Digital advertising delivery method based on big data

Abstract

The present invention relates to the field of digital marketing technology, specifically to a digital advertising delivery method based on big data, comprising the following steps: initializing an advertising model, importing advertising request time series data, performing timing analysis, adjusting queue entry rules and processing rates, calculating request distribution within differentiated time periods, and generating model parameter configurations. In the present invention, advertising timing and processing efficiency are optimized through queuing theory and time series data analysis. Timing analysis and parameter adjustment enhance the timeliness of advertising, ensure synchronization with user activity peaks, improve user engagement and advertising effectiveness, recalculate service time and estimate waiting time to support real-time monitoring of service efficiency, achieve precise matching of advertising resources with user needs, and quickly adapt to market changes through real-time strategy adjustment and simulation testing, thereby improving resource utilization efficiency and ROI. Refined management and data-driven optimization significantly enhance the accuracy and economic benefits of advertising delivery.

Description

Technical Field

[0001] The present invention relates to the field of digital marketing technology, and in particular to a digital advertising delivery method based on big data. Background Art

[0002] The field of digital marketing encompasses strategies and methods for promoting products and services using digital technologies, online platforms, and big data analytics. This field focuses on optimizing the efficiency and effectiveness of advertising content through interaction with internet users via electronic devices. Technical approaches include search engine optimization (SEO), content marketing, social media marketing, email marketing, affiliate marketing, and programmatic ad buying. Digital marketing also leverages user behavior data to personalize and precisely target advertising. By analyzing users' online behavior and preferences, advertisers can more effectively achieve their marketing goals and enhance interaction and engagement with consumers.

[0003] Among them, digital advertising methods refer to advertising activities implemented on digital platforms, which display brand or product information to the target audience through various online channels. Its main purpose is to attract potential customers, enhance brand awareness and promote sales. This method relies on Internet technology, including search engine advertising, social media advertising, mobile advertising and video advertising, etc. It can formulate personalized advertising strategies based on users' Internet usage habits and preferences and insights obtained through data analysis to improve advertising conversion rates and ROI (return on investment).

[0004] While existing technologies are widely used across various online platforms and tools, common issues include insufficient timeliness and precision in advertising delivery, and an inability to fully achieve the optimal match between ad content and user behavior. For example, traditional digital advertising methods rely heavily on fixed ad delivery rules and simple behavioral targeting, leading to inefficient use of advertising resources during off-peak periods and insufficient resources during peak periods. Existing technologies also fail to fully leverage real-time data for dynamic adjustments, lacking the ability to provide immediate feedback and adjustments to advertising effectiveness. This is particularly disadvantageous in a dynamic and volatile market environment, resulting in a decline in advertising ROI and a degraded user experience, making it difficult to effectively attract and retain potential customers. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the existing technology and propose a digital advertising method based on big data.

[0006] To achieve the above objectives, the present invention adopts the following technical solution, a digital advertising method based on big data, comprising the following steps:

[0007] S1: Initialize the advertising model, import advertising request time series data, perform timing analysis, adjust queuing rules and processing rates, calculate request distribution within differentiated time periods, and generate model parameter configuration;

[0008] S2: Calculate the arrival rate and service rate using the model parameter configuration, extract the timestamp and duration of each ad request, perform cluster analysis, adjust the timing parameters, and generate a timing parameter adjustment plan;

[0009] S3: Apply the timing parameter adjustment solution, recalculate the service time, use a simplified model to estimate the waiting time, analyze the difference between the estimated result and the real-time data, make fine adjustments, and generate performance tuning results;

[0010] S4: Based on the performance tuning results, adjust the advertising resources in real time, reconfigure advertising resource allocation and cache management, implement a new loading strategy, and generate a strategy implementation blueprint;

[0011] S5: Adopt the strategy implementation blueprint, design simulation tests, capture advertising performance data during the test, conduct effect analysis, identify advertising delivery patterns, and generate advertising delivery optimization results;

[0012] S6: Based on the advertising delivery optimization results, review the advertising content and user interaction, identify key advertising elements and viewing peaks, adjust the copy and visual elements, and synchronously adjust the delivery time to generate an optimal advertising delivery plan.

[0013] As a further solution of the present invention, the model parameter configuration includes the advertising request time series, the queue processing rate, and the advertising display time period; the timing parameter adjustment scheme includes the advertising reach rate adjustment, the advertising service rate optimization, and the advertising display time synchronization; the performance tuning results include the waiting time reduction, the service time optimization, and the efficiency optimization indicators; the strategy implementation blueprint includes the advertising loading speed optimization, the response speed improvement, and the resource reconfiguration parameters; the advertising delivery optimization results include the advertising effect improvement indicators, the advertising delivery effect comparison, and the advertising strategy adjustment feedback; the optimal advertising delivery plan includes the copy optimization parameters, the visual element adjustment plan, and the delivery time adjustment strategy.

[0014] As a further solution of the present invention, the steps of initializing the advertising model, importing the advertising request time series data, performing time series analysis, adjusting the queuing rules and processing rate, calculating the request distribution in differentiated time periods, and generating the model parameter configuration are as follows:

[0015] S101: Initialize the advertising model, import the advertising request time series data, perform data filtering operations, remove duplicate records, filter invalid and incomplete data items, and perform timestamp normalization on the remaining data to obtain a sorted data set;

[0016] S102: Based on the sorted data set, apply a moving average algorithm to analyze the request frequency in each time window, calculate the average request interval in each window, identify the peak request period, and make preliminary adjustments to the model's queuing rules and processing rate parameters based on the information to obtain request distribution characteristics;

[0017] S103: Optimize the advertisement model parameters based on the request distribution characteristics, adjust the queuing rules of advertisement requests, adjust the processing rate and process the request load, and build the model parameter configuration.

