CN118096266A - Intelligent advertising marketing system and method based on Internet - Google Patents

Abstract

The invention relates to the technical field of advertisement marketing, in particular to an intelligent advertisement marketing system and method based on the Internet, comprising a user analysis management unit, a data collection module, a user portrait construction module and a user portrait construction module, wherein the user analysis management unit is used for collecting user behavior data, the data analysis module is used for analyzing the user behavior data, and the user portrait construction module is used for classifying user groups according to user labels based on the result output by the data analysis module; the advertisement resource management unit is used for managing and distributing advertisement space resources for delivery; the targeted delivery unit matches advertisements with target audiences based on the user portraits, and adopts a targeted delivery algorithm to formulate an optimal advertisement delivery strategy; the putting optimization unit adjusts the bidding strategy based on the maximum expected benefits according to the net benefits of the advertisement position on each media platform resource, accurately reflects the liveness, loyalty and short-term buying habits of the user based on the buying frequency of the user, provides a basis for formulating a periodic marketing strategy, and further achieves accurate positioning of target audiences.

Description

Intelligent advertising marketing system and method based on Internet

Technical Field

The present invention relates to the field of advertising and marketing technology, and in particular to an intelligent advertising and marketing system and method based on the Internet.

Background Art

The Internet's intelligent advertising and marketing system is an advanced form of online advertising service. It uses advanced technologies such as big data, artificial intelligence (AI), and machine learning to create a complex system that integrates functions such as precise advertising targeting, intelligent creative generation, and automated delivery and optimization.

Intelligent advertising and marketing is highly dependent on machine learning and artificial intelligence algorithms, which may have a black box effect, that is, the decision-making process is difficult to explain, and sometimes advertisements may be mistakenly pushed to inappropriate target groups. In view of this, an Internet-based intelligent advertising and marketing system and method are provided.

Summary of the invention

The purpose of the present invention is to provide an Internet-based intelligent advertising and marketing system and method to solve the problem that the intelligent advertising and marketing proposed in the above background technology is highly dependent on machine learning and artificial intelligence algorithms, and these algorithms may have a black box effect, that is, the decision-making process is difficult to explain, and sometimes advertisements may be mistakenly pushed to inappropriate target groups.

To achieve the above-mentioned purpose, the present invention aims to provide an intelligent advertising and marketing system based on the Internet, including a user analysis management unit, wherein the user analysis management unit collects user behavior data through a data collection module, and the data analysis module analyzes the user behavior data to identify the user's consumption frequency, content preference, and product relevance characteristics. The user portrait construction module generates user tags based on the results output by the data analysis module, and classifies user groups according to the user tags to form different user group categories;

An advertisement resource management unit, the advertisement resource management unit is used to manage and allocate advertisement space resources for delivery, and the advertisement resource management unit includes an advertisement resource library, a media resource integration module and an advertisement resource management module;

The advertising resource management module (23) uses an advertising resource super-optimization algorithm to allocate advertising slots to advertisements in the advertising resource library (21) according to the traffic characteristics and values of different media channels based on the user portrait generated by the user analysis management unit (1);

A targeted delivery unit, which matches advertisements with target audiences based on user portraits, and uses a targeted delivery algorithm to formulate an optimal advertisement delivery strategy in combination with advertisement goals and budgets;

Considering the time sensitivity of user behavior, users' purchasing needs and interest preferences usually have a certain timeliness. Recent purchasing behavior may reflect the user's current hot spots of interest or demand trends. Pushing relevant advertising content is more likely to attract users' attention and response. At the same time, placing ads based on the user's active time period can increase the possibility of seeing the ads and interacting with them. The targeted delivery algorithm is optimized by introducing the recent active time period factor. The optimized targeted delivery algorithm is:

;

In the formula, Represents user portrait; Indicates the characteristic description of the advertisement; Represents the optimized user portrait and ad characterization The matching score between Represents the personalized attributes in the user portrait; Indicates the target population attributes in the advertising targeting strategy; Represents a collection of user's interest tags; Represents a set of interest keywords related to advertisements; Represents the user's historical consumption behavior data; Indicates the product category the ad is associated with; Represents the similarity function between two attributes; , and Respectively represent the weight factors of different matching dimensions; The time sensitivity of user behavior; Indicates the time information of advertisement delivery; Indicates the weight of the impact of the time sensitivity of user behavior on the matching results, and indicates the importance of time-related factors such as recent purchase behavior and active time periods in targeted delivery.

when When the value of is large, it means that time sensitivity has a greater impact on the matching results, and the model pays more attention to the user's recent behavior or active period; When the value of is small, the time factor has less impact on the matching results, and the model relies more on the matching results of other dimensions. Accurately push the latest user behavior, which can effectively improve the click-through rate and conversion rate of ads. By analyzing the user's active time period, advertisers can adjust the advertising delivery strategy according to the user traffic and activity in different time periods to achieve the optimal use of advertising budget.

A delivery optimization unit adjusts the bidding strategy of advertisements on each media platform resource based on the delivery optimization algorithm according to the net revenue of the advertisement positions on each media platform resource, and adopts a cross-platform joint optimization algorithm to track the behavior paths of users on different platforms, thereby optimizing the advertisement delivery strategy and achieving the overall optimal cross-platform advertising effect.

As a further improvement of the technical solution, the data analysis module analyzes the user behavior data based on the user behavior analysis algorithm, and the specific expression involved is:

User consumption frequency:

If the user set is U, user , where the time window , then the specific expression of user consumption frequency is: ;

In the formula, Indicates user Frequency of purchases in the past 7 days; Indicates user At the point in time The number of purchases made; Represents the time window, indicating the time from Starting from the day, the next 7 consecutive days; Indicates time point Belongs to the time window ;

User content preferences:

If there is a set of content to be interacted , where the content of user interaction , then the specific expression of user content preference is: ;

In the formula, Represents the content of user interaction; is the indicator function, indicating that at the time point ,user Whether it is related to the content There is interaction, if there is a better ,otherwise ; Indicates user About content degree of preference;

Product Relevance:

If there are two products E and F, their support and confidence , then the correlation between products E and F is calculated as: ;

In the formula, It indicates the increase multiple of the probability of purchasing product F compared to the base probability of purchasing product F when it is known that product E has been purchased.

