Flow advertisement prediction putting method and system platform based on AI artificial intelligence

Abstract

The invention discloses a traffic advertisement prediction putting method and system platform based on AI artificial intelligence, and belongs to the technical field of artificial intelligence, and the method comprises the steps: obtaining user behavior data, network traffic data and advertisement related data from a plurality of data sources, according to the method, key features such as user activeness, interest preference and the like are extracted through deep analysis by feature extraction and screening, normalization processing is performed on numeric features, and classification features are encoded, so that the effectiveness of model training is enhanced; the traffic advertisement prediction module is used for accurately predicting traffic through prediction models such as linear regression on the basis of extracted and screened features, the prediction models can be flexibly expanded and optimized according to actual requirements, and the prediction performance is improved; an optimal advertisement putting strategy is formulated through an advertisement putting decision module according to a flow prediction result and a putting target and budget of an advertiser, and personalized putting strategies are formulated for different user groups and time periods, so that the pertinence and effect of advertisements are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of artificial intelligence, and specifically refers to a method and system platform for predicting and placing traffic advertisements based on AI artificial intelligence. Background Art

[0002] With the development of the Internet, digital advertising has become an important way for enterprises to promote products and services. However, traditional advertising placement methods often lack accuracy, resulting in waste of advertising resources and low return on investment.

[0003] However, there are still certain defects in the existing methods for predicting and placing traffic advertisements. Once the existing prediction models are deployed, it is difficult to flexibly adjust them according to actual needs. They lack an effective feature selection mechanism and have weak adaptability to newly emerging trends. When facing a rapidly changing market environment, the prediction accuracy of the existing prediction models may be affected. The existing advertising placement strategies tend to adopt a relatively fixed set of solutions, failing to fully consider the characteristics of different user groups and the changes in time periods, resulting in lack of pertinence in advertising placement. There is a lack of an effective real-time monitoring and feedback mechanism, and it is impossible to obtain advertising performance data and user interaction data in a timely manner. Therefore, a method and system platform for predicting and placing traffic advertisements based on AI artificial intelligence are proposed. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system platform for predicting and placing traffic advertisements based on AI artificial intelligence to solve the problems raised in the above background art.

[0005] To achieve the above purpose, the present invention provides the following technical solutions: A method for predicting and placing traffic advertisements based on AI artificial intelligence, including the following steps:

[0006] S1. Obtain user behavior data, network traffic data, and advertisement-related data from multiple data sources;

[0007] S2. Perform data preprocessing on the obtained original data;

[0008] S3. Obtain the preprocessed data for feature extraction and screening;

[0009] S4. Perform traffic prediction based on the extracted and screened features;

[0010] S5. According to the traffic prediction results, formulate an optimal advertising placement strategy in combination with the advertising owner's placement goals and budget;

[0011] S6. Real-time monitor the placement effect of the advertisement and collect user feedback data;

[0012] S7. Feed back the obtained feedback data to the prediction model for dynamic adjustment.

[0013] Among them, for S1, obtain user behavior data, network traffic data, and advertisement-related data from multiple data sources; scrape data from various online platforms through API interfaces, obtain real-time network traffic data through traffic statistics of network devices, obtain data such as the display times, click-through rates, and conversion rates of advertisements from advertisement platforms, set an interval time for data collection, and after the data collection is completed, perform integrity checks on the collected data, and perform accuracy verification on the collected data by establishing data verification rules.

[0014] Among them, for S2, obtain the collected data and perform data cleaning, data transformation, and data integration preprocessing. Data cleaning includes duplicate data identification and removal, missing value processing, and outlier detection and correction. For duplicate data identification and removal, for the collected user behavior data, network traffic data, and advertisement-related data, identify completely duplicate records through data comparison. For outlier detection and correction, in the network traffic data, identify abnormal traffic values that are significantly different from other traffic data points through a clustering analysis algorithm, and communicate with the network service provider to confirm whether it is caused by reasons such as network failures.

[0015] Among them, for S2, data transformation includes numerical data standardization, categorical data encoding, and time series data processing. Numerical data standardization converts the data into standard normal distribution data with a mean of 0 and a standard deviation of 1 through a standardization method. Categorical data encoding converts categorical data such as gender and region in user behavior data, and advertisement type and industry in advertisement-related data using one-hot encoding technology. Time series data processing: For time series data such as search time in user behavior data, collection time in network traffic data, and exposure time in advertisement delivery effect data, extract time features; data integration integrates user behavior data, network traffic data, and advertisement-related data from different data sources according to predefined data association rules, and stores the preprocessed data in a database.

