad resource allocation. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an AI-based intelligent outdoor advertising placement optimization method and system to address the shortcomings of existing outdoor advertising placement optimization methods. These methods often neglect the dynamic control capabilities of advertising levels and the global optimization function across regions, resulting in poor resource allocation flexibility. Secondly, the causal relationship between advertisers' choices and conversion efficiency is often oversimplified or ignored. Most existing methods are based on static regression models or fixed rules, failing to fully integrate real-time changes in advertising prices, resource load, and user behavior preferences, making it difficult to obtain accurate conversion efficiency prediction models. Furthermore, in terms of advertisers' resource utilization rate, existing technologies lack dynamic measurement and verification standards, leading to instability in the operation of advertising resources under high or low load conditions. This makes it impossible to set optimal target utilization rates based on real historical data to guide the optimization of advertising resource allocation.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides an intelligent outdoor advertising delivery optimization method based on artificial intelligence, comprising:

[0008] Define outdoor advertising levels, collect basic advertising data, define four neighboring observation points of the current working point, obtain the selection function and output the selection probability of different levels according to the linear interpolation ratio of price and load, take the maximum number of advertising conversions as the advertising objective function, define the total revenue of the platform, use the snapshot evaluation method to extract time snapshot records for each level, allocate the advertiser's budget, and update the remaining budget.

[0009] The system measures the utilization rate of ad placement resources currently used by advertisers and updates the total revenue of the platform. It uses historical data of each level as discrete utilization points to find the point with the highest corresponding revenue value as the target utilization rate. It enumerates the price and frequency pairs observed in history for each level, interpolates to calculate the probability of the advertiser's selection, re-estimates the budget-averaged price and execution price, updates the total revenue of the platform, and calculates the deviation between the predicted parameter combination and the target utilization rate.

[0010] The average conversion rate and contrast effect value are calculated based on high-frequency samples and regular samples respectively. Combined with dual gating to screen candidate schemes, a set of candidate frequencies that meet the target utilization deviation limit is output.

[0011] For each frequency, the total revenue of the updated platform is calculated and updated, and the corresponding price combination is determined. The difference between the actual utilization rate and the expected resource utilization rate is calculated and compared.

[0012] As a preferred embodiment of the AI-based intelligent outdoor advertising placement optimization method of the present invention, the method involves: obtaining a selection function based on a biaxial linear interpolation ratio of price and load, outputting selection probabilities for different levels, maximizing ad conversions as the ad objective function, defining the platform's total revenue, extracting time snapshot records for each level using a snapshot evaluation method, allocating the advertiser's budget, and updating the remaining budget, including...

[0013] The advertising load is calculated as the ratio of the total number of ad impressions in a given month to the total ad viewing time in a given month for different tiers.

[0014] Define four neighboring observation points for the current working point, including the most recent value that is no greater than the current price and the most recent value that is no less than the current price from the set of historically actual advertising prices, and the most recent value that is no greater than the current advertising load and the most recent value that is no less than the current advertising load from the set of historically actual advertising loads.

[0015] For each level, the average percentage of samples from four neighboring observation points is directly calculated from historical data and used as the four-corner empirical percentage.

[0016] Based on the linear interpolation ratio between price and load, the selection function is obtained and the selection probabilities of different levels are output;

[0017] Calculate the maximum monthly sales display volume for each level based on the selection probability of different levels, determine the uniform execution price under the supply and budget caliber, and ensure that the price is neither higher than the historical transaction level, and that the budget is absorbed according to the supply scale, and calculate the uniform price per thousand times used for execution.

[0018] Define the objective function as maximizing the number of ad conversions without exceeding the budget, and define constraints accordingly;

[0019] The platform's total revenue is defined based on paid placement revenue and advertising sales revenue.

[0020] The snapshot evaluation method is used to extract time snapshot records for each level from the historical observation set, including the level placement price at the time of the snapshot, the ad load at the time of the snapshot, the observed level selection ratio at the time of the snapshot, and the average daily viewing time per person at the time of the snapshot.

[0021] Allocate advertiser budgets, calculate the maximum number of ad impressions that can be purchased, the maximum number of ad impressions that can be afforded with the current remaining budget, and the actual number of impressions allocated for each tier, and update the remaining budget;

[0022] Synchronously update the snapshot with advertising revenue, paid placement revenue, total platform revenue, and advertiser's expected total conversion;

[0023] Based on the collection of time snapshot records, the total platform revenue for each snapshot record is calculated and the maximum value is selected as the snapshot index. The actual allocated impressions are used as the recommended ad allocation, and the corresponding calculated recommended ad revenue, recommended ad revenue, and advertiser's expected total conversion are used as the expected results.

