Cross-media advertisement putting method and system

Abstract

The invention discloses a cross-media advertisement putting method and system, and relates to the technical field of advertisement putting, and the method comprises the steps: collecting advertisement putting data according to different advertisement media, building a response relation function between the advertisement putting amount and a conversion effect, and determining a cross-media putting response relation value, and calculating correlation coefficients to determine a synergistic effect value of each pair of advertisement media combinations. The method disclosed by the invention not only reveals the actual linkage synergy relationship between different advertisement media combinations through quantification of the synergy effect, but also can monitor the release marginal return of each media in real time through marginal benefit analysis of the first derivative of the response function, discover the release saturation point in time, and restrain the introduction of the optimization method, thereby improving the efficiency of the advertisement media. Therefore, on the premise that the budget distribution process meets the total budget and the actual delivery capacity of each media, the global optimal delivery amount distribution scheme can be obtained through derivation and simultaneous solution of the target function.

Description

Technical Field

[0001] This invention relates to the field of advertising delivery technology, and in particular to a cross-media advertising delivery method and system. Background Technology

[0002] With the rapid development of digital media and information technology, advertising channels are becoming increasingly diversified. Advertisers often need to allocate resources across multiple media platforms, including television, radio, the internet, social media, and mobile applications, when conducting marketing campaigns. As data-driven advertising decisions gradually become mainstream, advertisers are focusing on how to accurately evaluate the conversion effect of advertising campaigns and achieve optimal budget allocation through big data analysis and modeling. Existing technologies include some solutions that attempt to establish a response relationship between advertising volume and conversion effect through statistical analysis or machine learning methods. However, existing cross-media advertising technologies neglect the synergistic effect between advertising media and fail to effectively quantify the improvement of conversion effect by different media combinations. This results in a lack of scientific basis for budget allocation and makes it difficult to maximize the overall advertising effectiveness. Secondly, existing attribution analysis is mostly based on simple rules or linear models, which makes it difficult to accurately reflect the user's real conversion contribution in the multimedia path. Especially in scenarios with complex user behavior paths and diverse media combinations, the attribution results are prone to distortion and it is difficult to take into account both marginal benefits and synergistic effects. Summary of the Invention

[0003] In view of the aforementioned existing problems, the present invention is proposed.

[0004] Therefore, the present invention provides a method and system for cross-media advertising delivery; The existing cross-media advertising technologies neglect the synergistic effects between advertising media and fail to effectively quantify the conversion improvement of different media combinations. This results in a lack of scientific basis for budget allocation and makes it difficult to maximize the overall advertising effectiveness. Secondly, existing attribution analysis is mostly based on simple rules or linear models, which makes it difficult to accurately reflect the user's real conversion contribution in the multimedia path. Especially in scenarios with complex user behavior paths and diverse media combinations, the attribution results are prone to distortion and it is difficult to take into account the marginal benefits and synergistic effects.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a cross-media advertising delivery method, comprising: Based on the collected advertising data from different advertising media, a response relationship function between advertising volume and conversion effect is established to determine the cross-media placement response relationship value. The correlation coefficient is calculated to determine the synergistic effect value of each pair of advertising media combinations. The first derivative of the response relationship function is used as the marginal benefit value. The budget is allocated by using a constrained optimization method based on the comprehensive synergistic effect value. Lagrange multipliers are introduced to calculate the partial derivative of the advertising volume for each media to determine the optimal advertising volume for each media. Collect user behavior data during the advertising campaign period, construct a directed graph, embed the paths in the directed graph using the Node2Vec algorithm, and perform path clustering on the resulting advertising media node vectors; The conversion rates of all paths within a cluster are statistically analyzed, the causal contributions of different media to the cluster are analyzed, all advertising media subsets are enumerated, the conversion rates of each subset and the conversion rates after adding advertising media m are calculated, the attribution allocation is performed using Shapley value theory, the attribution ratio is calculated, the distribution ratio of all user paths in each cluster is statistically analyzed, and the optimal delivery volume is adjusted based on attribution. Distortion detection is performed based on attribution percentage, and reports are generated based on the final delivery volume data and actual conversions, and then transmitted in encrypted form.