Lift To help analysts understand the degree of correlation between two products:

If the lift is greater than 1, it means that there is a positive correlation between product E and product F, and customers who buy product E are more likely to buy product F;

If the lift is equal to 1, it means there is no association between product E and product F;

If the lift is less than 1, it means that there is a negative correlation between product E and product F, and customers who buy product E are less likely to buy product F.

As a further improvement of the technical solution, the specific steps involved in the user portrait construction module generating user tags are:

S3.1. Analyze the user behavior data output from the data analysis module to obtain consumption frequency A, content preference B, and product relevance characteristics C;

S3.2. Define a set of user tags based on business needs and analysis results, and map consumption frequency A, content preference B, and product relevance characteristics C to the defined user tags;

The specific mapping rules are as follows:

Consumption frequency A mapping rules:

If the user's purchase frequency in the past 7 days If it is high, it can be mapped to the label of "high frequency consumer" ;

If the purchase frequency is moderate, it can be mapped to the "regular consumer" label ;

If the purchase frequency is low, it may be mapped to the "low-frequency consumer" label ;

Content preference B mapping rules:

Based on user's content Degree of preference If a user frequently interacts with a certain type of content, then this type of content preference is mapped to the corresponding interest tag On, among them, Represents an index variable representing different interest types, such as Express interest in technology, Indicates fashion trends;

Product association feature C mapping rules:

If the user buys product E and the lift between product E and F Higher, such users can be mapped to the "Associated Product Purchaser" label ;

If users tend to buy complete sets of products or highly complementary product combinations, and often buy product B right after purchasing product A, users with this behavior pattern can be mapped to the "Package Buyer" label. ;

Based on the above mapping rules, the user portrait construction module can accurately label users in a diversified manner based on their performance in different dimensions, which helps the advertising system to perform refined operations for different user groups.

S3.3. Based on the defined user tags, user tag classification is used to classify user groups and divide users into groups with similar characteristics. Dividing users into groups with similar characteristics can achieve accurate positioning of target audiences, which helps to allocate advertising budgets more effectively and reduce ineffective exposure and waste. By identifying users' common interests, behavior patterns or consumption habits, the system can push products, content or services that they may be interested in to various user groups in a targeted manner, increase click-through rate and conversion rate, and provide enterprises with a powerful tool to design promotion strategies for cross-selling, bundling or other related product combinations.

In the above steps, the specific steps involved in user tag classification are:

S3.31. Randomly select K users as the initial cluster centers, denoted as ;

S3.32. For each user , calculate the distance between it and the K cluster centers, and its distance expression is:

;

In the formula, The standard configuration is the feature dimension; Indicates user In the The value of a feature; Represents the cluster center In the The value of a feature;

The user Assign to the cluster corresponding to the nearest cluster center:

That is ;

In the formula, Indicates user The index of the cluster to which the user should be assigned. The final cluster category number; For a given user , find a cluster K such that the distance between the user and the cluster center is the smallest; K represents the total number of clusters; Indicates users or data points; represents the cluster center of the Kth cluster; for user Assign a cluster label , so that it corresponds to the cluster center closest to the user , in each iteration, each user will be reallocated to the nearest cluster according to this principle until the cluster center no longer changes significantly;

S3.33. Calculate the average value of all users in each cluster and use it as the new cluster center;

Among them, the update formula of the new cluster center is:

;

In the formula, represents the set of all users in the currently assigned cluster K; Indicates that for all users in the current cluster K The new cluster center is obtained by averaging the characteristic vectors of represents the number of users in cluster K;

The cluster center is made as close as possible to the members of the cluster to which it belongs, so as to optimize the clustering effect. In each iteration, the algorithm converges to a local optimal solution by continuously updating the cluster center and redistributing the sample points.

S3.34, make convergence judgment:

If the cluster center reaches the preset threshold or the clustering result remains unchanged after several consecutive iterations, the iteration is stopped;

Otherwise, return to step S3.32 to continue to reallocate users and update cluster centers;

S3.34. After the iteration is completed, the final clustering result is obtained, that is, the cluster to which each user belongs, and user portraits can be constructed based on these clusters;

By continuously adjusting the position of the cluster center, the user behavior data in each cluster is made as close as possible, thereby achieving effective segmentation of user groups.

S3.4. Integrate user tags and group information to generate user portraits. User portraits include the user's basic information, interests and hobbies, and purchasing behavior characteristics to form a comprehensive user description. Each user portrait represents a comprehensive and personalized user description, which is used in application scenarios such as personalized recommendations, precision marketing, and customer relationship management.

As a further improvement of the technical solution, the advertising resource library is used to store all advertising resources for delivery;

The media resource integration module is connected to various media platforms through an API interface to manage media platform resources used for advertising, thereby achieving unified management and scheduling of advertising resources across platforms;

The advertising resource management module allocates advertising slots to advertisements in the advertising resource library according to the traffic characteristics and values of different media channels, based on the user portrait generated by the user analysis management unit and using an advertising resource super-optimization algorithm.

As a further improvement of the technical solution, the advertising resource super optimization algorithm is specifically as follows:

;

In the formula, Indicates Advertisement slot in Exposure on media platforms; Indicates Advertisement slot in Click-through rate on media platform resources; Indicates Advertisement slot in Conversion rate on media platform resources; Indicates Advertisement slot in Average revenue per user generated by media platform resources; Indicates Advertisement slot in Cost per thousand impressions on each media platform property; Indicates the number of ad slots; Indicates the number of media platform resources; It represents the net revenue obtained by deducting the sum of the corresponding delivery costs from the sum of the expected revenue generated by all advertising positions on various media platform resources;

The constraints are:

The total budget constraint is: ;

In the formula, represents the total budget;

Non-negativity constraints:

;

In the formula, Expressing to all and all , the above inequalities must all hold;

Time period delivery quota constraints:

;

In the formula, Indicates A collection of all advertisements included in a time period; Indicates time period budget ceiling within the Indicated in The total exposure of all advertisements on various media platform resources in a time period; Expressing to all , the above inequalities must all hold.

