Intelligent online advertisement pushing system based on big data

Abstract

The invention discloses an online advertisement intelligent pushing system based on big data, and relates to the technical field of advertisement pushing, the online advertisement intelligent pushing system comprises a data acquisition module, a data processing module, a data storage module, an intelligent pushing module and a user feedback module, the intelligent pushing module comprises a user interest analysis unit, a pushing opportunity unit and an effect evaluation unit, the user interest analysis unit analyzes the interest degrees of different advertisement categories, the pushing opportunity unit divides a day into a plurality of time periods, and determines the advertisement pushing strength and the advertisement pushing time periods, so that the preference of the user for the advertisement categories is analyzed through an interest evaluation algorithm, the most popular advertisement category is selected for pushing, and the advertisement pushing efficiency is improved. The optimal advertisement pushing time period and pushing strength are determined through the segmentation evaluation algorithm, the pushing effect is evaluated through the advertisement effect evaluation algorithm, the threshold value is set, the competitive advertisement influence is adjusted in real time, the advertisement strategy is dynamically adjusted in the high competitive time period, and the advertisement pushing accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of advertising push, and specifically to an intelligent online advertising push system based on big data. Background Art

[0002] Advertising has always been a very important part of marketing and is one of the most widely covered marketing means in the prior art. Each merchant uses advertising to attract customers, sell products or provide services, and achieve profitability. With the rapid development of Internet technology and big data analysis, online advertising push has become one of the important means for enterprises to promote products and services. In the early Internet advertising era, the push method of advertising was relatively simple, usually based on page content or simple user behavior for display. Although this method improves the relevance of advertising to a certain extent, there is still much room for improvement in its accuracy and user experience.

[0003] Currently, the advertising push system usually only relies on simple binary classification to judge whether an advertisement is relevant to a user, lacks detailed analysis of the advertisement, and is difficult to accurately capture the user's interest points, resulting in one-sided interest evaluation and inaccurate advertisement push. Moreover, the advertising push timing is often fixed, unable to fully consider the complexity and diversity of user behavior and the advertising market, resulting in poor effectiveness of advertising push. At the same time, it is unable to adjust strategies in a timely manner according to the advertising effect, reducing the advertising push effect. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent online advertising push system based on big data, which solves the problems raised in the above background art.

[0005] To achieve the above purpose, the present invention provides the following technical solution: An intelligent online advertising push system based on big data, including a data collection module, a data processing module, a data storage module, an intelligent push module, and a user feedback module;

[0006] The data collection module is used to collect user data through various channels, and the user data includes user behavior data, historical advertising data, social media interaction data, and user identity information;

[0007] The intelligent push module includes a user interest analysis unit, a push timing unit, and an effect evaluation unit;

[0008] The user interest analysis unit uses an interest evaluation algorithm to analyze the degree of interest in different advertisement categories based on user data within a past unit of time, and determines the preference degree of the user for each type of advertisement. The push timing unit uses a segmented evaluation algorithm to divide a day into multiple time periods, calculates the activity level of the user in each time period respectively, and determines the advertisement push intensity and the advertisement push time period according to the high or low activity level of the user in each time period. The effect evaluation unit evaluates the push effect of each advertisement push time period by using an advertisement effect evaluation algorithm, sets a threshold for the result of the advertisement effect evaluation algorithm, and thereby determines whether the advertisement push effect is qualified. When it is unqualified, the data in the segmented evaluation algorithm is adjusted.

[0009] Optionally, the analysis process of the interest evaluation algorithm is as follows:

[0010]

[0011] Where S u,b represents the interest score of user u for advertisement category b;

[0012] CA u,b represents the number of times user u clicks on advertisement category b;

[0013] CA ave,b represents the average value of the number of clicks on advertisement category b;

[0014] W1 represents the click count influence coefficient, and its value range is from 0 to 1;

[0015] T u,b represents the duration that user u views advertisement category b;

[0016] T u,all represents the total duration that user u views advertisements;

[0017] W2 represents the advertisement viewing duration influence coefficient, and its value range is from 0 to 1;

[0018] D u,b represents the quantity of goods of advertisement category b purchased by user u;

[0019] D ave,b represents the average value of the quantity of goods of advertisement category b purchased;

[0020] W3 represents the goods purchase influence coefficient, and its value range is from 0 to 1;

[0021] The interest score S of user u for advertisement category b u,b represents the preference degree of the user for the advertisement. The larger the interest score S of user u for advertisement category b u,b is, the higher the acceptance degree of the user for this type of advertisement. On the contrary, it is lower. When pushing advertisements, the advertisement category with the highest score is pushed.