[0006] As a preferred embodiment of the cross-media advertising placement method of the present invention, the steps include: determining the cross-media placement response relationship value, calculating the correlation coefficient to determine the synergistic effect value of each pair of advertising media combinations, using the first derivative of the response relationship function as the marginal benefit value, and using a constrained optimization method to allocate the budget based on the comprehensive synergistic effect value, including... For each advertising medium, a parameterizable Logistic function is used, and an adaptive parameter update mechanism is introduced to establish a response relationship function between advertising volume and conversion effect; Training is performed using the training set data, and the loss is calculated based on the calibrated training data. The parameters of the response function are updated online using the recursive least squares (RLS) method, and the parameters are automatically adjusted using the RLS algorithm to minimize the computational loss when updating the data. For each pair of advertising media combinations, the Pearson correlation coefficient of the normalized series of historical ad placements is calculated to determine the synergistic effect value of each pair of advertising media combinations, and the sum of synergistic effects is calculated for all advertising media combinations. For each advertising medium, calculate the first derivative of the response function under the current delivery volume as the marginal benefit value; The saturation threshold is based on the sum of the mean and standard deviation, using historical data. If the marginal benefit value is greater than the saturation threshold, it is determined that the corresponding media needs to continue to increase its investment. Based on the combined response values ​​of single media, synergistic effects, and saturation detection results, a constrained optimization method is adopted for budget allocation. The constraints are defined as follows: the sum of the ad spend of all media must not exceed the total budget; the ad spend of each media must not be negative; if the marginal benefit value is less than or equal to the saturation threshold, it is determined that the corresponding media has reached the saturation point and no further budget is allocated. A Lagrange multiplier λ is introduced to construct a Lagrange function. The partial derivative of the ad spend of each media is calculated and set to zero to obtain the optimality conditions. The optimal advertising volume is determined by combining the optimality conditions and budget constraints for the number of ad placements across all advertising media.

[0007] As a preferred embodiment of the cross-media advertising delivery method of the present invention, the steps of collecting user behavior data during the advertising delivery period, constructing a directed graph and embedding the paths in the directed graph using the Node2Vec algorithm, and performing path clustering on the obtained advertising media node vectors include: Collect all user behavior sequence data during the advertising campaign period, including: user unique identifier, advertising media m for each ad exposure, exposure timestamp, and behavior type; For each user, all their ad exposure events are arranged in chronological order to form an ordered path sequence; Merge all user path sequences to construct a directed graph, where node V is the number of all advertising media, edges represent user transitions from one media to another, and edge weights are the frequency of that transition in all user paths; Node attributes include: media ID, the number of times the media is first encountered in all paths, and the number of times it is included in the final conversion path; The Node2Vec algorithm is used to embed the directed graph path to obtain a low-dimensional vector representation of each advertising media node; For each user path, the path embedding vector is calculated as the mean of the embedding vectors of each node; K-means clustering is performed on all user path embedding vectors, and the number of clusters K is determined by the silhouette coefficient method.

[0008] As a preferred embodiment of the cross-media advertising delivery method of the present invention, the following steps are included: calculating the conversion rate of each subset and the conversion rate after adding advertising media m, performing attribution allocation based on Shapley value theory, calculating the attribution ratio, statistically analyzing the distribution ratio of all user paths in each cluster, and adjusting the optimal delivery volume based on attribution, including... For each path cluster, the conversion rate of all paths within the cluster is calculated based on the ratio of converted users to total users. For each advertising medium m, the conversion rate of paths containing the corresponding advertising medium within the cluster is calculated, and the causal contribution of different media under the cluster is calculated. For each user path, enumerate all subsets of advertising media, and calculate the conversion rate of each subset and the conversion rate after adding advertising media m; Attribution assignment is performed using Shapley value theory, and the sum of the average Shapley values ​​of all media is calculated as the total attribution value of the corresponding cluster. Then, the attribution proportion of advertising media m in cluster k is calculated. The distribution ratio of all user paths in each cluster is statistically analyzed, the global attribution weight of each advertising medium is calculated, and the optimal delivery volume is adjusted based on attribution.

[0009] As a preferred embodiment of the cross-media advertising delivery method of the present invention, the step of distortion detection based on attribution ratio includes: For each advertising medium, the attribution percentage is generated as a time series. The mean and standard deviation are calculated using a sliding window to detect attribution distortion. If the attribution percentage deviates from the mean by more than twice the standard deviation, it is judged as attribution distortion.

[0010] As a preferred embodiment of the cross-media advertising delivery method of the present invention, the step of collecting advertising delivery data according to different advertising media includes: Collect and organize historical advertising data, including: the number of ads placed for each advertising medium, the actual conversions for each advertising medium, the timestamp of the ad placement, user group characteristics, and the combination of advertising media for each placement; The number of deliverables and conversions are normalized, and the user group characteristics are converted into numerical features using one-hot encoding.