[0018] As a further solution of the present invention, the formula of the moving average algorithm is as follows:

[0019]

[0020] Calculate the average request interval time for each time window, where Represents the average request interval time, Representative The sum of the request intervals in the time windows, Representative The weight coefficient of the time window, Represents the number of time windows.

[0021] As a further solution of the present invention, the steps of calculating the arrival rate and service rate using the model parameter configuration, extracting the timestamp and duration of each advertising request, performing cluster analysis, adjusting the timing parameters, and generating a timing parameter adjustment solution are as follows:

[0022] S201: Filtering advertisement request records including timestamps and durations from the advertisement server log according to the model parameter configuration, extracting the data for unified storage, and generating a time data set;

[0023] S202: Based on the time data set, grouping technology is used to divide the timestamps and durations into differentiated categories, independent statistics are performed on each category, the activity distribution of the advertising requests is determined, labels are assigned to the grouped data, and a clustering result set is generated;

[0024] S203: Using the tag information in the clustering result set, adjust the time parameters of the advertisement display, and optimize the utilization efficiency of the advertisement resources by comparing the peak activity times of multiple categories, matching the advertisement release time and the user active time, and generating a timing parameter adjustment plan.

[0025] As a further solution of the present invention, the steps of applying the timing parameter adjustment solution, recalculating the service time, estimating the waiting time using a simplified model, analyzing the difference between the estimated result and the real-time data, and performing fine-tuning to generate the performance tuning result are as follows:

[0026] S301: recalculating the service time according to the timing parameter adjustment plan, estimating the preliminary waiting time using the baseline model, inputting the original service time data into the model, adjusting the input parameters of the time series, performing initialization calculations and predicting the waiting time, and obtaining waiting time estimation data;

[0027] S302: Using the waiting time estimation data, performing a difference comparison between the real-time data and the predicted data, including calculating the deviation between the predicted value and the real-time observed value, adjusting key parameters in the model according to the deviation result, and generating an adjusted parameter solution;

[0028] S303: Using the adjusted parameter solution, recalculate the model and optimize the service efficiency, compare the service time and processing rate of the model before and after the adjustment, perform efficiency evaluation, and obtain performance tuning results.

[0029] As a further solution of the present invention, based on the performance tuning results, real-time adjustment of advertising resources is performed, advertising resource allocation and cache management are reconfigured, and a new loading strategy is implemented. The steps of generating a strategy implementation blueprint are as follows:

[0030] S401: Analyze the performance tuning results, identify areas of low utilization efficiency of advertising resources in different time periods, adjust the advertising resource allocation strategy based on the data, optimize coverage and response time, and generate a resource reconfiguration plan;

[0031] S402: Using the resource reconfiguration solution, adjust the cache management rules for advertisements, including updating preloading conditions for advertisement content and setting cache priorities for differentiated advertisement types, monitor the impact of the adjustments on advertisement loading speed and user response, and generate cache management optimization results.

[0032] S403: Based on the cache management optimization results, refine the adjustment strategy and match it with the peak period of user activity, including adjusting the loading strategy and time window of demand ads, continuously monitoring the ad display efficiency and adjusting it in real time, and generating a strategy implementation blueprint.

[0033] As a further solution of the present invention, the steps of adopting the strategy implementation blueprint, designing simulation tests, capturing advertising effect data during the test, performing effect analysis, identifying advertising delivery patterns, and generating advertising delivery optimization results are as follows:

[0034] S501: According to the strategy implementation blueprint, key parameters for the simulation test are set, and an advertising delivery test is conducted based on the set advertising frequency and target audience. Multiple rounds of testing are conducted using differentiated advertising display formats, and advertising effectiveness data from each round of testing is captured, including user clicks and ad interaction duration, to obtain simulation test data records.

[0035] S502: Based on the simulated test data records, a linear regression algorithm is applied to classify the data of the differentiated advertising formats, perform group analysis of advertising effects, analyze the performance of the advertising placements in turn, identify the effectiveness differences of the differentiated advertising formats by calculating key indicators such as click volume and interaction time, and generate advertising effect analysis results;

[0036] S503: Based on the advertising effectiveness analysis results, identify poorly performing advertising formats, fine-tune the advertising strategy by adjusting the advertising content, display order and delivery time, retest the advertising delivery model, capture new advertising effectiveness data, and generate advertising delivery optimization results.

[0037] As a further solution of the present invention, the formula of the linear regression algorithm is as follows:

[0038]

[0039] Calculate the advertising effect and get the advertising efficiency value ,in, Represents the number of clicks, Represents the interaction time, Represents the number of interactions, Represents the total duration of the test period, Indicates the frequency of interaction per unit time. is the intercept of the regression model, 、 and They are the weight coefficients of click volume, interaction duration and interaction frequency per unit time respectively.

[0040] As a further embodiment of the present invention, the steps of reviewing the ad content and user interaction, identifying key ad elements and viewing peaks, adjusting the copy and visual elements, and synchronously adjusting the ad delivery time to generate the optimal ad delivery plan based on the ad delivery optimization results are as follows:

[0041] S601: Based on the advertising delivery optimization results, user interaction data in the advertisement is captured. By recording the number of user clicks and comments on the advertisement, time periods with high user activity are identified. Interaction data of the advertisement elements during the peak period is recorded to obtain advertisement element interaction records.