[0022] Optionally, the process of the segmented evaluation algorithm is as follows:

[0023]

[0024] where S j,u represents the activity score of user u in time period j;

[0025] RT j represents the platform access duration in time period j;

[0026] T j，all represents the total duration of time period j, that is, the length of a time period;

[0027] W 4,j represents the influence coefficient of access duration in time period j, and its value range is from 0 to 1;

[0028] L j represents the occupied advertisement quantity in time period j, that is, the number of advertisement positions pushed by competitors in time period j;

[0029] L all represents the total quantity of advertisement positions provided by the platform;

[0030] W 5,j represents the influence coefficient of competitive advertisements in time period j, and its value range is from 0 to 1;

[0031] The higher the activity score S of user u in time period j j,u , the less the user activity and the proportion of competitive advertisements in the current time period. The first y advertisement push time periods are selected according to the activity score S of user u in time period j j,u , and the advertisement budget investment intensity is determined according to the ranking of the advertisement push time periods. Pushing advertisements in this time period can increase the click-through rate and conversion rate of the advertisements, and introduce the factor of the proportion of competitive advertisements to simulate the competition environment in the advertisement market and improve the accuracy of the segmented evaluation algorithm.

[0032] Optionally, the process of the advertisement effect evaluation algorithm is as follows:

[0033]

[0034] where E y represents the effect score of the yth advertisement push time period;

[0035] C y represents the click volume of the yth advertisement push time period;

[0036] B y represents the exposure volume of the yth advertisement push time period, that is, the number of times the advertisement is displayed;

[0037] α represents the advertising click impact coefficient, and its value range is from 0 to 1;

[0038] R y represents the purchase volume in the y-th advertising push time period, that is, the number of purchases completed by clicking on the advertisement;

[0039] β represents the advertising purchase impact coefficient, and its value range is from 0 to 1;

[0040] P ave represents the average order price;

[0041] S y represents the budget in the y-th advertising push time period;

[0042] γ represents the advertising return impact coefficient, and its value range is from 0 to 1;

[0043] S u,b represents the interest score of user u for advertisement category b;

[0044] The interest score S of user u for advertisement category b u,b is substituted into the advertisement effect evaluation algorithm to indicate the highest interest score of user u for advertisement category b;

[0045] The effect score E of the y-th advertising push time period y The higher it is, the better the overall attractiveness of the advertisement and the higher the conversion rate. Conversely, it is worse. The value range of y is from 1 to 3. When y is equal to 1, E1 represents the effect score of the first advertising push time period. The advertising budget investment intensity in this time period is 0.5, and the highest score time period of the activity score S of user u in time period j corresponding to the first advertising push time period j,u ;

[0046] When y is equal to 2, E2 represents the effect score of the second advertising push time period. The advertising budget investment intensity is 0.3, and the second highest score time period of the activity score S of user u in time period j corresponding to the second advertising push time period j,u ;

[0047] When y is equal to 3, E3 represents the effect score of the third advertising push time period. The advertising budget investment intensity is 0.2, and the third highest score time period of the activity score S of user u in time period j corresponding to the third advertising push time period j,u ;

[0048] Optionally, the set effect threshold of the effect score E of the y-th advertising push time period is Z y , when the effect score E of the y-th advertising push time period y < the effect threshold Z y y ​When it indicates that the advertising effect is too low, the influence coefficient W5 of competing advertisements within the corresponding time period j is increased at this time, and the adjustment process is as follows:

[0049]

[0050] Among them, NW 5,j represents the influence coefficient of competing advertisements within the new time period j;

[0051] Z y represents the effect threshold;

[0052] In the effect score E of the y-th advertisement push time period y <effect threshold Z y When, the influence coefficient NW of competing advertisements within the new time period j is used 5,j to replace the influence coefficient W5 of competing advertisements within the corresponding time period j, which can adjust the influence of the advertisements pushed by competitors on its own advertising effect according to the actual situation, reduce the user activity score in the high-competition time period, avoid advertising push in the high-competition time period, improve the effectiveness of advertising push, and the advertisement push time period will be rearranged during subsequent advertisement push.