[0011] As a preferred embodiment of the cross-media advertising delivery method of the present invention, the step of generating a report based on the final delivery volume data and actual conversion numbers, and transmitting it in encrypted form, includes: The system monitors the actual conversion rate of each media in real time and records the corresponding final delivery volume data. It generates conversion reports using a report generation tool, encrypts the conversion reports based on the TLS encryption protocol, and protects the transmitted data with encryption using the AES encryption algorithm.

[0012] Secondly, the present invention provides a cross-media advertising delivery system, comprising, The response function module collects data on advertising across different media and establishes a response relationship function between advertising volume and conversion rate. The budget allocation optimization module calculates the marginal benefit value based on the first derivative of the response relationship function, and introduces Lagrange multipliers to calculate the partial derivative with respect to the advertising amount to determine the optimal advertising amount for each advertising media. The user behavior analysis module collects user behavior data during the advertising campaign period, uses the Node2Vec algorithm to embed directed graph paths, and performs path clustering on the resulting advertising media node vectors. The conversion rate statistics module performs statistical analysis on the path clustering results, calculates the conversion rate of all paths within a cluster, and analyzes the causal contribution of different advertising media within each cluster. The attribution adjustment module enumerates all subsets of advertising media, calculates the conversion rate of each subset, calculates the attribution allocation of advertising media based on the Shapley value theory, counts the attribution ratio of advertising media, and performs attribution-driven adjustments to the optimal delivery volume. The report generation module performs distortion detection based on the attribution ratio results, generates statistical reports of the final delivery volume data and actual conversion data, and encrypts the transmission of the reports.

[0013] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein the computer program, when executed by the processor, implements any step of the cross-media advertising delivery method as described in the first aspect of the present invention.

[0014] 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 cross-media advertising delivery method as described in the first aspect of the present invention.

[0015] The beneficial effects of this invention are as follows: By quantifying the synergistic effect, it not only reveals the actual synergistic relationship between different advertising media combinations, but also, through marginal benefit analysis of the first derivative of the response function, it can monitor the marginal return of each media in real time and promptly identify the saturation point of the placement. The introduction of the constraint optimization method ensures that the budget allocation process can meet the total budget and the actual placement capacity of each media. At the same time, by differentiating the objective function and solving the simultaneous equations, it can obtain the globally optimal placement allocation scheme. By collecting user behavior data during the advertising placement period and transforming it into a directed graph structure, it can comprehensively capture the real transfer paths of users between different advertising media. By embedding the directed graph using the Node2Vec algorithm, it can effectively extract the complex relationships and potential path patterns between advertising media. Based on clustering, it can statistically analyze the conversion rate of various paths and the causal contribution of advertising media, revealing the actual influence of different media in different conversion paths in more detail, avoiding the error caused by average attribution. Combining Shapley value theory for attribution allocation, it quantifies the marginal contribution of each advertising media in different path patterns, overcoming the problem of neglecting the path order and combination in traditional attribution methods. Attached Figure Description

[0016] 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.

[0017] Figure 1 This is a flowchart illustrating the cross-media advertising delivery method in Example 1.

[0018] Figure 2 This is a schematic diagram of the cross-media advertising delivery system in Example 1. Detailed Implementation

[0019] 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.

[0020] 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.

[0021] 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.

[0022] Example 1, referring to Figures 1 to 2 This is the first embodiment of the present invention, which provides a cross-media advertising delivery method, including the following steps: S1. Collect advertising data from different advertising media, establish a response relationship function between advertising volume and conversion effect, determine the cross-media placement response relationship value, calculate the correlation coefficient to determine the synergistic effect value of each pair of advertising media combinations, use the first derivative of the response relationship function as the marginal benefit value, and use the constrained optimization method to allocate the budget based on the comprehensive synergistic effect value. Introduce Lagrange multipliers to calculate the partial derivative of the advertising volume for each media to determine the optimal placement volume for the advertising media. Preferably, advertising data is collected based on different advertising media, including: Collect and organize historical advertising data, including: the number of ads placed for each advertising medium (unit: thousands of impressions), the actual number of conversions for each advertising medium (unit: times), the timestamp of the ad placement, user group characteristics (such as age, gender, region), and the combination of advertising media for each placement (such as "TV + social media"). The number of deliverables and conversions were normalized, and one-hot encoding was used to convert them into numerical features based on user group characteristics.