[0024] As a preferred embodiment of the AI-based intelligent outdoor advertising delivery optimization method of the present invention, the following steps are included: finding the point with the highest corresponding revenue value as the target utilization rate; enumerating historical real-observed price and frequency pairs for each level; interpolating to calculate the advertiser's selection probability; re-estimating the budget-averaged price and execution price; updating the platform's total revenue; and calculating the deviation between the predicted parameter combination and the target utilization rate, including...

[0025] Based on the expected results and the resource allocation chosen by the advertiser, the utilization rate of the ad placement resources currently used by the advertiser is measured, and the total revenue of the platform is updated and calculated.

[0026] The historical data of the hierarchy is analyzed according to discrete utilization rate points. The revenue contribution of each utilization rate point is analyzed, the corresponding average revenue value is calculated, and the point with the largest corresponding revenue value is found from the set of discrete utilization rate points as the target utilization rate.

[0027] For each level, enumerate the price and frequency pairs of historical real observations;

[0028] Based on the ad placement selection function, the probability of the advertiser's selection is calculated by interpolation, and the supply ceiling after the tier adjustment is simulated.

[0029] Based on the adjusted supply forecast, the budget-averaged price and execution price are re-estimated, and the advertiser's actual spending allocation under the adjusted parameters is calculated according to the budget allocation rules. The remaining budget is updated gradually until it is exhausted, the platform's total revenue is updated, and the deviation between the predicted parameter combination and the target utilization rate is calculated.

[0030] As a preferred embodiment of the AI-based intelligent outdoor advertising delivery optimization method of the present invention, the step of calculating the average conversion rate and contrast effect value based on high-frequency samples and regular samples respectively, and combining dual gating to screen candidate schemes, outputs a candidate frequency set that meets the target utilization deviation limit, including:

[0031] Based on the historical exposure frequency set of each level, a high-frequency threshold is calibrated according to historical experience, and exposure frequencies above the high-frequency threshold are defined as high-frequency samples, while exposure frequencies below the high-frequency threshold are defined as regular samples.

[0032] The average conversion rate was calculated based on both high-frequency and regular samples, and the contrast effect value was also calculated.

[0033] The candidate scheme is screened by combining dual gating, including eliminating combinations that do not meet the target utilization rate and eliminating frequency values ​​that are causally invalid, and outputting a set of candidate frequencies that meet the target utilization rate deviation limit.

[0034] As a preferred embodiment of the AI-based intelligent outdoor advertising delivery optimization method of the present invention, the step of calculating and updating the platform's total revenue for each frequency and determining the corresponding price combination includes:

[0035] The candidate frequency set is sorted according to the platform's total revenue contribution index, and the platform's total revenue is updated by substituting each frequency into the calculation. The corresponding price combination is determined, and the final frequency adjustment value and corresponding revenue price at the tier are determined.

[0036] As a preferred embodiment of the AI-based intelligent outdoor advertising delivery optimization method of the present invention, the step of comparing the difference between the actual utilization rate and the expected resource utilization rate includes:

[0037] Based on the frequency and price combination of execution, the actual utilization rate is calculated and compared with the expected resource utilization rate. The sum of the mean and standard deviation of the historical differences is used as the comparison threshold. If the actual calculated difference is less than or equal to the comparison threshold, the comparison is considered successful.

[0038] As a preferred embodiment of the AI-based intelligent outdoor advertising delivery optimization method of the present invention, the step of defining outdoor advertising levels and collecting basic advertising data includes:

[0039] The definition of outdoor advertising levels includes ad placement level, ad placement rental price, and ad placement listing price;

[0040] The collected basic advertising data includes the total number of target audiences, the number of times ads are displayed at different levels, the total viewing time and display frequency, ad placement pricing, the total budget for advertising and historical transaction prices, and the conversion rate of ads at each level.

[0041] Secondly, the present invention provides an intelligent outdoor advertising delivery optimization system based on artificial intelligence, comprising,

[0042] Data acquisition module: Collects basic data on outdoor advertising and defines the hierarchy; collects and calibrates the four neighboring observation points of the current working point according to the price and load axes.

[0043] Conversion module: Constructs a selection function by using the linear interpolation ratio of price and ad load, calculates the selection probability of different levels, and sets maximizing the number of ad conversions as the conversion objective function;

[0044] Total Revenue Calculation Module: With the platform's total revenue as the core objective, it uses a snapshot evaluation method to extract time snapshot records for verification and optimization of total revenue;

[0045] Budget allocation module: Allocates advertiser budgets and updates remaining budgets in real time;

[0046] Historical data analysis module: Calculates the target utilization rate based on the evaluation of advertising resource consumption and utilization rate, combined with hierarchically discretized historical utilization rate data;

[0047] Frequency Enumeration Module: Enumerates historical price and frequency combinations, interpolates to calculate the advertiser's selection probability, and re-outputs the budget-averaged price and execution price;

[0048] Conversion rate calculation module: Calculates the contrast effect value by comparing the average conversion rates of high-frequency and regular samples, and combines dual gating to screen the frequency set that meets the resource optimization requirements;

[0049] Utilization verification module: Calculates the total revenue contribution corresponding to frequency adjustment, generates recommended price combinations, and verifies and calibrates the deviation between actual utilization rate and resource utilization rate.