[0016] Among them, for S3, obtain the preprocessed data for feature extraction and screening; obtain the preprocessed data, extract features such as user activity, user interest preferences, and user purchase frequency from user behavior data, extract the type, creativity, and visual elements of advertisements from advertisement content, extract features such as advertisement delivery time, delivery channels, delivery budget, and delivery frequency from advertisement delivery data, extract the trend of traffic, analyze the source of traffic, and extract corresponding features from network traffic data, normalize numerical features, convert categorical features into one-hot encoding, and for ordered categorical features, label encoding can be performed. Calculate the correlation between each feature and the target variable through the Pearson correlation coefficient, and through correlation analysis, screen out features that are closely related to the target variable.

[0017] Among them, in S3, analysis of variance is used to evaluate whether the differences in different eigenvalues on the target variable are significant. If, for a certain feature, the means of the target variable show significant differences under different values, then this feature has a strong influence on the target variable. The importance of the features is evaluated through the random forest algorithm, ranked according to the evaluation results, and the top key features are retained. Combining the professional knowledge and business experience in the field of advertising placement, the features are manually reviewed and screened.

[0018] Among them, in S4, traffic prediction is performed based on the extracted and screened features; the extracted feature x n is used for prediction through linear regression, and the implementation formula is:

[0019] y = β0 + β1x1 + β2x2 +... + β n x n + ∈,

[0020] In the formula, y represents the traffic prediction result, x n represents the input feature, β0, β1,..., β n represent the model parameters, and ∈ represents the error term.

[0021] Among them, in S5, based on the traffic prediction result, the optimal advertising placement strategy is formulated by combining the advertising owner's placement target and budget; the traffic prediction result is obtained to analyze the network traffic trends of the placement target, different time periods, different regions, and different user groups. According to the target audience characteristics of the advertising owner and combining the user group traffic trends in the traffic prediction, the key user groups for placement are determined. Referring to the traffic distribution of different channels in the traffic prediction result and combining the active channels of the advertising owner's target audience, appropriate advertising placement channels are selected. According to the traffic prediction result and user behavior data analysis, the target audience is segmented into different groups, and personalized advertising content and placement strategies are formulated for each group. According to the peak and trough periods of the traffic prediction, the advertising placement time is determined, the formulated personalized strategy is executed, and real-time monitoring is carried out.

[0022] Among them, in S7, the obtained feedback data is fed back to the prediction model for dynamic adjustment. Real-time collection of relevant feedback on advertising performance data and user interaction data, and the real-time feedback data is fed back to the prediction module. The prediction model optimizes the parameter information according to the feedback data, and the optimized result is fed back to S5 to dynamically adjust the advertising placement decision.

[0023] It includes a data collection module, a data preprocessing module, a feature extraction module, a traffic advertising prediction module, an advertising placement decision module, a real-time monitoring feedback module, and a dynamic advertising placement optimization module.

[0024] Compared with the prior art, the beneficial effects of the present invention are:

[0025] 1. The present invention enhances the effectiveness of model training by extracting and screening features. Through in-depth analysis, key features such as user activity and interest preferences are extracted, and numerical features are normalized while categorical features are encoded. The importance of features is evaluated using the Pearson correlation coefficient and random forest, improving the accuracy and efficiency of the model.

[0026] 2. The present invention accurately predicts traffic through a traffic advertising prediction module based on the extracted and screened features. The prediction model can be flexibly extended and optimized according to actual needs to improve prediction performance.

[0027] 3. The present invention formulates an optimal advertising placement strategy through an advertising placement decision-making module based on the traffic prediction results, the advertising owner's placement objectives and budget, including placement time, placement channels, placement content, and placement frequency, and formulates personalized placement strategies for different user groups and time periods to improve the targeting and effectiveness of advertisements.

[0028] 4. The present invention optimizes and adjusts the prediction model according to the data feedback from real-time monitoring through a dynamic advertising placement optimization module to improve prediction accuracy, and dynamically optimizes the advertising placement strategy through intelligent algorithms to ensure the maximization of advertising placement effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is the flow chart of the traffic advertising prediction and placement method based on AI artificial intelligence of the present invention Figure 1 ;

[0030] Figure 2 is the flow chart of the traffic advertising prediction and placement method based on AI artificial intelligence of the present invention Figure 2 ;

[0031] Figure 3 is the flow chart of the traffic advertising prediction and placement method based on AI artificial intelligence of the present invention Figure 3 ;

[0032] Figure 4 is the schematic diagram of the traffic advertising prediction and placement system platform based on AI artificial intelligence of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0033] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0034] Embodiment

[0035] Please refer to Figures 1-4 as shown in the figure, the present invention provides a technical solution, including the following steps:

[0036] S1. Obtain user behavior data, network traffic data, and advertisement-related data from multiple data sources;

[0037] S2. Perform data preprocessing on the obtained original data;

[0038] S3. Obtain the preprocessed data for feature extraction and screening;

[0039] S4. Perform traffic prediction based on the extracted and screened features;

[0040] S5. According to the traffic prediction results, formulate an optimal advertisement placement strategy in combination with the advertiser's placement goals and budget;

[0041] S6. Real-time monitor the placement effect of the advertisement and collect user feedback data;

[0042] S7. Feed the obtained feedback data back to the prediction model for dynamic adjustment.