[0042] S602: Based on the interaction records of the advertising elements, screen the copy and visual elements, adjust the layout and content of the elements, adjust the advertising delivery period by matching the user's active time period, conduct advertising effectiveness testing within the period, and evaluate the adjustment effect by capturing user feedback and interaction data during the testing period to obtain an advertising test optimization plan;

[0043] S603: Apply the advertising test optimization plan to adjust the advertising strategy, unify the advertising content and delivery time on differentiated platforms, and generate the optimal advertising delivery plan through cross-platform data collection and analysis.

[0044] Compared with the prior art, the advantages and positive effects of the present invention are:

[0045] In the present invention, by introducing the queuing theory framework and advertising request time series data analysis, the timing accuracy and processing efficiency of advertising delivery are optimized. By performing timing analysis and parameter adjustment on the advertising request data, the timeliness of advertising delivery is enhanced, ensuring that the advertising display is synchronized with the peak of user activity, thereby improving user participation and advertising effect. By recalculating the service time and estimating the waiting time, the service efficiency can be monitored and adjusted in real time, thereby more accurately matching advertising resources with user needs. Through real-time adjustment of strategies and simulation testing, it is possible to quickly respond to market changes, improve the utilization efficiency and ROI of advertising resources, and significantly improve the accuracy and economic benefits of advertising delivery through refined management and data-driven optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a schematic diagram of the main steps of the present invention;

[0047] Figure 2 This is a schematic diagram of the refinement of S1 of the present invention;

[0048] Figure 3 This is a schematic diagram of the refinement of S2 of the present invention;

[0049] Figure 4 This is a schematic diagram of the refinement of S3 of the present invention;

[0050] Figure 5 This is a schematic diagram of the refinement of S4 of the present invention;

[0051] Figure 6 This is a schematic diagram of the refinement of S5 of the present invention;

[0052] Figure 7 This is a detailed schematic diagram of S6 of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0054] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined. Example

[0055] See also Figure 1 The present invention provides a technical solution, a digital advertising method based on big data, comprising the following steps:

[0056] S1: Initialize the advertising model. Using the basic framework of queuing theory, import ad request time series data, perform time series analysis on the data, adjust the queuing rules and processing rate, calculate the request distribution within differentiated time periods, adjust model parameters, and generate the model parameter configuration.

[0057] S2: Calculate the reach and service rates using model parameter configuration. For each ad request, extract the timestamp and duration, perform cluster analysis on the data, and adjust timing parameters based on the clustering results to synchronize ad display with user activity peaks, generating a timing parameter adjustment plan.

[0058] S3: Apply the timing parameter adjustment plan to recalculate service time. Use a simplified model to estimate waiting time. Analyze the difference between the estimated results and real-time data, fine-tune the parameters, and compare the service efficiency before and after the adjustment to evaluate the adjustment effect and generate performance tuning results.

[0059] S4: Based on the performance tuning results, real-time adjustments are made to ad resources, ad resource allocation and cache management are reconfigured, new loading strategies are implemented, ad loading and response speeds are monitored in real time, strategies are adjusted based on the monitoring data, and a blueprint for strategy implementation is generated.

[0060] S5: Adopt the strategy implementation blueprint, design simulation tests, capture advertising performance data during the testing process, analyze the data, identify advertising delivery patterns, fine-tune the advertising strategy based on the analysis results, optimize advertising delivery, and generate advertising delivery optimization results;

[0061] S6: Based on the results of ad delivery optimization, review ad content and user interactions, identify key ad elements and viewing peaks, adjust copy and visual elements, and simultaneously adjust delivery time to match user activities. Conduct small-scale tests and collect feedback to generate the optimal ad delivery plan.

[0062] Model parameter configuration includes ad request time series, queue processing rate, and ad display time period. Timing parameter adjustment plans include ad reach rate adjustment, ad service rate optimization, and ad display time synchronization. Performance tuning results include waiting time reduction, service time optimization, and efficiency optimization indicators. The strategy implementation blueprint includes ad loading speed optimization, response speed improvement, and resource reconfiguration parameters. Ad delivery optimization results include ad effect improvement indicators, ad delivery effect comparison, and ad strategy adjustment feedback. The optimal ad delivery plan includes copy optimization parameters, visual element adjustment plan, and delivery time adjustment strategy.

[0063] Specifically, if Figure 2 As shown in the figure, the steps to initialize the advertising model, import the advertising request time series data, perform time series analysis, adjust the queue rules and processing rate, calculate the request distribution in different time periods, and generate the model parameter configuration are as follows:

[0064] S101: Initialize the advertising model, import the advertising request time series data, perform data filtering operations, remove duplicate records, filter invalid and incomplete data items, and perform timestamp normalization on the remaining data to obtain the sorted data set. The execution process is as follows;

[0065] Initialize the advertising model, import ad request time series data, and preprocess the input ad request time series data. During the data screening phase, a specific algorithm is used to identify and remove duplicate records to ensure data uniqueness. The data is further screened to exclude invalid and incomplete data items. For example, each record is checked for completeness to ensure all required fields are filled in, and records missing key information are deleted. After completing these steps, the remaining data undergoes timestamp normalization, which involves converting all date and time information to a unified format and time zone for subsequent data processing and analysis. After timestamp normalization, the system sorts the data to ensure it is correctly arranged in chronological order, facilitating subsequent data processing operations. Detailed data processing logs are recorded, including the number and reasons for data removal, as well as the results of data normalization, to facilitate issue tracking and subsequent audits. Successful execution of these steps is crucial for building an accurate advertising model, laying a solid foundation for subsequent data analysis and model building, ultimately resulting in a consolidated dataset.