[0050] Thirdly, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the intelligent outdoor advertising delivery optimization method based on artificial intelligence as described in the first aspect of the present invention.

[0051] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the intelligent outdoor advertising delivery optimization method based on artificial intelligence as described in the first aspect of the present invention.

[0052] The beneficial effects of this invention are as follows: By flexibly capturing and calculating the adjusted unified execution price that conforms to actual economic constraints through linear interpolation and combined with budget allocation rules, the invention optimizes the allocation of ad display. By combining causal effect verification and utilization deviation screening, it ensures that the optimization of the solution is not only from the perspective of conversion rate, but also from the optimal solution of global supply and demand matching and economic benefit improvement. By selecting the maximum benefit point as the execution solution through snapshot index, the invention solves the subjective problem of weight assumptions in traditional optimization, ensuring that the actual adjusted results can return to the historical database to form a closed loop. By measuring resource consumption utilization and updating the platform's total revenue, the technical solution calculates resource utilization efficiency in real time based on the ratio between the advertiser's actual display volume and the maximum available resource volume. By eliminating the frequency combinations of utilization deviation and causal invalid pairs, it generates a candidate frequency set that meets multiple constraints. Attached Figure Description

[0053] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is a flowchart illustrating the AI-based intelligent outdoor advertising delivery optimization method in Example 1.

[0055] Figure 2 This is a schematic diagram of the structure of the AI-based intelligent outdoor advertising delivery optimization system in Example 1. Detailed Implementation

[0056] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0057] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0058] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0059] Example 1, referring to Figures 1 to 2 This is the first embodiment of the present invention, which provides an intelligent outdoor advertising delivery optimization method based on artificial intelligence, including the following steps:

[0060] S1 defines the outdoor advertising hierarchy, collects basic advertising data, defines four neighboring observation points of the current working point, obtains the selection function and outputs the selection probability of different levels according to the linear interpolation ratio of price and load, takes the maximum number of advertising conversions as the advertising objective function, defines the total revenue of the platform, uses the snapshot evaluation method to extract time snapshot records for each level, allocates the advertiser's budget, and updates the remaining budget.

[0061] Preferably, outdoor advertising levels are defined, and basic advertising data is collected, including:

[0062] The definition of outdoor advertising levels includes ad placement level, ad placement rental price, and ad placement listing price;

[0063] The collected basic advertising data includes the total number of target audiences, the number of times ads are displayed at different levels, the total viewing time and display frequency, ad placement pricing, the total budget for advertising and historical transaction prices, and the conversion rate of ads at each level.

[0064] Furthermore, based on the linear interpolation ratio between price and load, a selection function is obtained, and the selection probabilities for different levels are output. Maximizing the number of ad conversions is used as the ad objective function. The platform's total revenue is defined, and a snapshot evaluation method is used to extract time snapshot records for each level. Advertiser budgets are allocated, and remaining budgets are updated, including...

[0065] The advertising load is calculated as the ratio of the total number of ad impressions in a given month to the total ad viewing time in a given month for different tiers.

[0066] Define four neighboring observation points for the current working point, including the most recent value that is no greater than the current price and the most recent value that is no less than the current price from the set of historically actual advertising prices, and the most recent value that is no greater than the current advertising load and the most recent value that is no less than the current advertising load from the set of historically actual advertising loads.

[0067] For each level, the average percentage of samples from four neighboring observation points is directly calculated from historical data and used as the four-corner empirical percentage.

[0068] Based on the linear interpolation ratio between price and load axes, the selection function is obtained, and the selection probabilities at different levels are output, as follows:

[0069]

[0070]

[0071]

[0072] in, and These represent the linear interpolation ratios along the price and load axes, respectively. This indicates the price of a space at level L. and They respectively represent not greater than the current value. The most recent value of the price of a display space is not less than the current value. The most recent value of the price of a display space. This represents the ad load at level L. and They respectively represent not greater than the current value. The most recent value of the ad load is not less than the current value. The most recent value of ad load, This represents a selection function. This represents the sample average at the current price and load.

[0073] Based on the selection probability of different levels, the maximum monthly sales display volume for each level is calculated. A uniform execution price is determined under the supply and budget framework, ensuring that the price neither exceeds historical transaction levels nor exceeds the budget based on supply scale. The uniform price per thousand executions is calculated and expressed as follows:

[0074]

[0075]

[0076]

[0077] in, This represents the maximum monthly sales display volume for tier L. This represents the total number of unique target audience members within the statistics window. express That is, the probability that the target audience will choose level L under the current ad placement price and ad load conditions. This indicates the average monthly viewing time per user at each page level. This indicates the average exposure frequency for page layout level L. This represents the budgeted average price for level L. This indicates the advertiser's monthly budget on this platform. Indicates the total number of levels. This indicates the uniform price per thousand executions used. This represents the revenue per thousand effective ad impressions that has been used in actual historical transactions.