[0023] By collecting and organizing historical advertising data and normalizing the number of placements and conversions, we can eliminate the impact of scale differences between different advertising media, making subsequent analysis more fair and comparable. By using one-hot encoding to transform user group characteristics into numerical features, we can effectively improve the usability of data and the expressive power of the model, enabling subsequent data mining and modeling processes to better capture the differences in advertising effectiveness among different user groups.

[0024] Furthermore, the cross-media placement response relationship value is determined, the correlation coefficient is calculated to determine the synergistic effect value of each pair of advertising media combinations, the first derivative of the response relationship function is used as the marginal benefit value, and the budget is allocated using a constrained optimization method based on the comprehensive synergistic effect value, including... For each advertising medium, a parameterizable Logistic function is used, and an adaptive parameter update mechanism is introduced to establish a response relationship function between advertising volume and conversion effect, expressed as: ; in This represents the response relationship value between time t, user group characteristics u, and the number of ad placements for the m-th advertising media. This represents the maximum number of conversions for advertising medium m in historical data. This represents the number of ads placed on the m-th media. , and These represent the bias term, the number of deliveries coefficient, and the time coefficient, respectively. This indicates the calculation of the transpose of user characteristic coefficients; Training is performed using the training set data, and the loss is calculated based on the calibrated training data. The parameters of the response function are updated online using the recursive least squares (RLS) method, and the parameters are automatically adjusted using the RLS algorithm to minimize the computational loss when updating the data. For each advertising media combination, the Pearson correlation coefficient of the normalized historical ad spend series is calculated to determine the synergistic effect value of each advertising media combination. The sum of the synergistic effects for all advertising media combinations is then calculated and expressed as: ; in This represents the synergistic effect value of the i-th and j-th advertising media. This represents the relevance values ​​of the i-th and j-th advertising media. and Let i and j represent the number of times the advertisement is placed on the i-th and j-th advertising media, respectively. For each advertising medium, the first derivative of the response function under the current ad volume is calculated as the marginal benefit value, expressed as: ; ; Where z represents the linear combination of inputs to the Logistic function. This represents the marginal benefit value of the m-th advertising medium. This represents the response relationship value of the m-th advertising medium; The saturation threshold is based on the sum of the mean and standard deviation, using historical data. If the marginal benefit value is greater than the saturation threshold, it is determined that the corresponding media needs to continue to increase its investment. Based on the combined response values ​​of individual media, synergistic effects, and saturation detection results, a constrained optimization method is used for budget allocation. The constraints are defined as follows: the sum of the ad spend across all media must not exceed the total budget; the ad spend for each media must not be negative; and if the marginal benefit value is less than or equal to the saturation threshold, the corresponding media is considered to have reached its saturation point, and no further budget allocation is made. The objective formula for the constrained optimization is expressed as: ; in This represents the total expected conversions across all media combinations, where M represents the total number of advertising media. This represents the sum of the synergistic effects of all media combinations; Introducing the Lagrange multiplier λ, we construct the Lagrange function, expressed as: ; in Represent the Lagrange function, Let λ represent the total advertising budget, and λ represent the Lagrange multiplier. Taking the partial derivative of the advertising expenditure for each media and setting it to zero yields the optimality condition, expressed as:

[0025] ; in This represents the derivative of the single-media response function with respect to the amount of content delivered. This represents the derivative of the synergistic effect with respect to the amount of product delivered; The optimal advertising volume is determined by combining the optimality conditions and budget constraints for the number of ad placements across all advertising media.

[0026] By accurately modeling the response relationship of cross-media advertising, the system can dynamically reflect the actual impact of multi-dimensional factors such as advertising volume, time, and user groups on conversion results. By introducing an adaptive parameter update mechanism, the system can continuously optimize its response function parameters based on real-time data, ensuring that the model always keeps up with the latest market and user behavior changes and avoids model aging or failure. The quantification of synergy not only reveals the actual synergistic effect between different advertising media combinations, but also effectively identifies which media combinations can bring higher conversion rates, thus providing a scientific basis for the efficient allocation of resources. Using statistical measures such as the Pearson correlation coefficient, the calculation process of synergy is based entirely on historical data, ensuring its objectivity and traceability. By analyzing the marginal benefit of the first derivative of the response function, the marginal return on each media placement can be monitored in real time, and the saturation point of placement can be identified in a timely manner. This avoids wasting resources on placements with diminishing marginal benefits or even ineffective placements. The saturation threshold is set based on historical mean and standard deviation, exhibiting data-driven adaptive characteristics that can adapt to changes in different media and market environments. The introduction of constrained optimization methods enables the budget allocation process to maximize overall advertising conversion effects while meeting the total budget and the actual placement capacity of each media. The application of Lagrange multipliers ensures the strict execution of budget constraints. At the same time, by differentiating the objective function and solving the simultaneous equations, the globally optimal placement allocation scheme can be obtained. This achieves an organic unity of data-driven, dynamic adaptive, synergistic efficiency, and optimal resource allocation, significantly improving the return on investment and market response speed of advertising placements.