[0043] Among them, for S1, to obtain user behavior data, network traffic data, and advertisement-related data from multiple data sources; data is scraped from various online platforms through an API interface, real-time network traffic data is obtained through traffic statistics of network devices, data such as the display times, click-through rate, and conversion rate of advertisements are obtained from advertisement platforms, data collection is performed at set intervals, and after the data collection is completed, the integrity of the collected data is checked, and the accuracy of the collected data is verified by establishing data verification rules.

[0044] Among them, for S2, the collected data is obtained and data cleaning, data transformation, and data integration preprocessing are performed. Data cleaning includes duplicate data identification and removal, missing value processing, and outlier detection and correction. For duplicate data identification and removal of the collected user behavior data, network traffic data, and advertisement-related data, completely duplicate records are identified through data comparison. For outlier detection and correction, in the network traffic data, through a clustering analysis algorithm, outlier traffic values that are significantly different from other traffic data points are identified, and through communication and confirmation with the network service provider, it is judged whether it is caused by reasons such as network failures.

[0045] Among them, for S2, data conversion includes numerical data standardization, categorical data encoding, and time series data processing. Numerical data standardization uses a standardization method to convert data into standard normal distribution data with a mean of 0 and a standard deviation of 1. Categorical data encoding uses one-hot encoding technology to convert categorical data such as gender and region in user behavior data, and advertisement type and industry in advertisement-related data. Time series data processing: For time series data such as search time in user behavior data, collection time in network traffic data, and exposure time in advertisement delivery effect data, time features are extracted. Data integration integrates user behavior data, network traffic data, and advertisement-related data from different data sources according to predefined data association rules, and stores the preprocessed data in a database.

[0046] Among them, for S3, the preprocessed data is obtained for feature extraction and screening; the preprocessed data is obtained, and user activity, user interest preferences, and user purchase frequency features are extracted from user behavior data, the type, creativity, and visual elements of advertisements are extracted from advertisement content, the delivery time, delivery channel, delivery budget, and delivery frequency features of advertisements are extracted from advertisement delivery data, the trend of traffic is extracted from network traffic data, the source of traffic is analyzed, and corresponding features are extracted. Numerical features are normalized, and categorical features are converted into one-hot encoding. For ordered categorical features, label encoding can be performed. The correlation between each feature and the target variable is calculated through the Pearson correlation coefficient. Through correlation analysis, features closely associated with the target variable are screened out.

[0047] Among them, for S3, analysis of variance is used to evaluate whether the differences in different feature values on the target variable are significant. If the means of the target variable show significant differences for a certain feature under different values, then this feature has a strong influence on the target variable. The importance of features is evaluated through the random forest algorithm, ranked according to the evaluation results, and the top key features are retained. Combining professional knowledge and business experience in the field of advertisement delivery, the features are manually reviewed and screened.

[0048] Among them, for S4, traffic prediction is performed based on the extracted and screened features; the extracted feature x n is used for prediction through linear regression, and the implementation formula is:

[0049] y = β0 + β1x1 + β2x2 +... + β n x n + ∈,

[0050] In the formula, y represents the traffic prediction result, x n represents the input feature, β0, β1,..., β n represents the model parameters, and ∈ represents the error term.

[0051] Among them, S5 formulates the optimal advertising placement strategy according to the traffic prediction result, in combination with the advertising owner's placement target and budget; obtains the traffic prediction result, analyzes the network traffic trends of the placement target, different time periods, different regions, and different user groups, determines the user groups to be focused on according to the target audience characteristics of the advertising owner and in combination with the user group traffic trends in the traffic prediction, selects the appropriate advertising placement channels by referring to the traffic distribution of different channels in the traffic prediction result and in combination with the active channels of the advertising owner's target audience, divides the target audience into different groups according to the traffic prediction result and user behavior data analysis, formulates personalized advertising content and placement strategies for each group, formulates the advertising placement time according to the peak and trough periods of the traffic prediction, executes the formulated personalized strategy, and monitors it in real time.