[0078] Define the objective function as maximizing the number of ad conversions within the budget limit, and define the constraints as follows:

[0079]

[0080]

[0081]

[0082] in, This represents the value of the advertising objective function. Indicates ad conversion rate, Indicates the number of times an ad is displayed at different levels;

[0083] The platform's total revenue is defined based on paid placement revenue and advertising sales revenue, and is expressed as follows:

[0084]

[0085]

[0086]

[0087] in, This indicates revenue from paid placements. These represent the advertising levels: no ads (paid) and ads (paid). Indicates advertising revenue. This represents the platform's total revenue;

[0088] The snapshot evaluation method is used to extract time snapshot records from the historical observation set for each level, including the level and position price at the time of the snapshot. Ad load at snapshot time The percentage of hierarchical selections observed at snapshot time And the average daily viewing time per person at the time of the snapshot. ;

[0089] The advertiser's budget is allocated, and the maximum number of ad impressions that can be purchased, the maximum number of ad impressions that can be afforded with the current remaining budget, and the actual number of impressions allocated are calculated for each tier. The remaining budget is then updated, as shown below:

[0090]

[0091]

[0092]

[0093] in, This represents the maximum number of impressions that can be afforded with the current remaining budget, where R represents the initial remaining budget. As initial value This indicates the uniform price per thousand used during snapshot execution. This indicates the maximum number of images that can be purchased, and can be expressed as follows: As initial value This indicates the actual number of views allocated. Indicates the updated remaining budget;

[0094] The synchronized snapshot shows the following: ad revenue, placement revenue, total platform revenue, and advertiser's expected total conversions, represented as follows:

[0095]

[0096] in, This indicates the advertiser's expected total conversions;

[0097] Based on the collection of time snapshot records, the total platform revenue for each snapshot record is calculated and the maximum value is selected as the snapshot index. The actual allocated impressions are used as the recommended ad allocation, and the corresponding calculated recommended ad revenue, recommended ad revenue, and advertiser's expected total conversion are used as the expected results.

[0098] By defining ad load and combining it with frequency adjustment schemes, the technical solution effectively utilizes business data to support frequency and supply optimization, while avoiding the negative impact of overloaded ad delivery on user experience. The combination of technologies ensures a balance between ad resource utilization efficiency and dynamic budget allocation.

[0099] By using ad placement price and ad load as dual axes, and combining neighboring observation points and empirical proportions with historical sample data, a linear interpolation selection model is designed. Compared with the traditional method of directly selecting a scheme using a single variable, this combination of technologies significantly improves the complexity modeling of selection probability.

[0100] The impact of frequency and price adjustments on advertiser selection behavior, as well as changes in resource supply scale, are captured flexibly through linear interpolation and combined with budget allocation rules to calculate an adjusted unified execution price that conforms to actual economic constraints. Ad display allocation is optimized, and causal effect verification and utilization deviation screening are combined to ensure that the optimization of the solution is not only from the perspective of conversion rate, but also from the optimal solution of global supply and demand matching and economic benefit improvement. This solution design with dual-axis verification screening has non-obvious robustness and value.

[0101] By segmenting historical real data through snapshot records and generating time segments for frequency quantitative analysis, the interference of data noise on prediction results is significantly reduced. By selecting the maximum benefit point as the execution plan through snapshot index, the subjectivity problem of weight assumptions in traditional optimization is solved, ensuring that the actual adjusted results can return to the historical database to form a closed loop, while dynamically ensuring the optimization effect of the combination of advertising scale and budget in the next cycle.

[0102] S2 measures the utilization rate of ad placement resources currently used by advertisers and updates the total revenue of the platform. It uses historical data of each level as discrete utilization points to find the point with the highest corresponding revenue value as the target utilization rate. It enumerates the price and frequency pairs of historical observations for each level, interpolates to calculate the probability of the advertiser's selection, re-estimates the budget-averaged price and execution price, updates the total revenue of the platform, and calculates the deviation between the predicted parameter combination and the target utilization rate.

[0103] Preferably, the point with the highest corresponding revenue value is selected as the target utilization rate. For each level, historical real-observed price and frequency pairs are enumerated, and the advertiser's selection probability is calculated by interpolation. The budget-averaged price and execution price are re-estimated, the platform's total revenue is updated, and the deviation between the predicted parameter combination and the target utilization rate is calculated, including...