As a further improvement of the technical solution, the directional delivery unit includes a matching delivery module. The matching delivery module is based on a directional delivery algorithm, accurately matches the user portrait with the advertising positioning strategy, and pushes the most suitable advertising content to the corresponding user. The specific expression of the directional delivery algorithm is:

;

The most suitable advertising content is determined based on the user's characteristics and the advertising positioning strategy; the most suitable advertising content is determined by comprehensively considering multiple factors such as the user's personalized attributes, interest tags, historical consumption behavior, etc., by calculating the similarity between them and multiplying them by the corresponding weight factor. The weight factor can be adjusted according to the specific situation so that the matching priority can be adjusted according to different situations.

As a further improvement of the technical solution, the delivery optimization unit includes a data monitoring module, a target setting module and a bidding strategy optimization module;

Among them, the data monitoring module is used to collect and process data related to advertising delivery, the target setting module determines the optimization target based on the cross-platform joint optimization algorithm, and the bidding strategy optimization module adjusts the bidding strategy of advertisements on each media platform resource based on the expected revenue and target setting through the delivery optimization algorithm based on the optimization target. The delivery optimization algorithm dynamically adjusts the bidding strategy based on the expected revenue and real-time data to achieve the best delivery effect.

As a further improvement of the technical solution, the specific expression of the delivery optimization algorithm is:

;

In the formula, Represents the global bid coefficient, which is used to uniformly adjust the cost ratio of all resources; Indicated in The amount of ads delivered on a media resource, that is, the number of times an ad is displayed and clicked on that resource; Indicated in Average revenue on each media source; Indicates the number of media resources; is an index variable, which is used to traverse each media resource from 1 to M; Indicated in Average cost per media resource;

Among them, the constraint condition is that the total cost does not exceed the budget :

;

In the formula, represents the maximum total cost allowed;

By adjusting the global bid modifier and the amount served on each property , to maximize the total net benefit;

Based on the optimization goal of maximizing total revenue, then:

;

In the formula, It means that maximizing the total benefit is taken as the optimization goal;

Introduce constraints into the objective function to form an objective function to dynamically adjust the global bid coefficient and the amount served on each property :

;

In the formula, represents the Lagrangian function;

By corresponding and Find the partial derivatives and set them to 0 to find the extreme points:

;

;

Solve the above system of equations, for Amount of delivery on media properties :

;

For global bid modifiers :

;

In the formula, represents the Lagrange multiplier, which is used to deal with the constraints of the problem; Denotes the Lagrangian function L for the amount of delivery The partial derivative of Represents the Lagrangian function L for the global bid coefficient The partial derivative of .

As a further improvement of the technical solution, the specific expression involved in the cross-platform joint optimization algorithm is:

Media Resources Exposure to media resources The impact of clicks is:

;

Media Resources Exposure to media resources The impact of clicks is:

;

In the formula, Indicates that the ad is in the media resource Exposure to media resources The impact of clicks; Indicated in media resources Click-through rate; Indicated in media resources Click-through rate; Indicated in media resources After exposure in media resources Increased probability of clicks; Indicated in media resources After exposure in media resources Increased probability of clicks; Indicated in media resources Number of exposures; Indicated in media resources Number of exposures; The intensity coefficient indicating the influence; The intensity coefficient indicating the influence;

Then the joint optimization objective function K is:

;

In the formula, Represents media resources In the amount of delivery and bid modifier Independent advertising effect income under Indicates the delivery volume of each media resource in its respective and bid modifier The sum of direct advertising effect benefits generated under and All are index variables; Represents media resources Media Resources The weight coefficient between ; Represents media resources Media Resources The weight coefficient between ; It is the overall optimization goal, which is used to maximize the function value to achieve the best advertising effect across media resource platforms.

By considering the synergy between platforms, the exposure between platforms can increase the probability of users clicking on other platforms. This synergy helps to improve the overall advertising effect. It comprehensively considers the benefits of independent delivery on each platform and the gains from mutual influence between platforms to achieve the overall optimal cross-platform advertising effect.

On the other hand, the present invention provides an Internet-based intelligent advertising and marketing method, which is used in the above-mentioned Internet-based intelligent advertising and marketing system, and comprises the following steps:

S10.1. The data collection module collects user behavior data, and the data analysis module uses the user behavior analysis algorithm to calculate the user's consumption frequency, content preference and content interaction. The user portrait construction module generates user tags including consumption frequency, content preference and product relevance based on the user's behavior characteristics, and classifies the users into groups with similar characteristics to form user portraits.

S10.2. The media resource integration module connects multiple media platform resources to uniformly manage and schedule cross-platform advertising resources. The advertising resource management module uses the advertising resource super-optimization algorithm to maximize the expected revenue and dynamically adjust the advertising allocation strategy in different media channels, taking into account budget constraints and time period quota factors;

S10.3, and through the matching delivery module, use the targeted delivery algorithm to calculate the matching degree between the user portrait and the advertising features, combine personalized attributes, interest tags, and historical consumption behaviors for accurate matching, and formulate the best advertising delivery strategy to push the most suitable advertising content to the corresponding users, while considering time-sensitive factors such as the most recent active period and purchase behavior to improve the relevance and click-through conversion rate of the advertisement;

S10.4. Finally, the delivery optimization unit monitors the advertising delivery effect in real time, collects and processes data related to advertising clicks and conversions, sets the maximum total revenue as the optimization goal through the goal setting module, and uses the delivery optimization algorithm to dynamically adjust the global bid coefficient and the delivery volume on each media resource to ensure that the advertising benefits are maximized while meeting the budget constraints.

Compared with the prior art, the present invention has the following beneficial effects:

1. In the Internet-based intelligent advertising and marketing system and method, the data analysis module builds accurate user portraits based on user behavior analysis algorithms, identifies user consumption frequency, content preferences, and product relevance characteristics, and provides strong support for personalized advertising;

Also based on the user's purchase frequency in the past 7 days , accurately reflects the user's activity, loyalty and short-term purchasing habits, and provides a basis for formulating weekly or periodic marketing strategies; and adopts user label classification algorithm to group users, continuously optimizes clustering centers, so that users in each cluster have highly similar behavior patterns and consumption characteristics, thereby achieving accurate positioning of the target audience.