[0027] S2: Collect user behavior data during the advertising campaign period, construct a directed graph, embed the paths in the directed graph using the Node2Vec algorithm, and perform path clustering on the obtained advertising media node vectors. Preferably, user behavior data is collected during the advertising campaign period, a directed graph is constructed, and the Node2Vec algorithm is used to embed the paths in the directed graph. Then, path clustering is performed on the resulting advertising media node vectors, including... Collect all user behavior sequence data during the advertising campaign period, including: user unique identifier (such as encrypted ID), advertising media m for each ad exposure, exposure timestamp, and behavior type (exposure, click, conversion). For each user, all their ad exposure events are arranged in chronological order to form an ordered path sequence; Merge all user path sequences to construct a directed graph, where node V is the number of all advertising media, edges represent user transitions from one media to another, and edge weights are the frequency of that transition in all user paths; Node attributes include: media ID, the number of times the media is first encountered in all paths, and the number of times it is included in the final conversion path; The Node2Vec algorithm is used to embed the directed graph path to obtain a low-dimensional vector representation of each advertising media node. The Node2Vec parameters (such as walk step size and window size) are all determined using historical calibration values. For each user path, the path embedding vector is calculated as the mean of the embedding vectors of each node; K-means clustering is performed on all user path embedding vectors. The number of clusters K is determined by the silhouette coefficient method (each cluster center represents a typical conversion path pattern).

[0028] By collecting user behavior data during the advertising campaign period and transforming it into a directed graph structure, we can comprehensively capture the actual transfer paths of users across different advertising media, avoiding the bias of a single exposure or click perspective. Embedding the directed graph using the Node2Vec algorithm effectively extracts the complex relationships and potential path patterns between advertising media, ensuring that each advertising media node obtains a low-dimensional vector representation reflecting its actual role in the user conversion path. Performing K-means clustering on the embedded user path vectors helps discover and summarize various typical conversion path patterns, thus providing a structured foundation for subsequent attribution analysis and campaign optimization.

[0029] S3: Calculate the conversion rate of all paths within the cluster, analyze the causal contribution of different media in the cluster, enumerate all advertising media subsets, calculate the conversion rate of each subset and the conversion rate after adding advertising media m, combine Shapley value theory to perform attribution allocation, calculate the attribution ratio, calculate the distribution ratio of all user paths in each cluster, and adjust the optimal delivery volume based on attribution. Preferably, the conversion rate of each subset and the conversion rate after adding advertising media m are calculated. Attribution allocation is performed using Shapley value theory, attribution percentages are calculated, and the distribution ratio of all user paths in each cluster is statistically analyzed. Attribution-driven adjustments are then made to the optimal ad delivery volume, including... For each path cluster, the conversion rate of all paths within that cluster is calculated based on the ratio of converted users to total users. For each advertising medium *m*, the conversion rate of paths containing that advertising medium within the cluster is calculated, and the causal contribution of different media within the cluster is calculated, expressed as:

[0030] in This represents the causal contribution value of advertising medium m in cluster k. This represents the conversion rate of the path of advertising media m under cluster k. This represents the conversion rate of all paths within cluster k; For each user path, enumerate all subsets of advertising media, and calculate the conversion rate of each subset and the conversion rate after adding advertising media m; Attribution assignment is performed using Shapley score theory, and the sum of the average Shapley scores of all media is calculated as the total attribution value for the corresponding cluster. Then, the attribution percentage of advertising media m in cluster k is calculated, expressed as: ; ; ; in Let represent the Shapley value attribution assignment of media m in cluster k, and let represent the average marginal contribution of media m in this path pattern. This represents a subset of media S that does not contain media m. This represents the factorial of the number of media in subset S. This represents the factorial of the remaining media in the path after removing subset S and media m. This represents the factorial of the total number of media in the path. This represents the conversion rate of paths that contain only a subset S. This represents the conversion rate of the path containing subset S and media m. This represents the sum of the Shapley values ​​of all media in cluster k. This represents the average Shapley value of media m in cluster k, that is, the average Shapley value of media m across all paths in cluster k. This indicates the attribution percentage of media m under this path pattern; The distribution ratio of all user paths in each cluster is statistically analyzed, and the global attribution weight of each advertising medium is calculated. The optimal delivery volume is then adjusted based on attribution, as shown below: ; ; in This represents the global attribution weight of advertising medium m. This represents the proportion of paths in cluster k to the total number of paths. This represents the optimal placement volume for advertising media m. This represents the final delivery volume driven by attribution for media m.