[0066] S102: Based on the sorted data set, apply the moving average algorithm to analyze the request frequency within each time window, calculate the average request interval in each window, identify the peak request period, and make preliminary adjustments to the model's queuing rules and processing rate parameters based on this information. The execution process for obtaining request distribution characteristics is as follows;

[0067] The formula for the moving average algorithm is as follows:

[0068]

[0069] Calculate the average request interval time for each time window, where Represents the average request interval time, Representative The sum of the request intervals in the time windows, Representative The weight coefficient of the time window, Represents the number of time windows.

[0070] Detailed explanation of the formula and the process of formula calculation and derivation:

[0071] The calculated average request interval for each time window is a key indicator for evaluating request frequency and adjusting queue rules.

[0072] For the The sum of the request intervals within a time window, which is obtained by actually collecting and time-stamping historical log data.

[0073] It is The weight coefficient of each time window. This coefficient is set based on the reference that the data of the most recent time window is more important than the data of the more distant time window. It is generally defined using an exponential decay model. For example, the weight is proportional to the inverse of the time window from now, ensuring that the recent data has a greater impact on the average value.

[0074] is the total number of time windows, determined by the duration and granularity of data collection;

[0075] There are three time windows, and the sum of the request intervals in each window is They are 100 seconds, 150 seconds and 200 seconds respectively, and the corresponding weights are Set to 0.5, 1, and 1.5. The weight setting is based on the distance from the current time. The closer the time window, the higher the weight. The value of is 3;

[0076] Compute the product of each weight term:

[0077] Find the sum of the weights:

[0078]

[0079] Calculated according to the formula :

[0080]

[0081] The results show that after referring to the time window weight, the weighted average request interval is 166.67 seconds. This indicates that under the current settings, the average interval between each request is 166.67 seconds. This value reflects the combined impact of recent and long-term data, which helps to more accurately identify peak request periods and adjust the model's queuing rules and processing rate parameters, thereby optimizing overall request processing performance.

[0082] S103: Based on the request distribution characteristics, the advertising model parameters are optimized, the queuing rules for advertising requests are adjusted, the processing rate is adjusted, and the request load is handled. The execution process of constructing the model parameter configuration is as follows;

[0083] Based on request distribution characteristics, we optimize ad model parameters and adjust the enqueuing rules for ad requests. We first analyze the distribution characteristics of ad requests in the collated dataset. By analyzing multiple dimensions, including temporal distribution, frequency, and type, we identify high-frequency request periods and inefficient request types. Based on these results, we optimize model parameters. For example, we adjust ad delivery windows to match periods of high user activity, increasing ad exposure and click-through rates. We optimize the relevance and appeal of ad content based on the specific content of the requests, improving user engagement and satisfaction. We also adjust the enqueuing rules for ad requests and set priority queues to ensure that high-value ad requests are prioritized, maximizing ad revenue. To adjust the processing rate, the system employs a dynamic adjustment strategy that automatically adjusts the processing rate based on real-time system load and request volume, ensuring it avoids overload during sudden bursts of traffic while conserving resources during periods of low load. To manage request load, we implement load balancing technology to distribute requests to differentiated processing nodes, avoiding overloading of any single node. This allows us to build model parameter configurations.

[0084] Specifically, if Figure 3 As shown in the figure, the steps for calculating the arrival rate and service rate by configuring the model parameters, extracting the timestamp and duration of each ad request, performing cluster analysis, adjusting the timing parameters, and generating a timing parameter adjustment plan are as follows:

[0085] S201: Based on the model parameter configuration, ad request records including timestamps and durations are filtered from the ad server logs, and the data is extracted for unified storage. The execution process for generating a time dataset is as follows;

[0086] According to the model parameter configuration, we first filter the ad request records containing timestamps and durations from the ad server logs. During the screening process, regular expressions are used to match the corresponding log formats to ensure the accuracy and completeness of the acquired data. Then, the data is extracted for unified storage. The storage involved includes data formatting, cleaning, and conversion to ensure the consistency and availability of the data set. The extracted data includes the specific timestamp of the ad request, the specific duration of the request, the associated user information, and the category of the ad. The data will be stored in a pre-set database, using efficient database storage technologies, such as distributed storage systems, to ensure fast data reading and writing and high reliability, and generate a time data set.

[0087] S202: Based on the time dataset, a grouping technique is used to group timestamps and durations into differentiated categories. Independent statistics are performed on each category to determine the activity distribution of ad requests. Labels are assigned to the grouped data to generate a clustering result set. The execution process is as follows:

[0088] Independent statistics are performed for each category according to the formula:

[0089]

[0090] Calculate the activity distribution of ad requests, where Represents the percentage of advertising activity. Represents duration, Represents the total number of records;

[0091] Set the ad request record per category to , the sum of the duration is The total number of records is , then the advertising activity of each category The above formula can be used to calculate that, assuming there are 500 ad request records of a certain category, the sum of the duration is 10,000 seconds, and the total number of records is 1,000, the ad activity is calculated as ;

[0092] This result shows that ad requests in this category are very active and the frequency of ad delivery should be increased accordingly.

[0093] S203: Using the tag information in the clustering result set, adjust the timing parameters of the ad display. By comparing the peak activity times of multiple categories, matching the ad publishing time with the user's active time, the utilization efficiency of the ad resources is optimized. The execution process of generating the timing parameter adjustment plan is as follows;

[0094] Utilize the label information in the clustering result set and adjust the time parameters of advertising display by comparing the peak activity times of multiple categories. The label information in the clustering result set includes the category label of each advertising request and its corresponding activity. The information is classified and clustered through data analysis tools. The data analysis tools used include K-means clustering algorithm or hierarchical clustering algorithm to ensure that the label of each category is correct. For each category, calculate the peak activity time of its advertising request. This calculation is achieved by counting the frequency of advertising requests in each time period, and then match the advertising release time according to the user's active time. This process involves complex data comparison and time series analysis, and finally generates a timing parameter adjustment plan to optimize the efficiency of advertising resource utilization.