2. In the Internet-based intelligent advertising marketing system and method, the targeted delivery unit adopts a targeted delivery algorithm to formulate the optimal advertising delivery strategy, comprehensively considering the user's personalized attributes, interest tags, historical consumption behavior and time sensitivity factors, to ensure that the most suitable advertising content is pushed to the target user group at the right time, thereby significantly improving the advertising click-through rate and conversion rate;

The delivery optimization unit monitors the advertising delivery effect in real time, combines the preset optimization goals, and uses the delivery optimization algorithm to automatically adjust the bidding strategy and delivery volume to ensure that the delivery effect is maximized within a limited budget.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a flowchart of the overall process of the present invention.

The meaning of each number in the figure is:

1. User analysis management unit; 11. Data collection module; 12. Data analysis module; 13. User portrait construction module;

2. Advertising resource management unit; 21. Advertising resource library; 22. Media resource integration module; 23. Advertising resource management module;

3. Targeted delivery unit; 4. Delivery optimization unit.

DETAILED DESCRIPTION

The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

Example 1

Please refer to FIG1 , which provides an intelligent advertising and marketing system based on the Internet, including a user analysis management unit 1 , which collects user behavior data through a data collection module 11 ;

The data collection module 11 obtains the user's cookie information and device identifier; the user behavior data includes the user's browsing history, search history, purchase behavior, and social media interaction data on the APP;

The data analysis module 12 analyzes the user behavior data to identify the user's consumption frequency, content preference, and product relevance characteristics. The user portrait construction module 13 generates user tags based on the results output by the data analysis module 12, and classifies user groups according to the user tags to form different user group categories, so as to customize personalized advertising strategies and service plans for each group.

Collect user information through cookies, device identifiers, and user behavior data, analyze user attributes, interest preferences, and consumption habits based on the user information, form a user tag system, and complete the construction of user portraits, and the user analysis management unit 1 includes a data collection module 11, a data analysis module 12, and a user portrait construction module 13;

The data analysis module 12 analyzes the user behavior data based on the user behavior analysis algorithm, and the specific expression involved is:

User consumption frequency:

If the user set is U, user , where the time window , then the specific expression of user consumption frequency is:

;

In the formula, Indicates user Frequency of purchases in the past 7 days; Indicates user At the point in time The number of purchases made; Represents the time window, indicating the time from Starting from the day, the next 7 consecutive days; Indicates time point Belongs to the time window ;

pass Parameter indicators can help you understand the user's activity, loyalty and purchasing habits within a specific time window, and use 7 days as a time window. During this period, the user's purchasing habits, active time periods, interest preferences and other characteristics are relatively stable and easy to observe. By analyzing the consumption frequency within 7 days, you can identify the user's weekly consumption pattern, which provides a basis for formulating precision marketing strategies. For online advertising, 7 days is short enough to quickly evaluate the short-term effect of advertising. You can adjust the advertising content, delivery time and frequency in time according to the data changes in these 7 days to improve conversion rate and return on investment.

User content preferences:

If there is a set of content to be interacted , where the content of user interaction , then the specific expression of user content preference is:

;

In the formula, Represents the content of user interaction; is the indicator function, indicating that at the time point ,user Whether it is related to the content There is interaction, if there is a better ,otherwise ; Indicates user About content degree of preference;

The above expression can calculate the user's preference for each content, which can be used to build a user's content preference profile, thereby providing a basis for personalized advertising recommendations;

Product Relevance:

If there are two products E and F, their support and confidence , then the correlation between products E and F is calculated as: ;

In the formula, It indicates the increase multiple of the probability of purchasing product F compared to the base probability of purchasing product F when it is known that product E has been purchased.

Lift To help analysts understand the degree of correlation between two products:

If the lift is greater than 1, it means that there is a positive correlation between product E and product F, and customers who buy product E are more likely to buy product F;

If the lift is equal to 1, it means there is no association between product E and product F;

If the lift is less than 1, it means that there is a negative correlation between product E and product F, and customers who buy product E are less likely to buy product F.

By analyzing the user's purchase history data, the lift between different products can be calculated. Based on the correlation of products with high lift, the advertising system can more accurately target related product ads to potential customers, improving the effectiveness and conversion rate of advertising. By guiding advertising with lift, ads can be more accurately delivered to users who may be interested, thereby increasing the click-through rate and conversion rate of ads, reducing advertising costs, and thus improving the return on investment of ads.

The specific steps involved in the user portrait building module 13 generating user tags are:

S3.1. Analyze the user behavior data output from the data analysis module 12 to obtain consumption frequency A, content preference B, and product relevance feature C;

S3.2. Define a set of user tags based on business needs and analysis results, and map consumption frequency A, content preference B, and product relevance characteristics C to the defined user tags;

The specific mapping rules are as follows:

Consumption frequency A mapping rules:

If the user's purchase frequency in the past 7 days If it is high, it can be mapped to the label of "high frequency consumer" ;

If the purchase frequency is moderate, it can be mapped to the "regular consumer" label ;

If the purchase frequency is low, it may be mapped to the "low-frequency consumer" label ;

Content preference B mapping rules:

Based on user's content Degree of preference If a user frequently interacts with a certain type of content, then this type of content preference is mapped to the corresponding interest tag On, among them, Represents an index variable representing different interest types, such as Express interest in technology, Indicates fashion trends;

Product association feature C mapping rules:

If the user buys product E and the lift between product E and F Higher, such users can be mapped to the "Associated Product Purchaser" label ;

If users tend to buy complete sets of products or highly complementary product combinations, and often buy product B right after purchasing product A, users with this behavior pattern can be mapped to the "Package Buyer" label. ;

Based on the above mapping rules, the user portrait construction module 13 can accurately label users in a variety of ways based on their performance in different dimensions, which helps the advertising system to perform refined operations for different user groups.