[0052] Among them, S7 feeds the obtained feedback data back to the prediction model for dynamic adjustment, collects relevant feedback on advertising performance data and user interaction data in real time, feeds the real-time feedback data back to the prediction module, the prediction model optimizes the parameter information according to the feedback data, and the optimized result is fed back to S5 to dynamically adjust the advertising placement decision.

[0053] It includes a data collection module, a data preprocessing module, a feature extraction module, a traffic advertising prediction module, an advertising placement decision module, a real-time monitoring and feedback module, and a dynamic advertising placement optimization module.

[0054] Working principle: Data is scraped from various online platforms through API interfaces, and real-time network traffic data is obtained through traffic statistics of network devices; data such as the display times, click-through rates, and conversion rates of advertisements are obtained from advertising platforms. Data collection is performed at set intervals. After the collection is completed, the integrity of the data is checked and the accuracy is verified. The collected data is cleaned, transformed, and integrated. Data cleaning identifies and removes duplicate data, handles missing values and outliers. Data transformation standardizes numerical data into standard normal distribution data with a mean of 0 and a standard deviation of 1; for categorical data, one-hot encoding technology is used for transformation; for time series data, time features are extracted. Data integration integrates data from different data sources according to predefined data association rules and stores it in a database. Features such as user activity, interest preferences, and purchase frequency are extracted from user behavior data; features such as advertisement types, creativity, and visual elements are extracted from advertisement content; features such as placement time, channels, budget, and frequency are extracted from advertisement placement data; features such as traffic trends and sources are extracted from network traffic data. The extracted features are normalized, and categorical features are converted into one-hot encoding or label encoding. The Pearson correlation coefficient is used to calculate the correlation between each feature and the target variable, and features closely associated with the target variable are selected. The importance of features is evaluated through analysis of variance and random forest algorithms. Manual review and screening are carried out in combination with professional knowledge and business experience. Machine learning algorithms such as linear regression are used for traffic prediction. A prediction model is established through the relationship between input features and the target variable, and the traffic prediction results are output. According to the traffic prediction results, an optimal advertisement placement strategy is formulated in combination with the advertisement owner's placement goals and budget. The network traffic trends in different time periods, different regions, and different user groups are analyzed. According to the characteristics of the advertisement owner's target audience, combined with the user group traffic trends in the traffic prediction, the user groups for key placement are determined. Referring to the traffic distribution of different channels in the traffic prediction results, combined with the active channels of the advertisement owner's target audience, appropriate advertisement placement channels are selected. The target audience is segmented into different groups, and personalized advertisement content and placement strategies are formulated for each group. According to the peak and trough periods of the traffic prediction, the advertisement placement time is determined. The placement effect of the advertisement is monitored in real time, and user feedback data is collected. Through the real-time monitoring module, advertisement performance data and user interaction data are collected, and the real-time feedback data is input into the prediction model. The model optimizes the parameter information according to the feedback data, and the optimized results are fed back to the advertisement placement decision-making module to dynamically adjust the advertisement placement decision.

[0055] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

[0056] The above description of the present invention and its implementation manners is not restrictive. What is shown in the drawings is only one of the implementation manners of the present invention, and the actual structure is not limited thereto. In general, if those of ordinary skill in the art are inspired by it and design, without creative efforts, structural manners and embodiments similar to the technical solution without departing from the gist of the present invention, they shall fall within the protection scope of the present invention.

Claims

1. A traffic advertising prediction and delivery method based on AI artificial intelligence, characterized in that: The following steps are involved: S1. Obtain user behavior data, network traffic data, and advertising-related data from multiple data sources; S2. Perform data preprocessing based on the acquired raw data; S3, obtaining the preprocessed data for feature extraction and screening; S4, predicting the flow rate based on the extracted and screened features; S5. Develop the optimal advertising strategy based on traffic forecast results and advertiser’s advertising goals and budget; S6. Monitor the advertising effect in real time and collect user feedback data; S7. Feedback the acquired feedback data to the prediction model for dynamic adjustment.

2. The method for predicting and delivering traffic advertisements based on AI artificial intelligence according to claim 1 is characterized by: S1, obtaining user behavior data, network traffic data and advertising-related data from multiple data sources; Capture data from various online platforms through the API interface, obtain real-time network traffic data through the traffic statistics of network devices, obtain advertising display times, click-through rates and conversion rate data from advertising platforms, set intervals for data collection, and perform integrity checks on the collected data after data collection is completed. Verify the accuracy of the collected data by establishing data verification rules.