[0053] Optionally, the user feedback module is used to construct a feedback opinion platform and an intelligent customer unit. Users log in to the feedback opinion platform and the intelligent customer unit to send suggestions and inquire about advertising-related information. The problems collected by the intelligent customer unit and the suggestions collected by the user feedback module are both transmitted to the data storage module for storage.

[0054] Optionally, the data processing module is used to perform deduplication, denoising, and outlier detection processing on the user data collected by the data acquisition module.

[0055] Optionally, it further includes an encryption protection module. The encryption transmission module is used to encrypt the user data to ensure that it is not leaked during storage and transmission.

[0056] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0057] 1. The present invention evaluates the degree of user preference for each advertising category over a past period of time through an interest evaluation algorithm, and selects the advertising category with the highest degree of preference for pushing. In the interest evaluation algorithm, comprehensive analysis is carried out by combining factors such as the user's clicks, browsing duration, and purchase numbers for different advertising categories, and the results are quantified to achieve a detailed analysis of the advertisements, which can effectively capture the user's interests, improve the accuracy of advertisement pushing. After determining the advertising category to be pushed, the day is divided into multiple time periods through a segmented evaluation algorithm in the pushing timing unit, and the activity levels of the user in each time period are calculated respectively. The advertisement pushing intensity and the advertisement pushing time period are determined according to the high or low activity levels of the user in each time period, improving the accuracy of advertisement pushing. And the competitive advertisement pushing factors for each time period are introduced into the segmented evaluation algorithm, fully considering the influence of the complexity and diversity of user behavior and the advertising market on the advertisement pushing effect, and improving the effectiveness of advertisement pushing.

[0058] 2. After determining the advertisement pushing time period and the advertising category, the present invention evaluates the effect of advertisement pushing through an advertisement effect evaluation algorithm. The advertisement effect evaluation algorithm evaluates the pushing effect for each advertisement pushing time period based on the behavioral data of the user for the pushed advertisement, and sets a threshold for the result of the advertisement effect evaluation algorithm to judge whether the advertisement pushing effect is qualified. When it is unqualified, the competitive advertisement factors in the segmented evaluation algorithm are dynamically adjusted, which can adjust the influence of the advertisements pushed by competitors on its own advertisement effect according to the actual situation, reduce the user activity level score in the high competition time period, avoid pushing advertisements in the high competition time period, and can timely adjust the advertisement pushing strategy to improve the quality of advertisement pushing. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a block diagram of the system module of the present invention;

[0060] Figure 2 It is a flow chart of the pushing timing unit of the present invention;

[0061] Figure 3 It is a block diagram of the user feedback module of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] 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. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0063] Embodiment, please refer to Figures 1 to 3, this implementation provides an intelligent online advertising push system based on big data, including a data collection module, a data processing module, a data storage module, an intelligent push module, and a user feedback module;

[0064] The data collection module is used to collect user data through various channels. The user data includes user behavior data, historical advertising data, social media interaction data, and user identity information;

[0065] The data processing module is used to perform deduplication, denoising, and outlier detection on the user data collected by the data collection module to improve the quality of user data and ensure the accuracy of subsequent data analysis;

[0066] The user feedback module is used to build a feedback opinion platform and an intelligent customer unit. Users log in to the feedback opinion platform and the intelligent customer unit to send suggestions and inquire about advertising-related information. The problems collected by the intelligent customer unit and the suggestions collected by the user feedback module are both transmitted to the data storage module for storage for subsequent analysis;

[0067] The intelligent push module includes a user interest analysis unit, a push timing unit, and an effect evaluation unit;