S3.3. Based on the defined user tags, the user tag classification algorithm is used to classify user groups and divide users into groups with similar characteristics. Dividing users into groups with similar characteristics can achieve accurate positioning of target audiences, which helps to allocate advertising budgets more effectively and reduce ineffective exposure and waste. By identifying users' common interests, behavior patterns or consumption habits, the system can push products, content or services that they may be interested in to various user groups in a targeted manner, increase click-through rate and conversion rate, and provide companies with a powerful tool to design promotion strategies for cross-selling, bundling or other related product combinations.

In the above steps, the specific steps involved in user tag classification are:

S3.31. Randomly select K users as the initial cluster centers, denoted as ;

S3.32. For each user , calculate the distance between it and the K cluster centers, and its distance expression is:

;

In the formula, The standard configuration is the feature dimension; Indicates user In the The value of a feature; Represents the cluster center In the The value of a feature;

The user Assign to the cluster corresponding to the nearest cluster center:

That is ;

In the formula, Indicates user The index of the cluster to which the user should be assigned. The final cluster category number; For a given user , find a cluster K such that the distance between the user and the cluster center is the smallest; K represents the total number of clusters; Indicates users or data points; represents the cluster center of the Kth cluster; for user Assign a cluster label , so that it corresponds to the cluster center closest to the user , in each iteration, each user will be reallocated to the nearest cluster according to this principle until the cluster center no longer changes significantly;

S3.33. Calculate the average value of all users in each cluster and use it as the new cluster center;

Among them, the update formula of the new cluster center is:

;

In the formula, represents the set of all users in the currently assigned cluster K; Indicates that for all users in the current cluster K The new cluster center is obtained by averaging the characteristic vectors of represents the number of users in cluster K;

The cluster center is made as close as possible to the members of the cluster to which it belongs, so as to optimize the clustering effect. In each iteration, the algorithm converges to a local optimal solution by continuously updating the cluster center and redistributing the sample points.

S3.34, make convergence judgment:

If the cluster center reaches the preset threshold or the clustering result remains unchanged after several consecutive iterations, the iteration is stopped;

Otherwise, return to step S3.32 to continue to reallocate users and update cluster centers;

S3.34. After the iteration is completed, the final clustering result is obtained, that is, the cluster to which each user belongs, and user portraits can be constructed based on these clusters;

By continuously adjusting the position of the cluster center, the user behavior data in each cluster is made as close as possible, thereby achieving effective segmentation of user groups.

S3.4. Integrate user tags and group information to generate user portraits. User portraits include the user's basic information, interests and hobbies, and purchasing behavior characteristics to form a comprehensive user description. Each user portrait represents a comprehensive and personalized user description, which is used in application scenarios such as personalized recommendations, precision marketing, and customer relationship management.

The Internet-based intelligent advertising marketing system also includes an advertising resource management unit 2, which is used to manage and allocate advertising space resources for delivery, and the advertising resource management unit 2 includes an advertising resource library 21, a media resource integration module 22 and an advertising resource management module 23;

The advertising resource library 21 is used to store all advertising resources for delivery, including but not limited to website banner ads, social media interstitial ads, and APP splash screen ads;

The media resource integration module 22 connects with various media platforms through an API interface to manage media platform resources used for advertising, and realizes unified management and scheduling of advertising resources across platforms;

The advertising resource management module 23 allocates advertising slots to the advertisements in the advertising resource library 21 according to the traffic characteristics and values of different media channels, using an advertising resource super-optimization algorithm based on the user portrait generated by the user analysis management unit 1 .

In this embodiment, the advertising resource super optimization algorithm is specifically:

;

In the formula, Indicates Advertisement slot in Exposure on media platforms; Indicates Advertisement slot in Click-through rate on media platform resources; Indicates Advertisement slot in Conversion rate on media platform resources; Indicates Advertisement slot in Average revenue per user generated by media platform resources; Indicates Advertisement slot in Cost per thousand impressions on each media platform property; Indicates the number of ad slots; Indicates the number of media platform resources; It represents the net revenue obtained by deducting the sum of the corresponding delivery costs from the sum of the expected revenue generated by all advertising positions on various media platform resources;

The constraints are:

The total budget constraint is: ;

In the formula, represents the total budget;

Non-negativity constraints:

;

In the formula, Expressing to all and all , the above inequalities must all hold;

Time period delivery quota constraints:

;

In the formula, Indicates A collection of all advertisements included in a time period; Indicates time period budget ceiling within the Indicated in The total exposure of all advertisements on various media platform resources in a time period; Expressing to all , the above inequalities must all hold.

Using the advertising resource super-optimization algorithm described above, dynamically adjust the allocation of advertising resources in different media channels according to user portraits and traffic characteristics of media platforms to achieve maximum advertising benefits;

Considering the traffic characteristics and value of resources on different media platforms, accurately allocating advertising resources to the media platforms with the greatest potential value can avoid resource waste, improve the efficiency of advertising, and make advertising more accurate and effective. By optimizing the allocation of advertising positions and placing advertising resources on media platforms with higher traffic and greater value, it can increase advertising exposure and conversion rates, thereby increasing advertising revenue. This is beneficial to both advertisers and advertising platforms, and can increase the revenue levels of both parties;

By calculating the expected revenue of each ad space and making decisions based on budget constraints, when the preset threshold is reached or a specific rule is triggered, the system can respond immediately and adjust the delivery strategy, such as reducing budget allocation for ads with poor performance, increasing bids for high-quality ad spaces, or delivering more advertising content that meets the characteristics of the target user group.