[0031] By statistically analyzing the conversion rates and causal contributions of various advertising media across different paths based on clustering, this approach reveals the actual influence of different media on different conversion paths in greater detail. It avoids errors caused by average attribution. Combining this with Shapley value theory for attribution allocation allows for a fair and scientific quantification of the marginal contribution of each advertising media under different path patterns, overcoming the neglect of path order and combination issues in traditional attribution methods. By statistically analyzing the distribution ratio of all user paths in each cluster and adjusting advertising volume accordingly, attribution-driven resource optimization allocation can be achieved. This enables advertising budgets to be more accurately targeted at high-value media and paths, improving overall conversion efficiency and ROI. Finally, by synthesizing the attribution weights of each cluster, a reasonable global attribution weight can be assigned to each advertising media, ensuring that advertising decisions consider both the overall effect and the differences in user behavior patterns. This method can dynamically adapt to changes in the market and user behavior, continuously optimize advertising strategies, and achieve data-driven refined marketing.

[0032] S4 performs distortion detection based on attribution ratio, generates reports based on final delivery volume data and actual conversions, and transmits them in encrypted form. Preferably, distortion detection is performed based on attribution percentage, including: For each advertising medium, the attribution percentage is generated as a time series. The mean and standard deviation are calculated using a sliding window to detect attribution distortion. If the attribution percentage deviates from the mean by more than twice the standard deviation, it is judged as attribution distortion.

[0033] By creating a time series of the attribution percentage for each advertising medium and using a sliding window to calculate the mean and standard deviation for distortion detection, abnormal fluctuations or distortions in attribution allocation can be detected in a timely manner. This helps advertisers quickly locate potential data anomalies or market changes, improving the accuracy and reliability of attribution analysis.

[0034] Furthermore, reports are generated based on the final delivery volume data and actual conversion numbers, and transmitted in encrypted form, including... The system monitors the actual conversion rate of each media in real time and records the corresponding final delivery volume data. It generates conversion reports using a report generation tool, encrypts the conversion reports based on the TLS encryption protocol, and protects the transmitted data with encryption using the AES encryption algorithm.

[0035] By monitoring the actual conversion rate of each media in real time and recording the final delivery volume data, and automatically generating conversion reports through report generation tools, the efficiency of data processing and report output can be improved, reducing manual intervention and the probability of errors. By using TLS protocol and AES encryption algorithm to encrypt and protect the report data, the security and privacy of sensitive data during transmission can be effectively guaranteed, preventing data leakage and illegal tampering, and enhancing data compliance and customer trust.

[0036] This embodiment also provides a cross-media advertising delivery system, including, The response function module collects data on advertising across different media and establishes a response relationship function between advertising volume and conversion rate. The budget allocation optimization module calculates the marginal benefit value based on the first derivative of the response relationship function, and introduces Lagrange multipliers to calculate the partial derivative with respect to the advertising amount to determine the optimal advertising amount for each advertising media. The user behavior analysis module collects user behavior data during the advertising campaign period, uses the Node2Vec algorithm to embed directed graph paths, and performs path clustering on the resulting advertising media node vectors. The conversion rate statistics module performs statistical analysis on the path clustering results, calculates the conversion rate of all paths within a cluster, and analyzes the causal contribution of different advertising media within each cluster. The attribution adjustment module enumerates all subsets of advertising media, calculates the conversion rate of each subset, calculates the attribution allocation of advertising media based on the Shapley value theory, counts the attribution ratio of advertising media, and performs attribution-driven adjustments to the optimal delivery volume. The report generation module performs distortion detection based on the attribution ratio results, generates statistical reports of the final delivery volume data and actual conversion data, and encrypts the transmission of the reports.

[0037] This embodiment also provides a computer device applicable to cross-media advertising delivery methods, 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 implement the cross-media advertising delivery method proposed in the above embodiment.

[0038] 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.

[0039] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the cross-media advertising delivery method 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.