[0095] Specifically, if Figure 4 As shown in the figure, the steps for applying the timing parameter adjustment plan, recalculating the service time, estimating the waiting time using a simplified model, analyzing the difference between the estimated results and the real-time data, and making fine adjustments to generate the performance tuning results are as follows:

[0096] S301: Recalculate service time according to the timing parameter adjustment plan, use the baseline model to estimate preliminary waiting time, input the original service time data into the model, adjust the input parameters of the time series, perform initialization calculations and predict waiting time. The execution process for obtaining waiting time estimation data is as follows;

[0097] The baseline model is used to estimate the initial waiting time according to the formula:

[0098]

[0099] Calculate wait time estimates where represents the estimated waiting time, represents the initial service time, and are model parameters;

[0100] The initial service time is estimated based on the baseline of the reference model. Assuming the average time per request is 15 minutes (900 seconds), the model parameters and Represent the time adjustment coefficient and constant deviation respectively, let and Seconds, estimate the waiting time and calculate it as Second;

[0101] This indicates that the estimated waiting time increases slightly by adjusting the parameters, reflecting the adaptability of the model after referring to the actual operational impact.

[0102] S302: Using the waiting time estimation data, compare the difference between the real-time data and the predicted data, including calculating the deviation between the predicted value and the real-time observed value, and adjusting the key parameters in the model based on the deviation result. The execution process of generating the adjusted parameter solution is as follows;

[0103] Waiting time estimation data is used to compare the differences between real-time data and predicted data. During the calculation process, the difference between the actual waiting time and the predicted waiting time for each service is first calculated, and the size of the deviation is quantified by the standard deviation or the sum of squared errors. This calculation not only involves simple numerical subtraction, but also includes an overall performance evaluation of the data, such as statistical analysis of deviations. Based on the obtained deviation results, key parameters in the model are adjusted, such as adjusting the sliding window size or weight parameters in the time series model, to ensure that the accuracy of the model prediction is more consistent with the actual data, and generate an adjusted parameter scheme. The parameter adjustment is based on the analysis of the systematic deviation between the actual observed data and the predicted data, aiming to reduce the error of future predictions and improve the practicality and accuracy of the model.

[0104] S303: Using the adjusted parameter solution, recalculate the model and optimize the service efficiency. Compare the service time and processing rate of the model before and after the adjustment to evaluate the efficiency. The execution process to obtain the performance tuning results is as follows;

[0105] Use the adjusted parameter plan to recalculate the model and optimize service efficiency. Recalculate the model through data analysis software or customized scripts, including entering new parameter values ​​and monitoring resource consumption during model runtime, such as CPU and memory usage. Compare the service time and processing rate of the model before and after adjustment. This involves data aggregation, comparison and visualization. The process focuses not only on improving the model calculation speed, but also on the overall improvement of service quality. Efficiency assessment is performed to obtain performance tuning results. The results are obtained by comparing the model's response time and processing capacity. The optimization results show that the model can process requests faster after adjusting the parameters, thereby improving service efficiency and customer satisfaction.

[0106] Specifically, if Figure 5 As shown in the figure, based on the performance tuning results, real-time adjustments to ad resources are made, ad resource allocation and cache management are reconfigured, and a new loading strategy is implemented. The specific steps to generate a strategy implementation blueprint are as follows:

[0107] S401: Analyze performance tuning results, identify areas with low advertising resource utilization efficiency in different time periods, adjust advertising resource allocation strategies based on the data, optimize coverage and response time, and generate resource reconfiguration plans. The execution process is as follows:

[0108] Analyze performance tuning results and identify areas with low utilization efficiency of advertising resources in differentiated time periods. This analysis uses data mining techniques such as cluster analysis to identify changes in advertising display efficiency within differentiated time periods. Through the data, multi-dimensional efficiency charts are generated to display advertising response and coverage within differentiated time periods. Based on the analysis results, the advertising resource allocation strategy is adjusted, including reallocating the time and area of ​​advertising display to improve resource utilization in inefficient areas, optimize coverage and response time, and ultimately generate a resource reconfiguration plan. This plan is driven by real-time data to ensure that advertising resource allocation is better aligned with market demand and user behavior.

[0109] S402: Adopting a resource reconfiguration solution to adjust ad cache management rules, including updating ad content preloading conditions and setting cache priorities for differentiated ad types, monitoring the impact of the adjustments on ad loading speed and user response, and generating cache management optimization results. The execution process is as follows:

[0110] Adjust the ad cache management rules according to the formula:

[0111]

[0112] Calculate the cache management optimization results, where Represents the adjusted cache capacity, Represents the original cache capacity, represents the adjustment factor based on user response time;

[0113] Refer to cache management optimization, original cache capacity Set to 100 units, adjust the coefficient According to the average user response time increased by 20%, that is ;

[0114] Calculate the new cache capacity as units, which indicates that the adjusted cache capacity is increased to meet higher data requests and improve user experience.