The Internet-based intelligent advertising marketing system also includes a directional delivery unit 3, which matches advertisements with target audiences based on user portraits, and uses a directional delivery algorithm to formulate an optimal advertising delivery strategy in combination with advertising goals and budgets;

The directional delivery unit 3 includes a matching delivery module. The matching delivery module is based on a directional delivery algorithm, accurately matches the user portrait with the advertising positioning strategy, and pushes the most suitable advertising content to the corresponding user. The specific expression of the directional delivery algorithm is:

;

In the formula, Represents user portrait; Indicates the characteristic description of the advertisement; Represents user portrait and ad characterization The matching score between them is used to measure the suitability of an advertisement for a user. Represents the personalized attributes of the user portrait, such as age, gender, and occupation; Indicates the target population attributes in the advertising targeting strategy; Represents a collection of user's interest tags; Represents a set of interest keywords related to advertisements; Represents the user's historical consumption behavior data; Indicates the product category the ad is associated with; Represents the similarity function between two attributes; , and Respectively represent the weight factors of different matching dimensions;

The most suitable advertising content is determined based on the user's characteristics and the advertising positioning strategy; the most suitable advertising content is determined by comprehensively considering multiple factors such as the user's personalized attributes, interest tags, historical consumption behavior, etc., by calculating the similarity between them and multiplying them by the corresponding weight factor. The weight factor can be adjusted according to the specific situation so that the matching priority can be adjusted according to different situations.

In this embodiment, considering the time sensitivity of user behavior, the user's purchase needs and interest preferences usually have a certain timeliness. The recent purchase behavior may reflect the user's current interest hot spots or demand trends. Pushing related advertising content is more likely to attract the user's attention and response. At the same time, placing advertisements based on the user's active time period can increase the possibility of seeing the advertisement and interacting with it. The recent active time period factor is introduced to optimize the targeted delivery algorithm. The optimized targeted delivery algorithm is:

;

in, The time sensitivity of user behavior, indicating the most recent active period; Indicates the time information of advertisement delivery; Indicates the weight of the impact of the time sensitivity of user behavior on the matching results, and indicates the importance of time-related factors such as recent purchase behavior and active time periods in targeted delivery.

when When the value of is large, it means that time sensitivity has a greater impact on the matching results, and the model pays more attention to the user's recent behavior or active period; When the value of is small, the time factor has less impact on the matching results, and the model relies more on the matching results of other dimensions. Accurate push based on the user's latest behavior can effectively improve the click-through rate and conversion rate of ads. By analyzing the user's active time periods, advertisers can adjust their advertising delivery strategies based on user traffic and activity in different time periods to optimize the use of advertising budgets.

The Internet-based intelligent advertising marketing system also includes a delivery optimization unit 4, which adjusts the bidding strategy of the advertisement on each media platform resource based on the delivery optimization algorithm according to the net income of the advertisement position on each media platform resource, and adopts a cross-platform joint optimization algorithm to track the user's behavior path on different platforms, thereby optimizing the advertising delivery strategy and achieving the overall optimal cross-platform advertising effect;

Among them, the delivery optimization unit 4 includes a data monitoring module, a target setting module and a bidding strategy optimization module;

Among them, the data monitoring module is used to collect and process data related to advertising, including advertising clicks and conversion data. These data can come from the real-time data stream or historical data of the advertising platform, and are used to evaluate the performance of advertising on different media platform resources and calculate the net profit. The goal setting module determines the optimization goals based on the cross-platform joint optimization algorithm. The optimization goals include maximizing total revenue, maximizing return on investment (ROI), and minimizing costs. The bidding strategy optimization module sets the expected revenue and goals, and adjusts the bidding strategy of advertising on each media platform resource based on the optimization goals through the delivery optimization algorithm. The delivery optimization algorithm dynamically adjusts the bidding strategy based on the expected revenue and real-time data to achieve the best delivery effect.

In this embodiment, the specific expression of the delivery optimization algorithm is:

;

In the formula, Represents the global bid coefficient, which is used to uniformly adjust the cost ratio of all resources; Indicated in The amount of ads delivered on a media resource, that is, the number of times an ad is displayed or clicked on that resource; Indicated in Average revenue on each media source; Indicates the number of media resources; is an index variable, which is used to traverse each media resource from 1 to M; Indicated in Average cost per media resource;

Among them, the constraint condition is that the total cost does not exceed the budget :

;

In the formula, represents the maximum total cost allowed;

By adjusting the global bid modifier and the amount served on each property , to maximize the total net benefit;

Based on the optimization goal of maximizing total revenue, then:

;

In the formula, It means that maximizing the total benefit is taken as the optimization goal;

Introduce constraints into the objective function to form an objective function to dynamically adjust the global bid coefficient and the amount served on each property :

;

In the formula, represents the Lagrangian function;

By corresponding and Find the partial derivatives and set them to 0 to find the extreme points:

;

;

Solve the above system of equations, for Amount of delivery on media properties :

;

For global bid modifiers :

;

In the formula, represents the Lagrange multiplier, which is used to deal with the constraints of the problem; Denotes the Lagrangian function L for the amount of delivery The partial derivative of Represents the Lagrangian function L for the global bid coefficient The partial derivative of .

By dynamically adjusting the amount of advertising and bidding strategies on different media resources to maximize total revenue and ensure that the budget is not exceeded, the effectiveness and efficiency of advertising are improved, allowing advertisers to obtain the greatest return within a limited budget.

The specific expression involved in the cross-platform joint optimization algorithm is:

Media Resources Exposure to media resources The impact of clicks is:

;

Media Resources Exposure to media resources The impact of clicks is:

;

In the formula, Indicates that the ad is in the media resource Exposure to media resources The impact of clicks; Indicated in media resources Click-through rate; Indicated in media resources Click-through rate; Indicated in media resources After exposure in media resources Increased probability of clicks; Indicated in media resources After exposure in media resources Increased probability of clicks; Indicated in media resources Number of exposures; Indicated in media resources Number of exposures; The intensity coefficient indicating the influence; The intensity coefficient indicating the influence;

Then the joint optimization objective function K is:

;

In the formula, Represents media resources In the amount of delivery and bid modifier Independent advertising effect income under Indicates the delivery volume of each media resource in its respective and bid modifier The sum of direct advertising effect benefits generated under and All are index variables; Represents media resources Media Resources The weight coefficient between ; Represents media resources Media Resources The weight coefficient between ; It is the overall optimization goal, which is used to maximize the function value to achieve the best advertising effect across media resource platforms.