[0104] Based on the expected results and the resource allocation chosen by the advertiser, the utilization rate of the ad placement resources currently used by the advertiser is measured, and the platform's total revenue is updated and calculated, as follows:

[0105]

[0106]

[0107] in, Indicates resource consumption utilization rate, This represents the actual amount of data displayed at time t. This represents the maximum monthly sales display volume for time level L. This represents the total revenue of the platform at level L, representing the update time t. This indicates revenue from paid placements in recommended sections. This indicates revenue from referral advertising;

[0108] The historical data of each level is analyzed according to discrete utilization rate points. The revenue contribution of each utilization rate point is calculated, and the corresponding average revenue value is selected from the set of discrete utilization rate points as the target utilization rate, represented as follows:

[0109]

[0110]

[0111]

[0112] in, This represents the average revenue over a historical period corresponding to the utilization rate u in level L. This represents the number of time points in the history where the utilization rate u of level L has appeared. Indicates time Total revenue of the tiered platform This represents the optimal target utilization rate of level L;

[0113] For each level, enumerate the price and frequency pairs of historical real observations, represented as:

[0114]

[0115] in, This indicates a price and frequency pair. This represents the set of historical listing prices for each tier. This represents the set of historical exposure frequencies for each level;

[0116] Based on the ad placement selection function, the probability of the advertiser's selection is calculated by interpolation, and the supply ceiling after the tier adjustment is simulated, as follows:

[0117]

[0118]

[0119]

[0120] in, This represents the probability of an advertiser choosing the current tier L, price p, and exposure frequency f. express function, This represents the average viewing time per person at level L. Indicates the number of days in the statistical period. This indicates the average daily viewing time. This represents the adjusted supply ceiling of level L under candidate combinations of price p and frequency f.

[0121] Based on the adjusted supply forecast, the budget-averaged price and execution price are re-estimated, and the advertiser's actual spending allocation under the adjusted parameters is calculated according to the budget allocation rules. The remaining budget is updated incrementally until it is exhausted, and the platform's total revenue is updated. The deviation between the predicted parameter combination and the target utilization rate is calculated and expressed as:

[0122]

[0123]

[0124]

[0125]

[0126]

[0127]

[0128]

[0129] in, This indicates the price required based on the updated advertiser budget amortization. This represents the updated, uniformly executed price per thousand impressions, which is the final price the platform charges advertisers. This indicates the adjusted number of impressions after the advertiser's budget is allocated to tier L. This represents the revenue generated from paid placements based on the current tier (L), price (p), and exposure frequency (f). This represents the advertising revenue based on the current tier L, price p, and impression frequency f. This represents the dynamic combination (L, p, f) with respect to the platform's total revenue. This indicates the deviation between the actual utilization rate and the target utilization rate of the dynamic combination.

[0130] By measuring resource consumption utilization and updating the platform's total revenue, the technical solution calculates resource utilization efficiency in real time based on the ratio between the advertiser's actual impressions and the maximum available resources. It determines the target utilization rate by evaluating the revenue contribution of grouped and discretized utilization rates.

[0131] By enumerating the combination of historical price and frequency and comprehensively analyzing the impact on the selection probability model, the technical solution innovatively combines price and frequency to form a resource regulation mechanism driven by historical real data. At the same time, it combines the interpolation calculation of the selection probability model to accurately capture the advertiser's behavioral tendencies.

[0132] By dynamically adjusting supply forecasts and budget-allocated prices, the platform's supply and demand balance is optimized. The solution extends the ability to adjust dynamic prices and frequency combinations across cycles, including building a progressively updated resource allocation mechanism through supply forecasts to respond to advertisers' needs in real time and avoid situations of insufficient or excessive supply.

[0133] By calculating the deviation between the parameter combination (price, frequency, and placement allocation) and the target utilization rate, the solution completes a closed loop of supply and demand equilibrium in the optimization logic. By combining the tiered revenue calculation of placement fees and advertising sales revenue, the solution clearly depicts the two-way source logic of revenue from the platform level.

[0134] S3 calculates the average conversion rate and contrast effect value based on high-frequency samples and regular samples respectively, and outputs a set of candidate frequencies that meet the target utilization deviation limit by combining dual gating to screen candidate schemes.

[0135] Preferably, the average conversion rate and contrast effect value are calculated based on high-frequency samples and regular samples respectively. Combined with dual-gating screening of candidate schemes, a set of candidate frequencies that meet the target utilization deviation limit is output, including:

[0136] Based on the historical exposure frequency set of each level, a high-frequency threshold is calibrated according to historical experience, and exposure frequencies above the high-frequency threshold are defined as high-frequency samples, while exposure frequencies below the high-frequency threshold are defined as regular samples.

[0137] The average conversion rate was calculated based on both high-frequency and regular samples, and the contrast effect value was calculated and expressed as follows:

[0138]

[0139]

[0140]

[0141] in, This represents the average conversion rate of high-frequency samples. Represents a high-frequency sample set. This represents the ad conversion rate at time t. This represents the average conversion rate of a standard sample. Represents a regular sample set. This represents the average difference in conversion rate improvement between high-frequency and regular-frequency conversions, serving as a comparative effect value.