[0115] S403: Based on the cache management optimization results, refine the adjustment strategy to match the peak user activity period, including adjusting the loading strategy and time window of demand ads, continuously monitoring ad display efficiency and making real-time adjustments, and generating a strategy implementation blueprint. The execution process is as follows;

[0116] Based on the cache management optimization results, refine the adjustment strategy and match it with the peak period of user activity. The adjustment strategy includes analyzing user activity data, identifying high-demand ads, and reconfiguring the ad loading strategy. The adjustment strategy uses data analysis to determine the specific time period of user peak period, and then sets the time window for ad loading to ensure that the necessary ad support is provided during the user's most active period. Continuously monitor the ad display efficiency and adjust it in real time. The monitoring technologies involved include real-time data flow analysis and response time tracking to ensure the continuous optimization and adjustment of the advertising strategy, and finally generate a strategy implementation blueprint. The blueprint details the implementation steps and expected effects of various strategy adjustments to ensure the effective use and optimization of advertising resources.

[0117] Specifically, if Figure 6 As shown in the figure, the steps to adopt the strategy implementation blueprint, design simulation tests, capture advertising performance data during the test, conduct effect analysis, identify advertising delivery patterns, and generate advertising delivery optimization results are as follows:

[0118] S501: According to the strategy implementation blueprint, set the key parameters of the simulation test, conduct an ad placement test based on the set ad frequency and target audience, and conduct multiple rounds of testing using differentiated ad display formats. Capture ad performance data from each round of testing, including user clicks and ad interaction duration, to obtain simulation test data records. The execution process is as follows;

[0119] According to the strategy implementation blueprint, key parameters for the simulation test are set based on the data characteristics previously analyzed. These include ad display frequency (i.e., the number of impressions per hour and per day), as well as target audience definition, such as age, location, and interests. This ensures that these parameters accurately reflect the actual needs of the target market. Ad placement tests are then conducted based on the set frequency and target audience. The system utilizes differentiated ad display formats, such as static images, dynamic videos, or interactive ads, to test the effectiveness of these different formats with different target groups. Multiple rounds of testing are conducted, each targeting a different ad format or target group, ensuring the multidimensionality of the testing and the reliability of the results. During testing, the system uses advanced data capture tools to record every ad display and user interaction, capturing ad performance data from each test round, including user clicks and ad engagement duration. This data allows the system to monitor ad performance in real time and conduct preliminary analysis based on real-time data, generating simulated test data records. This record not only helps verify which ad formats are most effective but also guides future optimization directions.

[0120] S502: Based on the simulated test data records, a linear regression algorithm is applied to classify the data of differentiated advertising formats, perform group analysis of advertising effectiveness, analyze the performance of advertising placements in turn, and identify the effectiveness differences of differentiated advertising formats by calculating key indicators such as click volume and interaction time. The execution process for generating advertising effectiveness analysis results is as follows;

[0121] The formula for the linear regression algorithm is as follows:

[0122]

[0123] Calculate the advertising effect and get the advertising efficiency value ,in, Represents the number of clicks, Represents the interaction time, Represents the number of interactions, Represents the total duration of the test period, Indicates the frequency of interaction per unit time. is the intercept of the regression model, 、 and They are the weight coefficients of click volume, interaction duration and interaction frequency per unit time respectively.

[0124] Detailed explanation of the formula and the process of formula calculation and derivation:

[0125] This formula is used to calculate the advertising effect, where Represents the advertising efficiency value;

[0126] Indicates the number of clicks, which is obtained by monitoring the total number of clicks on a specific ad within a certain period of time. For example, set the cumulative number of clicks on an ad to 1200 within a week;

[0127] Interaction duration refers to the total duration of user interaction with an ad. This duration is calculated by recording the interaction time of all users with the ad through the data collection system. The total user interaction duration for the same ad within a week is set to 3600 seconds.

[0128] Indicates the number of interactions, obtained by counting the interaction events of a specific ad, and the number of interactions within a week is set to 300;

[0129] Indicates the total duration of the test period in seconds, for example, the total duration of one week is 604800 seconds (7 days);

[0130] Weight coefficient The intercept and the weight of each factor on the advertising effect are respectively set according to historical data analysis and linear regression model fitting. , , , ,The setting of the coefficient is based on the statistical analysis of actual advertising operation data,,reflecting the contribution of differentiation factors to advertising effects;

[0131] Substitute the specific values ​​into the formula for calculation:

[0132]

[0133] The results show that, given the number of clicks, interaction duration, and interaction frequency, the advertising efficiency value is 10.100744. This value reflects the comprehensive effect of the advertisement in attracting clicks and maintaining user interaction. The high advertising efficiency value indicates that the advertising performance was good during the test period and effectively promoted user interaction and participation.

[0134] S503: Based on the advertising effectiveness analysis results, identify poorly performing advertising formats, fine-tune the advertising strategy by adjusting the advertising content, display sequence, and delivery time, retest the advertising delivery model, capture new advertising effectiveness data, and generate advertising delivery optimization results. The execution process is as follows;

[0135] Based on the results of ad effectiveness analysis, we identify underperforming ad formats. Based on the results of simulation tests, we analyze the results to identify those with poor user engagement and click-through rates. Through a detailed data analysis module, we conduct in-depth analysis of the collected ad effectiveness data to identify common characteristics and potential causes of underperforming ad formats. Based on the analysis results, we adjust ad content, such as changing visual elements or messaging to better attract the target audience, adjusting the display order to ensure that more popular ad formats are shown more frequently, and adjusting the delivery time to target users during peak user activity periods to increase effective reach. After fine-tuning the ad strategy, we retest the ad delivery model. This retest not only verifies the effectiveness of the previous adjustments but also captures the improved performance of the ads through new data capture. The new test results are recorded and analyzed in detail by the system to generate optimized ad delivery results. This information provides a scientific basis for future advertising strategies and ensures the maximum return on advertising investment.