Among them, the delivery optimization algorithm adjusts the bidding strategy of advertisements on each media platform resource based on the joint optimization objective function K (i.e., the global bidding coefficient ) and the volume of delivery on each platform ; By considering the synergy between platforms, the exposure between platforms can increase the probability of users clicking on other platforms. This synergy helps to improve the overall advertising effect. It comprehensively considers the benefits of independent delivery on each platform and the gains from mutual influence between platforms to achieve the overall optimal cross-platform advertising effect.

Embodiment 2:

The difference between Example 2 of the present invention and Example 1 is that this example introduces an Internet-based intelligent advertising and marketing method used by an Internet-based intelligent advertising and marketing system.

The Internet-based intelligent advertising and marketing method is used in the above-mentioned Internet-based intelligent advertising and marketing system, and comprises the following steps:

S10.1, the data collection module 11 collects user behavior data, and the data analysis module 12 uses the user behavior analysis algorithm to calculate the user's consumption frequency, content preference and content interaction. The user portrait construction module 13 generates user tags including consumption frequency, content preference and product relevance according to the user's behavior characteristics, and classifies the users into groups with similar characteristics to form user portraits;

S10.2, the media resource integration module 22 is used to connect multiple media platform resources, uniformly manage and dispatch cross-platform advertising space resources, and the advertising resource management module 23 uses the advertising resource super-optimization algorithm to maximize the expected revenue and dynamically adjust the advertising allocation strategy in different media channels under the consideration of budget constraints and time period quota factors;

S10.3, and through the matching delivery module, use the targeted delivery algorithm to calculate the matching degree between the user portrait and the advertising features, combine personalized attributes, interest tags, and historical consumption behaviors for accurate matching, and formulate the best advertising delivery strategy to push the most suitable advertising content to the corresponding users, while considering time-sensitive factors such as the most recent active period and purchase behavior to improve the relevance and click-through conversion rate of the advertisement;

S10.4. Finally, the delivery optimization unit 4 monitors the advertising delivery effect in real time, collects and processes the data related to advertising clicks and conversions, sets the maximization of total revenue as the optimization goal through the goal setting module, and uses the delivery optimization algorithm to dynamically adjust the global bid coefficient and the delivery volume on each media resource to ensure that the advertising benefits are maximized while meeting the budget constraints.

The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims (10)