[0040] In summary, this invention not only reveals the actual synergistic effect between different advertising media combinations through the quantification of synergistic effects, but also enables real-time monitoring of the marginal return of each media through marginal benefit analysis of the first derivative of the response function, allowing for timely identification of the saturation point of the campaign. The introduction of constrained optimization methods ensures that the budget allocation process can meet the total budget and the actual deployment capacity of each media. Furthermore, by differentiating the objective function and solving the simultaneous equations, a globally optimal allocation scheme can be obtained. By collecting user behavior data during the advertising campaign period and transforming it into a directed graph structure, the true transfer paths of users between different advertising media can be comprehensively captured. Embedding the directed graph using the Node2Vec algorithm can effectively extract the complex relationships and potential path patterns between advertising media. Based on clustering, the conversion rates of various paths and the causal contributions of advertising media can be statistically analyzed, revealing the actual influence of different media in different conversion paths in greater detail, avoiding the errors caused by average attribution. Combining Shapley value theory for attribution allocation quantifies the marginal contribution of each advertising media in different path patterns, overcoming the problem of traditional attribution methods neglecting path order and combination.

[0041] 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. A cross-media advertising delivery method, characterized in that, include: Based on the collected advertising data from different advertising media, a response relationship function between advertising volume and conversion effect is established to determine the cross-media placement response relationship value. The correlation coefficient is calculated to determine the synergistic effect value of each pair of advertising media combinations. The first derivative of the response relationship function is used as the marginal benefit value. The budget is allocated by using a constrained optimization method based on the comprehensive synergistic effect value. Lagrange multipliers are introduced to calculate the partial derivative of the advertising volume for each media to determine the optimal advertising volume for each media. Collect user behavior data during the advertising campaign period, construct a directed graph, embed the paths in the directed graph using the Node2Vec algorithm, and perform path clustering on the resulting advertising media node vectors; The conversion rates of all paths within a cluster are statistically analyzed, the causal contributions of different media to the cluster are analyzed, all advertising media subsets are enumerated, the conversion rates of each subset and the conversion rates after adding advertising media m are calculated, the attribution allocation is performed using Shapley value theory, the attribution ratio is calculated, the distribution ratio of all user paths in each cluster is statistically analyzed, and the optimal delivery volume is adjusted based on attribution. Distortion detection is performed based on attribution percentage, and reports are generated based on the final delivery volume data and actual conversions, and then transmitted in encrypted form.

2. The cross-media advertising delivery method as described in claim 1, characterized in that: The process involves determining the cross-media placement response relationship value, calculating the correlation coefficient to determine the synergistic effect value of each advertising media combination, using the first derivative of the response relationship function as the marginal benefit value, and employing a constrained optimization method to allocate the budget based on the comprehensive synergistic effect value. For each advertising medium, a parameterizable Logistic function is used, and an adaptive parameter update mechanism is introduced to establish a response relationship function between advertising volume and conversion effect; Training is performed using the training set data, and the loss is calculated based on the calibrated training data. The parameters of the response function are updated online using the recursive least squares (RLS) method, and the parameters are automatically adjusted using the RLS algorithm to minimize the computational loss when updating the data. For each pair of advertising media combinations, the Pearson correlation coefficient of the normalized series of historical ad placements is calculated to determine the synergistic effect value of each pair of advertising media combinations, and the sum of synergistic effects is calculated for all advertising media combinations. For each advertising medium, calculate the first derivative of the response function under the current delivery volume as the marginal benefit value; The saturation threshold is based on the sum of the mean and standard deviation, using historical data. If the marginal benefit value is greater than the saturation threshold, it is determined that the corresponding media needs to continue to increase its investment. Based on the combined response values ​​of single media, synergistic effects, and saturation detection results, a constrained optimization method is adopted for budget allocation. The constraints are defined as follows: the sum of the ad spend of all media must not exceed the total budget; the ad spend of each media must not be negative; if the marginal benefit value is less than or equal to the saturation threshold, it is determined that the corresponding media has reached the saturation point and no further budget is allocated. A Lagrange multiplier λ is introduced to construct a Lagrange function. The partial derivative of the ad spend of each media is calculated and set to zero to obtain the optimality conditions. The optimal advertising volume is determined by combining the optimality conditions and budget constraints for the number of ad placements across all advertising media.