[0068] The user interest analysis unit uses an interest evaluation algorithm to analyze the degree of interest in different advertising categories based on user data within a unit of time, and determines the user's preference for advertisements. The push timing unit divides a day into multiple time periods using a segmented evaluation algorithm, calculates the activity level of users in each time period respectively, and determines the advertising push intensity and advertising push time period according to the high or low activity level of users in each time period. The effect evaluation unit uses an advertising effect evaluation algorithm and combines the data of the interest evaluation algorithm and the segmented evaluation algorithm to evaluate the push effect of each advertising push time period, and sets a threshold for the result of the advertising effect evaluation algorithm to determine whether the advertising push effect is qualified. When it is unqualified, the data in the segmented evaluation algorithm is adjusted.

[0069] It also includes an encryption protection module. The encryption transmission module is used to encrypt user data to ensure that it is not leaked during storage and transmission, and will remind users when collecting user data, and data collection can only be carried out after the user agrees, effectively protecting the privacy of users.

[0070] More specifically, in this embodiment: After the data collection module collects user data, due to the complexity and diversity of Internet users, the data processing module is required to preprocess the user data to ensure the quality of user data and improve the accuracy of subsequent analysis. Then, the user interest analysis unit in the intelligent push module analyzes the users, evaluates the degree of preference of the users for each advertising category over a past period of time, and selects the advertising category with the highest degree of preference for pushing. In the interest evaluation algorithm of the user interest analysis unit, through comprehensive analysis by combining factors such as the clicks, browsing duration, and purchase numbers of the users for different advertising categories, and quantifying the results, a detailed analysis of the advertisements can be achieved, effectively capturing the interests of the users and improving the accuracy of advertisement pushing. After determining the advertising category to be pushed, the segmented evaluation algorithm in the push timing unit divides a day into multiple time periods, calculates the activity levels of the users in each time period respectively, determines the advertisement pushing intensity and the advertisement pushing time period according to the high or low activity levels of the users in each time period, and introduces the competing advertisement pushing factors in each time period into the segmented evaluation algorithm, fully considering the impact of the complexity and diversity of user behavior and the advertising market on the advertisement pushing effect, and improving the effectiveness of advertisement pushing.

[0071] After determining the advertisement pushing time period and the advertising category, the advertisement pushing work can be carried out. Then, the advertisement effect evaluation algorithm in the effect evaluation unit evaluates the effect of the advertisement pushing. The advertisement effect evaluation algorithm evaluates the pushing effect of each advertisement pushing time period according to the behavioral data of the users for the pushed advertisements, and sets a threshold for the result of the advertisement effect evaluation algorithm to judge whether the advertisement pushing effect is qualified. When it is unqualified, the data in the segmented evaluation algorithm is dynamically adjusted to change the advertisement pushing intensity and the advertisement pushing time period, so that the system can be dynamically adjusted according to the actual advertising market situation, realizing the effect of intelligent advertisement pushing and effectively improving the quality of advertisement pushing.

[0072] Furthermore, the analysis process of the interest evaluation algorithm is as follows:

[0073]

[0074] Where S u,b represents the interest score of user u for advertising category b;

[0075] CA u,b represents the number of times user u clicks on advertising category b;

[0076] CA ave,b represents the average value of the number of clicks on advertising category b;

[0077] W1 represents the click count influence coefficient, and its value range is from 0 to 1;

[0078] T u,bIndicates the duration for user u to view ad category b;

[0079] T u,all Indicates the total duration for user u to view ads;

[0080] W2 represents the influence coefficient of ad viewing duration, with a value range from 0 to 1;

[0081] D u,b Indicates the quantity of products of ad category b purchased by user u;

[0082] D ave,b Indicates the average value of products of ad category b purchased;

[0083] W3 represents the influence coefficient of product purchase, with a value range from 0 to 1;

[0084] Ad categories include different types of products, such as electronic products, clothing, household items, etc.;

[0085] Specifically, the interest score S of user u for ad category b u,b Indicates the preference degree of the user for the ad. The higher the interest score S of user u for ad category b u,b the higher the acceptance degree of the user for this type of ad, and vice versa. By introducing multi-dimensional data to calculate the interest scores for different ad categories b and quantifying the results, the preference degrees of different users for ads can be analyzed more comprehensively and accurately, effectively capturing the interests of users, improving the accuracy of ad push, and pushing ads according to the ad category with the highest score during ad push.