[0142] Combining dual-gating screening of candidate schemes—including eliminating combinations that do not meet the target utilization rate and eliminating causal invalid frequency values—outputs a set of candidate frequencies that meet the target utilization rate deviation limit, represented as:

[0143]

[0144]

[0145] in, This represents the set of candidate frequencies that meet the target utilization deviation limit. This indicates the maximum permissible deviation from the target utilization range, determined based on historical experience. This represents the final set of frequencies that pass both causal and utilization verification. This represents the minimum contrast effect value.

[0146] By calibrating high-frequency thresholds and classifying samples based on the historical exposure frequency set of each level, the grouping criteria for high-frequency and regular samples are effectively determined through empirical data. This process avoids the risk of subjectively setting thresholds and directly uses the historical frequency distribution state to dynamically define high-frequency thresholds.

[0147] By calculating the average conversion rate of high-frequency and regular samples and comparing the conversion rate improvement benefits, the potential impact of high-frequency advertising on conversion rate benefits was further explored. The average improvement difference calculated between high-frequency and regular samples was used as the comparison benefit value. This not only directly determines whether high-frequency exposure leads to a significant increase in conversion rate, but also preliminarily verifies that high frequency is an important causal factor for improving advertising effectiveness.

[0148] By combining dual gating to screen candidate frequency sets and introducing target utilization deviation and causal verification constraints, the technical solution ensures the global robustness of frequency adjustment decisions. Specific effects include: limiting target utilization deviation to eliminate frequency combinations that are inefficient in resource use or deviate from the target, avoiding additional resource waste caused by high-frequency optimization; and limiting the minimum contrast effect value to verify causal effects, eliminating frequency values ​​that meet the utilization deviation requirements but do not significantly improve conversion efficiency, ensuring that the solution always aims for maximum effectiveness.

[0149] By eliminating frequency combinations with utilization bias and causal invalid pairs, a candidate frequency set that satisfies multiple constraints is generated. By introducing the minimum contrast effect value to eliminate invalid combinations in two-way verification, the scheme ensures that the selected frequencies remain stable in the adjustment of budget allocation and resource supply scale, while always operating towards the optimal point of resources and benefits.

[0150] S4, substitute each frequency into the calculation to update the platform's total revenue, determine the corresponding price combination, and calculate the difference between the actual utilization rate and the expected resource utilization rate.

[0151] Preferably, for each frequency, the total revenue of the platform is calculated and updated, and the corresponding price combination is determined, including:

[0152] The candidate frequency set is sorted according to the platform's total revenue contribution index, and the platform's total revenue is updated by substituting each frequency into the calculation. The corresponding price combination is determined, and the final frequency adjustment value and corresponding revenue price at the tier are determined.

[0153] By sorting the candidate frequency set according to the platform's total revenue contribution index and updating the platform's total revenue, the solution design establishes a direct correlation between frequency selection behavior data and total revenue contribution. It selects the combination with the highest economic benefits from the candidate set, and completes hierarchical revenue optimization by determining the final frequency adjustment value and combining it with the corresponding price combination. Under the goal of maximizing the platform's overall revenue, the control action is directly reduced to an executable frequency price adjustment, avoiding excessive complexity caused by multiple parameter combinations.

[0154] Furthermore, the difference between the actual utilization rate and the expected resource utilization rate is calculated and compared, including:

[0155] Based on the frequency and price combination of execution, the actual utilization rate is calculated and compared with the expected resource utilization rate. The sum of the mean and standard deviation of the historical differences is used as the comparison threshold. If the actual calculated difference is less than or equal to the comparison threshold, the comparison is considered successful.

[0156] After adjusting the frequency and price combination, the difference between the actual utilization rate and the expected resource utilization rate is calculated and compared. Dynamic verification logic is added after the plan is executed. This process effectively supplements the verification of the execution effect of the frequency and price plan. In particular, after the frequency plan is actually configured, the platform can dynamically track the operation effect of resource allocation.

[0157] This embodiment also provides an artificial intelligence-based intelligent outdoor advertising delivery optimization system, including:

[0158] Data acquisition module: Collects basic data on outdoor advertising and defines the hierarchy; collects and calibrates the four neighboring observation points of the current working point according to the price and load axes.

[0159] Conversion module: Constructs a selection function by using the linear interpolation ratio of price and ad load, calculates the selection probability of different levels, and sets maximizing the number of ad conversions as the conversion objective function;

[0160] Total Revenue Calculation Module: With the platform's total revenue as the core objective, it uses a snapshot evaluation method to extract time snapshot records for verification and optimization of total revenue;

[0161] Budget allocation module: Allocates advertiser budgets and updates remaining budgets in real time;

[0162] Historical data analysis module: Calculates the target utilization rate based on the evaluation of advertising resource consumption and utilization rate, combined with hierarchically discretized historical utilization rate data;

[0163] Frequency Enumeration Module: Enumerates historical price and frequency combinations, interpolates to calculate the advertiser's selection probability, and re-outputs the budget-averaged price and execution price;

[0164] Conversion rate calculation module: Calculates the contrast effect value by comparing the average conversion rates of high-frequency and regular samples, and combines dual gating to screen the frequency set that meets the resource optimization requirements;

[0165] Utilization verification module: Calculates the total revenue contribution corresponding to frequency adjustment, generates recommended price combinations, and verifies and calibrates the deviation between actual utilization rate and resource utilization rate.