3. The method for predicting and delivering traffic advertisements based on AI artificial intelligence according to claim 1 is characterized in that: The S2 obtains the collected data and performs data cleaning, data conversion and data integration preprocessing. Data cleaning includes duplicate data identification and removal, missing value processing and outlier detection and correction. Duplicate data identification and removal of collected user behavior data, network traffic data and advertising-related data, through data comparison, identify completely duplicate records, outlier detection and correction, in the network traffic data, through cluster analysis algorithm, identify abnormal traffic values ​​that are significantly different from other traffic data points, and communicate with the network service provider to confirm whether it is caused by network failure or other reasons.

4. The method for predicting and delivering traffic advertisements based on AI artificial intelligence according to claim 3 is characterized in that: The data conversion in S2 includes numerical data standardization, categorical data encoding and time series data processing. The numerical data standardization converts the data into standard normal distribution data with a mean of 0 and a standard deviation of 1 through a standardization method. The categorical data encoding converts the gender and region in the user behavior data, the advertisement type and the industry in the advertisement-related data using a unique hot encoding technique. The time series data processing: extracts time features from the time series data such as the search time in the user behavior data, the collection time in the network traffic data and the exposure time in the advertisement delivery effect data. Data integration integrates user behavior data, network traffic data, and advertising-related data from different data sources according to pre-defined data association rules, and stores the pre-processed data in the database.

5. The method for predicting and delivering traffic advertisements based on AI artificial intelligence according to claim 1 is characterized in that: S3, obtaining the preprocessed data for feature extraction and screening; Obtain preprocessed data, extract user activity, user interest preferences and user purchase frequency features from user behavior data, extract advertisement type, advertisement creativity and advertisement visual elements from advertisement content, extract advertisement delivery time, delivery channel, delivery budget and delivery frequency features from advertisement delivery data, extract traffic trends from network traffic data, analyze the source of traffic, and extract corresponding features, normalize numerical features, convert categorical features into one-hot encoding, perform label encoding on ordered categorical features, calculate the correlation between each feature and the target variable through Pearson correlation coefficient, and screen out features that are closely related to the target variable through correlation analysis.

6. The method for predicting and delivering traffic advertisements based on AI artificial intelligence according to claim 5 is characterized in that: S3, through variance analysis, evaluates whether the differences between different feature values ​​on the target variable are significant. If the mean of the target variable shows significant difference under different values ​​of a certain feature, then the feature has a strong influence on the target variable. The importance of the feature is evaluated by the random forest algorithm, and the features are ranked according to the evaluation results. The top key features are retained, and the features are manually reviewed and screened in combination with professional knowledge and business experience in the field of advertising.

7. The method for predicting and delivering traffic advertisements based on AI artificial intelligence according to claim 1 is characterized in that: S4: perform flow prediction based on the extracted and screened features; obtain the extracted features x n , prediction is performed through linear regression, and the implementation formula is: y=β0+β1x1+β2x2+...+β n x n +∈, In the formula, y represents the traffic prediction result, x n Represents the input features, β0, β1, ..., β n represents the model parameters, ∈ represents the error term.

8. The method for predicting and delivering traffic advertisements based on AI artificial intelligence according to claim 1 is characterized by: S5, according to the traffic prediction results, formulate the optimal advertising delivery strategy in combination with the advertising goals and budget of the advertiser; obtain the traffic prediction results to analyze the delivery goals, network traffic trends in different time periods, different regions and different user groups, determine the user groups for key delivery according to the characteristics of the advertiser's target audience and the traffic trends of the user groups in the traffic prediction, refer to the traffic distribution of different channels in the traffic prediction results, and select appropriate advertising delivery channels in combination with the active channels of the advertiser's target audience; according to the traffic prediction results and user behavior data analysis, segment the target audience into different groups, formulate personalized advertising content and delivery strategies for each group, formulate advertising delivery time according to the peak and trough periods of traffic prediction, execute the formulated personalized strategy and monitor in real time.

9. The method for predicting and delivering traffic advertisements based on AI artificial intelligence according to claim 1 is characterized in that: S7, based on the feedback data obtained, feeds back to the prediction model for dynamic adjustment, collects advertising performance data and user interaction data related feedback in real time, feeds back the real-time feedback data to the prediction module, and the prediction model optimizes parameter information based on the feedback data. The optimized results are fed back to S5 to dynamically adjust the advertising delivery decision.

10. The AI-based traffic advertising prediction and delivery system platform implemented by the method of claim 1 is characterized by: It includes data collection module, data preprocessing module, feature extraction module, traffic advertising prediction module, advertising delivery decision module, real-time monitoring feedback module and dynamic advertising delivery optimization module.

Images