[0086] Furthermore, the process of the segmented evaluation algorithm is as follows:

[0087]

[0088] where S j,u Indicates the activity score of user u in time period j;

[0089] RT j Indicates the platform access duration in time period j;

[0090] T j，all Indicates the total duration of time period j, that is, the length of a time period;

[0091] W 4,j Indicates the influence coefficient of access duration in time period j, with a value range from 0 to 1;

[0092] L j Indicates the occupied ad quantity in time period j, that is, the number of ad positions pushed by competitors within time period j;

[0093] L allRepresents the total number of ad slots provided by the platform;

[0094] W 5,j Represents the competitive ad influence coefficient in time period j, with a value range from 0 to 1;

[0095] The value range of j is from 1

[0096] Specifically, the activity score S of user u in time period j j,u The higher it is, the less the user activity and the proportion of competitive ads in the current time period. According to the activity score S of user u in time period j j,u Take the top y ad push time periods according to the level of S, and determine the advertising budget investment intensity according to the ranking of the ad push time periods. The total value of the advertising budget investment intensity is 1. By dividing a day into different time periods, accurately calculate the time periods when users are active, so as to improve the effectiveness of ad push, and introduce the factor of the proportion of competitive ads to simulate the competitive environment of the ad market, so that ads can be pushed in appropriate time periods, improve the accuracy of ad push, and ads pushed in this time period can be more easily noticed by users and reduce the influence of competitors' ads, so as to increase the click-through rate and conversion rate of ads.

[0097] Furthermore, the ad effect evaluation algorithm process is as follows:

[0098]

[0099] Among them, E y Represents the effect score of the y-th ad push time period;

[0100] C y Represents the click-through volume of the y-th ad push time period;

[0101] B y Represents the exposure volume of the y-th ad push time period, that is, the number of times the ad is displayed;

[0102] α represents the ad click influence coefficient, with a value range from 0 to 1;

[0103] R y Represents the purchase volume of the y-th ad push time period, that is, the number of purchases completed by clicking on the ad;

[0104] β represents the ad purchase influence coefficient, with a value range from 0 to 1;

[0105] P ave Represents the average order price;

[0106] S y Represents the budget of the y-th ad push time period;

[0107] γ represents the ad return influence coefficient, with a value range from 0 to 1;

[0108] S u,b represents the interest score of user u for ad category b;

[0109] The interest score S of user u for ad category b u,b is substituted into the ad effect evaluation algorithm, indicating that user u has the highest interest score for ad category b, because the interest evaluation algorithm will push ads according to the ad category with the highest score;

[0110] By multiplying the ad return rate with the interest score S representing user u for ad category b u,b it enables the evaluation of ad effect not only to depend on the performance data of the ad, but also to consider the impact of user interest preferences on the ad return rate, and can quantify the actual effect of each ad in a specific user group. Advertisers can increase investment in the user group of ads with high interest scores and high returns. By combining user interests and the actual returns of ads, advertisers can allocate more budgets to those ad categories that are most likely to generate high returns for pushing, while improving the overall ad return rate.

[0111] Specifically, the effect score E of the y-th ad push time period y The higher it is, the better the overall attractiveness of the ad and the higher the conversion rate, and vice versa. The value range of y is from 1 to 3. When y equals 1, E1 represents the effect score of the first ad push time period. The ad budget investment intensity for this time period is 0.5, and the highest score of the activity score S of user u in time period j for the first ad push time period j,u is the highest score;

[0112] When y equals 2, E2 represents the effect score of the second ad push time period. The ad budget investment intensity is 0.3, and the second highest score of the activity score S of user u in time period j for the second ad push time period j,u is the second highest score;

[0113] When y equals 3, E3 represents the effect score of the third ad push time period. The ad budget investment intensity is 0.2, and the third highest score of the activity score S of user u in time period j for the third ad push time period j,u is the third highest score;

[0114] The total value of the ad budget investment intensity is 1. The ad budget investment intensities in different ad push time periods are different. The ad budget investment intensity can be expressed as the budget cost. For example, the ad budget investment intensity of the first ad push time period is 0.5, that is, 50%, and 50% of the daily ad budget is spent in this time period.