[0136] Specifically, if Figure 7 As shown in the figure, based on the results of ad delivery optimization, the steps to generate the optimal ad delivery plan include reviewing ad content and user interaction, identifying key ad elements and viewing peaks, adjusting copy and visual elements, and synchronously adjusting delivery time:

[0137] S601: Based on the advertising delivery optimization results, capture user interaction data in the advertisement. By recording the number of user clicks and comments on the advertisement, identify time periods with high user activity, and record the interaction data of the advertisement elements during the peak time periods. The execution process for obtaining the advertisement element interaction record is as follows;

[0138] Identify the time period with high user activity, according to the formula:

[0139]

[0140] Calculate the interaction records of advertising elements, where Represents the time period when users are most active. Represents the number of clicks, Represents the number of comments, Represents the total time period;

[0141] Refer to monitoring users’ clicks and comments on advertisements. If 100 clicks and 50 comments are recorded within an hour, and the total time period is set to 1 hour (60 minutes), the time period with high user activity is calculated as ;

[0142] The results show that during this hour, there were an average of 2.5 interactions per minute, reflecting a period of high user activity.

[0143] S602: Based on the interaction records of advertising elements, screen the copy and visual elements, adjust the layout and content of the elements, adjust the advertising delivery period by matching the user's active time period, conduct advertising effectiveness testing within the period, and evaluate the adjustment effect by capturing user feedback and interaction data during the test period. The execution process of obtaining the advertising test optimization plan is as follows;

[0144] Based on the interaction records of advertising elements, screen the copy and visual elements, and adjust the layout and content of the elements. The process includes analyzing the interaction data, determining which copy and visual elements are most attractive to users, adjusting the content and visual layout of the advertisement to adapt to user preferences through data-driven methods, adjusting the advertising delivery time period by matching the user's active time period, implementing advertising effect testing within the time period, capturing user feedback and interaction data during the test to evaluate the adjustment effect, and obtaining an advertising test optimization plan, which is dynamically adjusted according to the specific response of users to ensure the optimal match between advertising content and delivery time.

[0145] S603: Apply the advertising test optimization plan to adjust the advertising strategy, unify the advertising content and delivery time on different platforms, and generate the optimal advertising delivery plan through cross-platform data collection and analysis. The execution process is as follows;

[0146] Advertisement testing and optimization plans are applied to adjust advertising strategies and unify advertising content and delivery schedules across differentiated platforms. This includes cross-platform data collection and analysis. Advertisement delivery data from differentiated platforms is collected through a unified interface, and the data is analyzed to identify performance differences between differentiated platforms. Advertisement content and delivery strategies are adjusted based on the analysis results to suit the specific needs and user behavior patterns of each platform, generating the optimal ad delivery plan. This plan aims to improve the overall effectiveness and ROI of advertising through fine-tuning.

[0147] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A digital advertising method based on big data, characterized in that: The following steps are involved: Initialize the advertising model, import ad request time series data, perform timing analysis, adjust queuing rules and processing rates, calculate request distribution within differentiated time periods, and generate model parameter configurations. Utilizing the model parameter configuration, calculating the arrival rate and service rate, extracting the timestamp and duration of each ad request, performing cluster analysis, adjusting the timing parameters, and generating a timing parameter adjustment plan; Apply the timing parameter adjustment solution, recalculate the service time, use a simplified model to estimate the waiting time, analyze the difference between the estimated results and the real-time data, make fine adjustments, and generate performance tuning results; Based on the performance tuning results, adjust advertising resources in real time, reconfigure advertising resource allocation and cache management, implement new loading strategies, and generate a strategy implementation blueprint; Adopt the strategy implementation blueprint, design simulation tests, capture advertising performance data during the test, conduct performance analysis, identify advertising delivery patterns, and generate advertising delivery optimization results; Based on the advertising delivery optimization results, review the advertising content and user interaction, identify key advertising elements and viewing peaks, adjust the copy and visual elements, and synchronously adjust the delivery time to generate the optimal advertising delivery plan.

2. The digital advertising method based on big data according to claim 1, characterized in that: The model parameter configuration includes the advertising request time series, queue processing rate, and advertising display time period. The timing parameter adjustment plan includes advertising reach rate adjustment, advertising service rate optimization, and advertising display time synchronization. The performance tuning results include waiting time reduction, service time optimization, and efficiency optimization indicators. The strategy implementation blueprint includes advertising loading speed optimization, response speed improvement, and resource reconfiguration parameters. The advertising delivery optimization results include advertising effect improvement indicators, advertising delivery effect comparison, and advertising strategy adjustment feedback. The optimal advertising delivery plan includes copy optimization parameters, visual element adjustment plan, and delivery time adjustment strategy.

3. The digital advertising method based on big data according to claim 1, characterized in that: The steps to initialize the advertising model, import advertising request time series data, perform time series analysis, adjust queuing rules and processing rates, calculate request distribution within differentiated time periods, and generate model parameter configuration are as follows: Initialize the advertising model, import the ad request time series data, perform data filtering operations, remove duplicate records, filter invalid and incomplete data items, and normalize the timestamps of the remaining data to obtain the sorted data set; Based on the collated data set, a moving average algorithm is applied to analyze the request frequency within each time window, calculate the average request interval in each window, identify the peak request period, and make preliminary adjustments to the model's queuing rules and processing rate parameters based on this information to obtain request distribution characteristics; Through the request distribution characteristics, the advertising model parameters are optimized, the queuing rules of the advertising requests are adjusted, the processing rate is adjusted and the request load is processed, and the model parameter configuration is constructed.