[0166] This embodiment also provides a computer device applicable to the intelligent outdoor advertising delivery optimization method based on artificial intelligence, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the intelligent outdoor advertising delivery optimization method based on artificial intelligence as proposed in the above embodiment.

[0167] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0168] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the intelligent outdoor advertising delivery optimization method based on artificial intelligence as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0169] In summary, this invention flexibly captures and calculates an adjusted unified execution price that conforms to actual economic constraints through linear interpolation and combined with budget allocation rules. It optimizes ad display allocation and combines causal effect verification and utilization deviation screening to ensure that the optimization of the solution is not only from the perspective of conversion rate, but also from the perspective of selecting the optimal solution for global supply and demand matching and economic benefit improvement. By selecting the maximum revenue point as the execution solution through snapshot index, it solves the subjectivity problem of weight assumptions in traditional optimization and ensures that the actual adjusted results can be returned to the historical database to form a closed loop. By measuring resource consumption utilization and updating the platform's total revenue, the technical solution calculates resource utilization efficiency in real time based on the ratio between the advertiser's actual display volume and the maximum available resource volume. By eliminating the frequency combinations of utilization deviation and causal invalid pairs, it generates a candidate frequency set that meets multiple constraints.

[0170] It should be noted that 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An artificial intelligence-based outdoor advertising intelligent delivery optimization method, characterized in that, The method comprises the following steps: Defining the outdoor advertising level, collecting advertising basic data, defining four adjacent observation points of the current working point, linearly interpolating the proportion of price and load to obtain the selection function and output the selection probability of different levels, making the number of advertising conversions maximum as the advertising objective function, defining the total revenue of the platform, extracting the time snapshot record of each level by using the snapshot evaluation method, allocating the budget of the advertiser, updating the remaining budget, measuring the consumption utilization rate of the current advertiser's placement resource, updating and calculating the total revenue of the platform, finding the point with the maximum corresponding revenue value as the target utilization rate according to the discrete utilization rate points of the historical data of the level, enumerating the real observation price and frequency pairs of each level, interpolating and calculating the selection probability of the advertiser, re-estimating the budget allocation price and the execution price, updating the total revenue of the platform, and calculating the deviation of the predicted parameter combination and the target utilization rate; According to the high-frequency sample and the conventional sample, the average conversion rate and the comparative effect value are calculated respectively, the candidate scheme is screened by combining double gating, and the candidate frequency set meeting the target utilization rate deviation limit is output; The platform total revenue is updated by substituting each frequency, and the corresponding price combination is determined, and the actual utilization rate and the expected resource utilization rate are compared by difference. The method comprises the following steps: 2.The AI-based outdoor advertisement intelligent delivery optimization method of claim 1, wherein: According to the ratio of the total number of advertising displays in the month and the advertising viewing time in the month of different levels as the advertising load; Defining four adjacent observation points of the current working point, including the nearest value not greater than the current price and the nearest value not less than the current price in the set of historical actual appearing placement pricing prices, and the nearest value not greater than the current advertising load and the nearest value not less than the current advertising load in the set of historical actual appearing advertising load; For each level, the sample average proportion of the four adjacent observation points is directly calculated from the historical data as the four-angle empirical proportion; Linearly interpolating the proportion of price and load to obtain the selection function and output the selection probability of different levels; Based on the selection probability of different levels, the monthly maximum sales display quantity of each level is calculated, the unified execution price is determined under the supply and budget caliber, and it is ensured that the price is not higher than the historical transaction level and the budget is digested according to the supply scale, and the unified thousand-time price of execution is calculated; Defining the maximum number of advertising conversions as the advertising objective function under the premise that the budget is not exceeded, and defining the constraint; Defining the total revenue of the platform based on the placement payment income and the advertising sales income; The time snapshot record of each level in the historical observation set is extracted by using the snapshot evaluation method, including the level placement price at the snapshot time, the advertising load at the snapshot time, the level selection proportion observed at the snapshot time, and the per capita daily viewing time at the snapshot time. ​ Allocating the budget of the advertiser, calculating the maximum number of purchasable advertisement displays, the maximum number of affordable advertisement displays under the current remaining budget, and the actual allocated displays for each level, and updating the remaining budget; Synchronously updating the advertisement revenue, the position payment revenue, the total platform revenue, and the total expected conversion of the advertiser under the snapshot; According to the set of time snapshot records, calculating the total platform revenue of each snapshot record and selecting the maximum value as the snapshot index, and taking the actual allocated display quantity as the recommended advertisement allocation quantity, and taking the calculated recommended position payment revenue, recommended advertisement revenue, and total expected conversion of the advertiser as the expected result. 