[0115] Furthermore, set the effect score E of the y-th ad push time periody The effect threshold is Z y When the effect score E of the y-th advertisement push time period y < the effect threshold Z y it indicates that the advertisement effect is too low. At this time, the influence coefficient W5 of competing advertisements in the corresponding time period j is increased. The adjustment process is as follows:

[0116]

[0117] where NW 5,j represents the influence coefficient of competing advertisements in the new time period j;

[0118] Z y represents the effect threshold;

[0119] Specifically, when the effect score E of the y-th advertisement push time period y < the effect threshold Z y the influence coefficient NW of competing advertisements in the new time period j is used 5,j to replace the influence coefficient W5 of competing advertisements in the corresponding time period j. When pushing advertisements subsequently, the advertisement push time periods will be rearranged again, which can adjust the influence of the advertisements pushed by competitors on the advertisement effect of its own according to the actual situation, reduce the user activity score in high-competition time periods, avoid pushing advertisements in high-competition time periods, and can timely adjust the advertisement push strategy to improve the advertisement push quality.

[0120] 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.

Claims

1. An intelligent online advertising push system based on big data, characterized in that, It includes a data collection module, a data processing module, a data storage module, an intelligent push module, and a user feedback module; The data collection module is used to collect user data through various channels. The user data includes user behavior data, historical advertisement data, social media interaction data, and user identity information; The intelligent push module includes a user interest analysis unit, a push timing unit, and an effect evaluation unit; The user interest analysis unit uses an interest evaluation algorithm to analyze the degree of interest in different advertisement categories based on the user data within a past unit time, and determines the preference degree of the user for each type of advertisement. The push timing unit uses a segmented evaluation algorithm to divide a day into multiple time periods, calculates the activity level of the user in each time period respectively, and determines the advertisement push intensity and the advertisement push time period according to the high or low activity level of the user in each time period. The effect evaluation unit evaluates the push effect of each advertisement push time period by using an advertisement effect evaluation algorithm, sets a threshold for the result of the advertisement effect evaluation algorithm, and judges whether the advertisement push effect is qualified. When it is unqualified, the data in the segmented evaluation algorithm is adjusted.

2. The intelligent online advertising intelligent push system based on big data according to claim 1, characterized in that: The analysis process of the interest evaluation algorithm is as follows: where S u,b represents the interest score of user u in ad category b; CA u,b represents the number of times user u clicks on ad category b; CA ave,b represents the average number of clicks on advertisement category b; W1 represents the click - through rate impact coefficient, and its value range is from 0 to 1; T u,b Indicates the duration for which user u views ad category b; T u,all Indicates the total duration of the advertisement viewed by user u; W2 represents the advertisement viewing duration impact coefficient, and its value range is from 0 to 1; D u,b Indicates the quantity of products of advertisement category b purchased by user u; D ave,b Represents the average value of purchasing products of advertising category b; W3 represents the product purchase impact coefficient, and its value range is from 0 to 1; The interest score S of user u for advertisement category b u,b represents the degree of preference of the user for the advertisement. The interest score S of user u for advertisement category b u,b The larger the score, the higher the acceptance degree of the user for this type of advertisement. Conversely, the lower the score. When pushing advertisements, the advertisement category with the highest score is pushed.

3. The intelligent online advertising intelligent push system based on big data according to claim 2, characterized in that: The process of the segmented evaluation algorithm is as follows: Among which S j,u represents the activity score of user u in time period j; RT j Represents the platform access duration for time period j; T j，all represents the total duration of time period j, i.e., the length of a time period; W 4,j Indicates the influence coefficient of access duration within time period j, with a value range of 0 to 1; L j Indicates the occupied ad volume in time period j, that is, the number of ad slots pushed by competitors within time period j; L all represents the total number of ad spaces provided by the platform; W 5,j Indicates the competitive advertisement influence coefficient within time period j, with a value range of 0 to 1; The activity score S of user u in time period j j,u The higher it is, the less the activity of the user and the proportion of competing advertisements in the current time period. According to the activity score S of user u in time period j j,u Take the top y advertisement push time periods according to the level of S, and determine the advertising budget investment intensity according to the ranking of the advertisement push time periods. Pushing advertisements during this time period can increase the click-through rate and conversion rate of the advertisements, and introduce the factor of the proportion of competing advertisements to simulate the competition environment in the advertising market and improve the accuracy of the segmented evaluation algorithm.