1. An intelligent advertising and marketing system based on the Internet, characterized by comprising: A user analysis management unit (1), wherein the user analysis management unit (1) collects user behavior data through a data collection module (11), and the data analysis module (12) analyzes the user behavior data to identify the user's consumption frequency, content preference, and product relevance characteristics; a user portrait construction module (13) generates user tags based on the results output by the data analysis module (12), and classifies user groups according to the user tags to form different user group categories; An advertising resource management unit (2), the advertising resource management unit (2) being used to manage and allocate advertising space resources for delivery, and the advertising resource management unit (2) comprising an advertising resource library (21), a media resource integration module (22) and an advertising resource management module (23); The advertising resource management module (23) uses an advertising resource super-optimization algorithm to allocate advertising slots to advertisements in the advertising resource library (21) according to the traffic characteristics and values of different media channels based on the user portrait generated by the user analysis management unit (1); A targeted delivery unit (3), wherein the targeted delivery unit (3) matches advertisements with target audiences based on user portraits, combines advertisement targets and budgets, adopts a targeted delivery algorithm to formulate an optimal advertisement delivery strategy, and introduces a recent active time period factor to optimize the targeted delivery algorithm. The optimized targeted delivery algorithm is: ; In the formula, Represents user portrait; Indicates the characteristic description of the advertisement; Represents the optimized user portrait and ad characterization The matching score between Represents the personalized attributes in the user portrait; Indicates the target population attributes in the advertising targeting strategy; Represents a collection of user's interest tags; Represents a set of interest keywords related to advertisements; Represents the user's historical consumption behavior data; Indicates the product category the ad is associated with; Represents the similarity function between two attributes; , and Respectively represent the weight factors of different matching dimensions; The time sensitivity of user behavior; Indicates the time information of advertisement delivery; Indicates the influence weight of the time sensitivity of user behavior on the matching results; The delivery optimization unit (4) adjusts the bidding strategy of advertisements on each media platform resource based on the delivery optimization algorithm according to the net revenue of the advertisement space on each media platform resource, and adopts a cross-platform joint optimization algorithm to track the behavior path of users on different platforms, thereby optimizing the advertisement delivery strategy. 2. The Internet-based intelligent advertising and marketing system according to claim 1 is characterized in that the data analysis module (12) analyzes the user behavior data based on a user behavior analysis algorithm, and the specific expression involved is: User consumption frequency: If the user set is U, user , where the time window , then the specific expression of user consumption frequency is: ; In the formula, Indicates user Frequency of purchases in the past 7 days; Indicates user At the point in time The number of purchases made; represents a time window; Indicates time point Belongs to the time window ; User content preferences: If there is a set of content to be interacted , where the content of user interaction , then the specific expression of user content preference is: ; In the formula, Represents the content of user interaction; is the indicator function; Indicates user About content degree of preference; Product Relevance: If there are two products E and F, their support and confidence , then the correlation between products E and F is calculated as: ; In the formula, It indicates the increase multiple of the probability of purchasing product F compared to the base probability of purchasing product F when it is known that product E has been purchased. 3. The intelligent advertising and marketing system based on the Internet according to claim 1 is characterized in that the specific steps involved in generating user tags by the user portrait building module (13) are: S3.1. Analyze the user behavior data output from the data analysis module (12) to obtain consumption frequency A, content preference B, and product relevance feature C; S3.2. Define a set of user tags based on business needs and analysis results, and map consumption frequency A, content preference B, and product relevance characteristics C to the defined user tags; S3.3, based on the defined user tags, user tag classification is used to classify user groups and divide users into groups with similar characteristics; S3.4. Integrate user tags and group information to generate user portraits. 4. The Internet-based intelligent advertising and marketing system according to claim 1, characterized in that the advertising resource library (21) is used to store all advertising resources for delivery; The media resource integration module (22) is connected to various media platforms through an API interface to manage media platform resources used for advertising, thereby achieving unified management and scheduling of advertising resources across platforms; The advertising resource management module (23) uses an advertising resource super-optimization algorithm to allocate advertising slots to advertisements in the advertising resource library (21) according to the traffic characteristics and values of different media channels based on the user portrait generated by the user analysis management unit (1). 5. The Internet-based intelligent advertising and marketing system according to claim 4, characterized in that the advertising resource super-optimization algorithm is specifically: ; In the formula, Indicates Advertisement slot in Exposure on media platforms; Indicates Advertisement slot in Click-through rate on media platform resources; Indicates Advertisement slot in Conversion rate on media platform resources; Indicates Advertisement slot in Average revenue per user generated by media platform resources; Indicates Advertisement slot in Cost per thousand impressions on each media platform property; Indicates the number of ad slots; Indicates the number of media platform resources; It represents the net revenue obtained by deducting the sum of the corresponding delivery costs from the sum of the expected revenue generated by all advertising positions on various media platform resources; The constraints are: The total budget constraint is: ; In the formula, represents the total budget; Non-negativity constraints: ; In the formula, Expressing to all and all , the above inequalities must all hold; Time period delivery quota constraints: ; In the formula, Indicates A collection of all advertisements included in a time period; Indicates time period budget ceiling within the Indicated in The total exposure of all advertisements on various media platform resources in a time period; Expressing to all , all the above inequalities must hold. 6. The Internet-based intelligent advertising marketing system according to claim 1 is characterized in that the directional delivery unit (3) includes a matching delivery module, which is based on a directional delivery algorithm and accurately matches user portraits with advertising positioning strategies to push the most suitable advertising content to the corresponding users. The specific expression of the directional delivery algorithm is: . 7. The Internet-based intelligent advertising and marketing system according to claim 1, characterized in that the delivery optimization unit (4) comprises a data monitoring module, a target setting module and a bidding strategy optimization module; Among them, the data monitoring module is used to collect and process data related to advertising delivery, the target setting module determines the optimization target based on the cross-platform joint optimization algorithm, and the bidding strategy optimization module adjusts the bidding strategy of advertisements on each media platform resources based on the expected revenue and target setting through the delivery optimization algorithm based on the optimization target. 8. The Internet-based intelligent advertising and marketing system according to claim 7, characterized in that the specific expression of the delivery optimization algorithm is: ; In the formula, Indicates the global bid coefficient; Indicated in The amount of delivery on each media resource; Indicated in Average revenue on each media source; Indicates the number of media resources; is the index variable; Indicated in Average cost per media source; Among them, the constraint condition is that the total cost does not exceed the budget : ; In the formula, represents the maximum total cost allowed; By adjusting the global bid modifier and the amount served on each property , to maximize the total net benefit; Based on the optimization goal of maximizing total revenue, then: ; In the formula, It means that maximizing the total benefit is taken as the optimization goal; Introduce constraints into the objective function to form an objective function to dynamically adjust the global bid coefficient and the amount served on each property : ; In the formula, represents the Lagrangian function; By corresponding and Find the partial derivatives and set them to 0 to find the extreme points: ; ; Solve the above system of equations, for Amount of delivery on media properties : ; For global bid modifiers : ; In the formula, represents the Lagrange multiplier; Denotes the Lagrangian function L for the amount of delivery The partial derivative of Represents the Lagrangian function L for the global bid coefficient The partial derivative of . 9. The Internet-based intelligent advertising and marketing system according to claim 8, characterized in that the specific expression involved in the cross-platform joint optimization algorithm is: Media Resources Exposure to media resources The impact of clicks is: ; Media Resources Exposure to media resources The impact of clicks is: ; In the formula, Indicates that the ad is in the media resource Exposure to media resources The impact of clicks; Indicated in media resources Click-through rate; Indicated in media resources Click-through rate; Indicated in media resources After exposure in media resources Increased probability of clicks; Indicated in media resources After exposure in media resources Increased probability of clicks; Indicated in media resources Number of exposures; Indicated in media resources Number of exposures; The intensity coefficient indicating the influence; The intensity coefficient indicating the influence; Then the joint optimization objective function K is: ; In the formula, Represents media resources In the amount of delivery and bid modifier Independent advertising effect income under Indicates the delivery volume of each media resource in its respective and bid modifier The sum of direct advertising effect benefits generated under and All are index variables; Represents media resources Media Resources The weight coefficient between ; Represents media resources Media Resources The weight coefficient between ; It is the overall optimization goal, which is used to maximize the function value to achieve the best advertising effect across media resource platforms. 10. An Internet-based intelligent advertising and marketing method, used in an Internet-based intelligent advertising and marketing system as claimed in any one of claims 1 to 9, characterized in that it comprises the following steps: S10.1, the data collection module (11) collects user behavior data, and the data analysis module (12) uses the user behavior analysis algorithm to calculate the user's consumption frequency, content preference and content interaction, and the user portrait construction module (13) generates user tags including consumption frequency, content preference and product relevance based on the user's behavior characteristics, and classifies the users into groups with similar characteristics to form user portraits; S10.2. The media resource integration module (22) is used to connect multiple media platform resources, uniformly manage and schedule cross-platform advertising space resources, and the advertising resource management module (23) uses an advertising resource super-optimization algorithm to maximize expected revenue and dynamically adjust the advertising distribution strategy in different media channels under the consideration of budget constraints and time period quota factors; S10.3, and through the matching delivery module, use the targeted delivery algorithm to calculate the matching degree between the user portrait and the advertising features, combine personalized attributes, interest tags, and historical consumption behaviors for accurate matching, and formulate the optimal advertising delivery strategy to push the most suitable advertising content to the corresponding users; S10.4. Finally, the delivery optimization unit (4) monitors the advertising delivery effect in real time, collects and processes the data related to advertising clicks and conversions, sets the maximization of total revenue as the optimization goal through the goal setting module, and uses the delivery optimization algorithm to dynamically adjust the global bid coefficient and the delivery volume on each media resource to ensure that the advertising benefit is maximized while meeting the budget limit.