3.The AI-based outdoor advertisement intelligent delivery optimization method of claim 2, wherein: The point with the maximum corresponding revenue value is taken as the target utilization rate, the historical real observed price and frequency pairs of each level are enumerated, the selection probability of the advertiser is calculated by interpolation, the budget allocation price and the execution price are re-estimated, the total platform revenue is updated, and the deviation of the predicted parameter combination from the target utilization rate is calculated, including, Based on the expected result, the position resource consumption utilization rate of the current advertiser is measured according to the resource allocation selected by the advertiser, and the total platform revenue is updated and calculated; The historical data of the level is analyzed according to the discrete utilization rate points, the revenue contribution of each utilization rate point is calculated, the corresponding average revenue value is calculated, and the point with the maximum corresponding revenue value is found from the discrete utilization rate point set as the target utilization rate; For each level, the historical real observed price and frequency pairs are enumerated; According to the position selection function, the selection probability of the advertiser is calculated by interpolation, and the supply upper limit after the level adjustment is simulated; Based on the adjusted supply prediction, the budget allocation price and the execution price are re-estimated, and the actual allocation of the advertiser under the adjusted parameter condition is calculated according to the budget allocation rule, the remaining budget is updated step by step until the budget is exhausted, the total platform revenue is updated, and the deviation of the predicted parameter combination from the target utilization rate is calculated. 4.The AI-based outdoor advertisement intelligent delivery optimization method of claim 3, wherein: The average conversion rate and the comparative effect value are calculated according to the high-frequency samples and the regular samples respectively, the candidate frequency set that meets the target utilization rate deviation limit is output by combining double-gated screening of candidate schemes, including, Based on the historical exposure frequency set of each level, the high-frequency threshold is calibrated according to historical experience, and the exposure frequency greater than the high-frequency threshold is defined as a high-frequency sample, and the exposure frequency lower than the high-frequency threshold is defined as a regular sample; The average conversion rate is calculated according to the high-frequency samples and the regular samples respectively, and the comparative effect value is calculated; The candidate frequency set that meets the target utilization rate deviation limit is output by combining double-gated screening of candidate schemes, including removing combinations that do not meet the target utilization rate, and removing frequency values that are causally ineffective. 5.The AI-based outdoor advertisement intelligent delivery optimization method of claim 4, wherein: The actual utilization rate and the expected resource utilization rate are compared by difference, including, ​ 6.The AI-based outdoor advertisement intelligent delivery optimization method of claim 5, wherein: ​ Based on the frequency and price combination, the actual utilization rate is compared with the expected resource utilization rate, and the sum of the mean and standard deviation of the historical difference value is used as the comparison threshold. If the actual calculated difference value is less than or equal to the comparison threshold, it means that the comparison is passed. 7.The AI-based outdoor advertisement intelligent delivery optimization method of claim 1, wherein: The method comprises the following steps of: The method comprises the following steps of: The method comprises the following steps of:

8. An artificial intelligence-based outdoor advertising intelligent delivery optimization system based on any one of the artificial intelligence-based outdoor advertising intelligent delivery optimization methods of claims 1-7. The method comprises the following steps of: The method comprises the following steps of: The data collection module collects outdoor advertising basic data and defines levels, and collects and calibrates four adjacent observation points of the current work point according to the price and load axes; The conversion module constructs a selection function by linear interpolation of the price and advertising load, calculates the selection probability of different levels, and sets the maximum number of advertising conversions as the conversion objective function; The total revenue calculation module takes the platform total revenue as the core target, uses the snapshot evaluation method to extract time snapshot records for verification and optimization of total revenue; The budget allocation module allocates the budget of the advertiser and updates the remaining budget in real time; The historical data analysis module calculates the target utilization rate according to the advertising resource consumption utilization rate evaluation and the level discretization historical utilization rate data; The frequency enumeration module enumerates the historical price and frequency combination, interpolates the advertising owner selection probability and re-outputs the budget allocation price and execution price; The conversion rate calculation module calculates the comparison effect value by comparing the average conversion rates of high-frequency and regular samples, and combines double-gated screening to select the frequency set that meets the resource optimization requirements; 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The utilization rate verification module calculates the total revenue contribution of the frequency adjustment, generates a recommended price combination, and verifies and calibrates the deviation between the actual utilization rate and the resource utilization rate.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: The processor executes the computer program to realize the steps of the outdoor advertising intelligent optimization method based on artificial intelligence according to any one of claims 1-7. The computer program is executed by the processor to realize the steps of the outdoor advertising intelligent optimization method based on artificial intelligence according to any one of claims 1-7.

Images