4. The intelligent online advertising intelligent push system based on big data according to claim 3, characterized in that: The process of the advertisement effect evaluation algorithm is as follows: Among which E y represents the effect score of the y-th advertisement push time period; C y represents the click volume of the y-th advertisement push time period; B y Represents the exposure volume in the y-th advertisement push time period, that is, the number of times the advertisement is displayed; α represents the advertisement click impact coefficient, and its value range is from 0 to 1; R y represents the purchase volume in the y-th advertisement push time period, that is, the number of purchases completed by clicking on the advertisement; β represents the advertisement purchase impact coefficient, and its value range is from 0 to 1; P ave represents the average order price; S y represents the budget for the y-th advertisement push time period; γ represents the advertisement return impact coefficient, and its value range is from 0 to 1; S u,b Indicates the interest score of user u in ad category b; The interest score S of user u for advertisement category b u,b is substituted into the advertisement effect evaluation algorithm, indicating that the interest score of user u for advertisement category b is the highest; The effectiveness score E of the y-th advertisement push time period y The higher it is, the better the overall attractiveness of the advertisement and the higher the conversion rate. Conversely, the worse it is. The value range of y is from 1 to 3. When y equals 1, E1 represents the effectiveness score of the first advertisement push time period. The advertising budget investment intensity in this time period is 0.

5. The activity score S of the corresponding user u in the time period j of the first advertisement push time period j,u The time period with the highest score; When y is equal to 2, E2 represents the effectiveness score of the second advertising push time period. The advertising budget investment intensity is 0.3, and the activity score S of the corresponding user u in time period j of the second advertising push time period j,u is the second highest score time period; When y equals 3, E3 represents the effectiveness score of the third advertisement push time period, the advertising budget investment intensity is 0.2, and the activity score S of the corresponding user u in time period j of the third advertisement push time period j,u is the third highest score time period.

5. The intelligent online advertising intelligent push system based on big data according to claim 4, wherein: Set the effect score E for the y-th advertisement push time period y The effect threshold is Z y , when the effect score E of the y-th advertisement push time period y <effect threshold Z y , it means that the advertisement effect is too low. At this time, increase the influence coefficient W5 of competing advertisements within the corresponding time period j. The adjustment process is as follows: Among them, NW 5,j represents the competitive advertisement influence coefficient within the new time period j; Z y represents an effect threshold; Effect score E in the y-th advertisement push time period y <Effect threshold Z y When this occurs, use the competitive advertisement influence coefficient NW in the new time period j 5,j to replace the competitive advertisement influence coefficient W5 in the corresponding time period j, which can adjust the influence of the advertisements already pushed by competitors on its own advertisement effect according to the actual situation, reduce the user activity score in high-competition time periods, avoid pushing advertisements during high-competition time periods, improve the effectiveness of advertisement pushing, and re-arrange the advertisement push time periods during subsequent advertisement pushing.

6. The intelligent online advertising intelligent push system based on big data according to claim 5, wherein: The user feedback module is used to build a feedback opinion platform and an intelligent customer unit. Users log in to the feedback opinion platform and the intelligent customer unit to send suggestions and inquire about advertisement - related information. The problems collected by the intelligent customer unit and the suggestions collected by the user feedback module are both transmitted to the data storage module for storage.

7. The intelligent online advertising intelligent push system based on big data according to claim 1, characterized in that: The data processing module is used to perform deduplication, denoising, and outlier detection processing on the user data collected by the data collection module.

8. The intelligent online advertising intelligent push system based on big data according to claim 1, wherein: It also includes an encryption protection module. The encryption transmission module is used to encrypt the user data to ensure that it is not leaked during storage and transmission.

Images