4. The digital advertising method based on big data according to claim 3, characterized in that: The formula of the moving average algorithm is as follows: Calculate the average request interval time for each time window, where Represents the average request interval time, Representative The sum of the request intervals in the time windows, Representative The weight coefficient of the time window, Represents the number of time windows.

5. The digital advertising method based on big data according to claim 1, characterized in that: The steps of using the model parameter configuration to calculate the arrival rate and service rate, extracting the timestamp and duration of each ad request, performing cluster analysis, adjusting the timing parameters, and generating a timing parameter adjustment plan are as follows: According to the model parameter configuration, the ad request records including timestamps and durations are filtered from the ad server logs, and the data is extracted for unified storage to generate a time data set; Based on the time dataset, a grouping technique is used to group timestamps and durations into differentiated categories, independent statistics are performed on each category to determine the activity distribution of ad requests, labels are assigned to the grouped data, and a clustering result set is generated; The tag information in the clustering result set is used to adjust the time parameters of the advertisement display. By comparing the peak activity times of multiple categories, the advertisement publishing time and the user active time are matched, the utilization efficiency of the advertisement resources is optimized, and a timing parameter adjustment plan is generated.

6. The digital advertising method based on big data according to claim 1, characterized in that: Apply the timing parameter adjustment solution, recalculate the service time, use a simplified model to estimate the waiting time, analyze the difference between the estimated results and the real-time data, make fine adjustments, and generate performance tuning results. The specific steps are as follows: Recalculating the service time according to the timing parameter adjustment plan, estimating the preliminary waiting time using the baseline model, inputting the original service time data into the model, adjusting the input parameters of the time series, performing initialization calculations and predicting the waiting time, and obtaining waiting time estimation data; Using the waiting time estimation data, performing a difference comparison between the real-time data and the predicted data, including calculating the deviation between the predicted value and the real-time observed value, adjusting key parameters in the model according to the deviation result, and generating an adjusted parameter solution; The adjusted parameter scheme is used to recalculate the model and optimize the service efficiency. The service time and processing rate of the model before and after the adjustment are compared to perform an efficiency assessment and obtain a performance tuning result.

7. The digital advertising method based on big data according to claim 1, characterized in that: Based on the performance tuning results, real-time adjustments to ad resources are made, ad resource allocation and cache management are reconfigured, and a new loading strategy is implemented. The specific steps for generating a strategy implementation blueprint are as follows: Analyze the performance tuning results, identify areas of low utilization efficiency of advertising resources in different time periods, adjust the allocation strategy of advertising resources based on the data, optimize coverage and response time, and generate a resource reconfiguration plan; Adopting the resource reconfiguration solution, adjusting the cache management rules for advertisements, including updating the preloading conditions for advertisement content and setting cache priorities for differentiated advertisement types, monitoring the impact of the adjustments on advertisement loading speed and user response, and generating cache management optimization results; Based on the cache management optimization results, refine the adjustment strategy and match it with the peak period of user activity, including adjusting the loading strategy and time window of demand ads, continuously monitoring the ad display efficiency and adjusting it in real time, and generating a strategy implementation blueprint.

8. The digital advertising method based on big data according to claim 1, characterized in that: The steps for implementing the strategy blueprint, designing simulation tests, capturing advertising performance data during the tests, conducting performance analysis, identifying advertising delivery patterns, and generating advertising delivery optimization results are as follows: According to the strategy implementation blueprint, set the key parameters of the simulation test, conduct advertising delivery tests based on the set advertising frequency and target audience, and conduct multiple rounds of testing using differentiated advertising display formats. Capture advertising performance data from each round of testing, including user clicks and ad interaction duration, to obtain simulation test data records; Based on the simulated test data records, a linear regression algorithm is applied to classify the data of the differentiated advertising formats, perform group analysis of advertising effects, analyze the performance of advertising placements in turn, and identify the effectiveness differences of the differentiated advertising formats by calculating key indicators such as click volume and interaction time, thereby generating advertising effect analysis results; Based on the advertising effectiveness analysis results, identify poorly performing advertising formats, fine-tune advertising strategies by adjusting advertising content, display sequence, and delivery time, retest advertising delivery models, capture new advertising effectiveness data, and generate advertising delivery optimization results.

9. The digital advertising method based on big data according to claim 8, characterized in that: The formula of the linear regression algorithm is as follows: Calculate the advertising effect and get the advertising efficiency value ,in, Represents the number of clicks, Represents the interaction time, Represents the number of interactions, Represents the total duration of the test period, Indicates the frequency of interaction per unit time. is the intercept of the regression model, 、 and They are the weight coefficients of click volume, interaction duration and interaction frequency per unit time respectively.

10. The digital advertising method based on big data according to claim 1, characterized in that: Based on the advertising optimization results, the steps for reviewing advertising content and user interactions, identifying key advertising elements and viewing peaks, adjusting copy and visual elements, and synchronously adjusting the advertising time to generate the optimal advertising delivery plan are as follows: Based on the advertising delivery optimization results, user interaction data in the advertisement is captured. By recording the number of user clicks and comments on the advertisement, time periods with high user activity are identified, and interaction data of the advertisement elements during peak hours is recorded to obtain interaction records of the advertisement elements. Based on the interaction records of the advertising elements, the copy and visual elements are screened, the layout and content of the elements are adjusted, the advertising delivery period is adjusted by matching the user's active time period, and the advertising effect is tested within the time period. The adjustment effect is evaluated by capturing user feedback and interaction data during the test period to obtain an advertising test optimization plan; Apply the advertising test optimization plan to adjust the advertising strategy, unify the advertising content and delivery time on differentiated platforms, and generate the optimal advertising delivery plan through cross-platform data collection and analysis.

Images