3. The cross-media advertising delivery method as described in claim 2, characterized in that: The process involves collecting user behavior data during the advertising campaign period, constructing a directed graph, embedding the paths in the directed graph using the Node2Vec algorithm, and then performing path clustering on the resulting advertising media node vectors. include, Collect all user behavior sequence data during the advertising campaign period, including: user unique identifier, advertising media m for each ad exposure, exposure timestamp, and behavior type; For each user, all their ad exposure events are arranged in chronological order to form an ordered path sequence; Merge all user path sequences to construct a directed graph, where node V is the number of all advertising media, edges represent user transitions from one media to another, and edge weights are the frequency of that transition in all user paths; Node attributes include: media ID, the number of times the media is first encountered in all paths, and the number of times it is included in the final conversion path; The Node2Vec algorithm is used to embed the directed graph path to obtain a low-dimensional vector representation of each advertising media node; For each user path, the path embedding vector is calculated as the mean of the embedding vectors of each node; K-means clustering is performed on all user path embedding vectors, and the number of clusters K is determined by the silhouette coefficient method.

4. The cross-media advertising delivery method as described in claim 3, characterized in that: The conversion rate of each subset and the conversion rate after adding advertising media m are calculated. Attribution allocation is performed using Shapley value theory, attribution percentages are calculated, and the distribution ratio of all user paths in each cluster is statistically analyzed. Attribution-driven adjustments are then made to the optimal ad delivery volume, including... For each path cluster, the conversion rate of all paths within the cluster is calculated based on the ratio of converted users to total users. For each advertising medium m, the conversion rate of paths containing the corresponding advertising medium within the cluster is calculated, and the causal contribution of different media under the cluster is calculated. For each user path, enumerate all subsets of advertising media, and calculate the conversion rate of each subset and the conversion rate after adding advertising media m; Attribution assignment is performed using Shapley value theory, and the sum of the average Shapley values ​​of all media is calculated as the total attribution value of the corresponding cluster. Then, the attribution proportion of advertising media m in cluster k is calculated. The distribution ratio of all user paths in each cluster is statistically analyzed, the global attribution weight of each advertising medium is calculated, and the optimal delivery volume is adjusted based on attribution.

5. The cross-media advertising delivery method as described in claim 4, characterized in that: The distortion detection based on attribution percentage includes, For each advertising medium, the attribution percentage is generated as a time series. The mean and standard deviation are calculated using a sliding window to detect attribution distortion. If the attribution percentage deviates from the mean by more than twice the standard deviation, it is judged as attribution distortion.

6. The cross-media advertising delivery method as described in claim 5, characterized in that: The method involves collecting advertising placement data based on different advertising media. include, Collect and organize historical advertising data, including: the number of ads placed for each advertising medium, the actual conversions for each advertising medium, the timestamp of the ad placement, user group characteristics, and the combination of advertising media for each placement; The number of deliverables and conversions are normalized, and the user group characteristics are converted into numerical features using one-hot encoding.

7. The cross-media advertising delivery method as described in claim 6, characterized in that: The process of generating reports based on the final delivery volume data and actual conversion numbers, and transmitting them in encrypted form, includes: The system monitors the actual conversion rate of each media in real time and records the corresponding final delivery volume data. It generates conversion reports using a report generation tool, encrypts the conversion reports based on the TLS encryption protocol, and protects the transmitted data with encryption using the AES encryption algorithm.

8. A cross-media advertising delivery system, based on the cross-media advertising delivery method according to any one of claims 1 to 7, characterized in that: include, The response function module collects data on advertising across different media and establishes a response relationship function between advertising volume and conversion rate. The budget allocation optimization module calculates the marginal benefit value based on the first derivative of the response relationship function, and introduces Lagrange multipliers to calculate the partial derivative with respect to the advertising amount to determine the optimal advertising amount for each advertising media. The user behavior analysis module collects user behavior data during the advertising campaign period, uses the Node2Vec algorithm to embed directed graph paths, and performs path clustering on the resulting advertising media node vectors. The conversion rate statistics module performs statistical analysis on the path clustering results, calculates the conversion rate of all paths within a cluster, and analyzes the causal contribution of different advertising media within each cluster. The attribution adjustment module enumerates all subsets of advertising media, calculates the conversion rate of each subset, calculates the attribution allocation of advertising media based on the Shapley value theory, counts the attribution ratio of advertising media, and performs attribution-driven adjustments to the optimal delivery volume. The report generation module performs distortion detection based on the attribution ratio results, generates statistical reports of the final delivery volume data and actual conversion data, and encrypts the transmission of the reports.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the cross-media advertising delivery method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the cross-media advertising delivery method according to any one of claims 1 to 7.

Images