Advertisement push marketing method and system based on big data

Abstract

The invention relates to the technical field of data processing, in particular to an advertisement push marketing method and system based on big data. The method comprises the following steps: acquiring behavior data when each advertisement is pushed to each user and the age of each user in recent time; determining a score value of the pushed user to the advertisement; determining accepted users in the pushed users of the target advertisement according to the score values of all the pushed users for the advertisement; determining a preliminary similarity between the advertisements; determining the final similarity between the advertisements; determining a recommendation prediction value of the advertisement to the non-pushing user; and according to the size sequence of the recommendation prediction values, advertisement pushing marketing is carried out on the user. According to the method, the advertisement similarity is determined by combining the acknowledged user behavior data of the advertisement, the average age and other factors, the difference of different advertisement audiences is fully considered, and the problem of low pushing accuracy caused by neglecting the objective characteristics of the advertisement is avoided.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an advertisement push marketing method and system based on big data. Background Art

[0002] With the rapid development of Internet technology, mobile communication technology, artificial intelligence, and the Internet of Things, the digital transformation of all walks of life in society is accelerating. Data has become one of the most important production factors in modern society, especially in the advertising and marketing industry. Big data technology provides accurate user portraits, behavior analysis, and decision support for advertising push. Traditional advertising marketing models usually rely on radio and television broadcasts, and the effect evaluation is not accurate enough. The advertising push method based on big data can provide users with customized advertising content through real-time data analysis and user behavior prediction, improve the efficiency and accuracy of advertising delivery, and reduce advertising costs. An existing algorithm for advertising push is the collaborative filtering algorithm, which has high recommendation accuracy and high computational efficiency, making the algorithm have good application prospects in the field of advertising push and can significantly improve advertising effects and user experience.

[0003] The patent document with announcement number CN106095841B discloses a mobile Internet advertising recommendation method based on collaborative filtering, the method comprising: step 1, obtaining user check-in data and microblog text data; step 2, preprocessing the user check-in data and microblog text data; step 3, calculating the user's interest similarity; step 4, using a hierarchical clustering method to analyze the user check-in data, comparing the similarity of the user behavior trajectories from both overall and local aspects according to the spatiotemporal intersection of the user's behavior trajectories, and obtaining the similarity of the user behavior trajectories through linear combination; step 5, comprehensively considering the user's interest similarity, the similarity of the user's behavior trajectories, the user's behavior similarity in different time periods and the influence of the popularity index of the advertisement, using a collaborative filtering recommendation method to calculate the recommendation value of the mobile Internet advertisement and recommend the top-ranked advertisements to the user.

[0004] However, the above patent document does not solve the problem that when using collaborative filtering algorithm to calculate the similarity between advertisements, it only relies on user interaction data and does not take into account the different target audiences of different advertisements. As a result, the algorithm can only calculate the similarity based on user behavior, ignoring the objective characteristics of the advertisements. Then, if some advertisements show similarities in user behavior but are completely different in target audiences, the algorithm may still recommend them, resulting in a decrease in the accuracy of advertising push. Summary of the invention

[0005] In order to solve the problem that traditional collaborative filtering algorithms rely on user interaction data, fail to take into account the different audiences of different advertisements, ignore the objective characteristics of advertisements, and thus reduce the accuracy of advertisement push, the present invention provides an advertisement push marketing method and system based on big data.

[0006] In a first aspect, the present invention provides an advertising push marketing method based on big data, which adopts the following technical solution: A method for advertising push marketing based on big data comprises: obtaining the behavior data of each advertisement when it is pushed to each user in the recent period and the age of each user; for a target advertisement in each advertisement, determining the score value of the pushed user for the target advertisement according to the behavior record of any pushed user; determining the recognized users among the pushed users of the target advertisement according to the size of the score values ​​of all pushed users for the target advertisement; for the target advertisement and any other advertisement, obtaining the similarity of the score values ​​of the two advertisements by all pushed users as the preliminary similarity of the two advertisements; determining the final similarity of the two advertisements according to the preliminary similarity, the total number of purchases of the purchase records in the behavior data of the recognized users of the two advertisements, and the average age of the recognized users of the two advertisements; taking any non-pushed user of the target advertisement as the target user, determining the recommendation prediction value of the target advertisement for the target user according to the score value of the pushed advertisement by the target user and the final similarity between the pushed advertisement of the target user and the target advertisement; and performing advertising push marketing on the target user according to the size order of the recommended prediction values.

[0007] The beneficial effects are: by obtaining the user's age and behavior data, it is possible to build a more comprehensive and detailed user portrait, thereby improving the matching degree of advertisements, which can not only reflect the user's interest preferences, but also capture the user's potential needs, and provide more efficient data support for advertisement recommendations; through comprehensive analysis of the score value, similarity analysis, recognized user behavior data and purchase records of the pushed advertisements, it is possible to effectively improve the accuracy of advertisement recommendations, no longer relying solely on the user's historical behavior data, but taking into account factors such as the characteristics of the advertisement content itself and the preferences of the audience group, so as to achieve accurate push; traditional advertisement push methods may cause information overload and advertisement fatigue to users, By pushing advertisements that are highly matched with target users, the push of irrelevant advertisements is reduced, avoiding users' negative emotions caused by excessive advertisement push. Since advertisement push is more accurate, users' interest and acceptance of advertisements will be significantly improved, thereby improving the conversion rate of advertisements. By combining users' historical behavior, ratings and advertisement similarity, users' reactions to advertisements can be predicted more accurately, thereby optimizing advertisement delivery strategies. Through real-time feedback and rating data analysis of big data, the system can continuously optimize advertisement push strategies. In actual operations, the results of advertisement push can be adjusted according to the ever-changing user behavior and advertisement effects, ensuring that advertising marketing activities always remain efficient.

[0008] Furthermore, the behavior data also includes display rating values ​​and browsing time.

[0009] Furthermore, the score satisfies: ; In the formula, For the Sent to users The rating value of the advertisements. For the Sent to users The explicit rating value in the behavioral data of the advertisement, For the Sent to users The viewing time in the behavioral data of ads, For the Sent to users The contribution value of purchase records in the behavioral data of advertisements, For the preset contribution time, Contribute to the preset viewing time. Contribute to the preset viewing time limit. is the standard normalization function, is the minimum function, The floor symbol.

[0010] The beneficial effects are as follows: by combining multiple factors such as users' explicit ratings, viewing time, and purchase records, the rating value is more comprehensive and can more accurately reflect users' interest in and interaction with advertisements; the preset contribution value and upper limit of viewing time are used to effectively avoid the excessive influence of single duration data on the rating, making the rating more reasonable and stable; through the standard normalization function, the differences between different users and advertisements are eliminated, making the ratings more consistent and comparable, and improving the fairness of the system; the flexibility of the rating mechanism can be adjusted according to the needs of different advertisements and users, thereby achieving more personalized and accurate advertising recommendations, and improving advertising conversion rates and user satisfaction.

[0011] Furthermore, the approved users are selected by sorting the ratings of the target advertisements by all the users who have pushed the advertisements, and a preset percentage of users are selected as approved users.

[0012] The beneficial effect is: by utilizing the behavioral data and interest preferences of recognized users (i.e., users who have a high score for the target advertisement), it is possible to recommend advertisements that are more in line with the interests and needs of users who are not subsequently pushed the advertisement, thereby improving the relevance of the advertisement and user engagement; for users who are not pushed the advertisement, directly pushing the advertisement may involve the risk of inaccuracy. By drawing on the preferences of recognized users, it is possible to more quickly find a suitable advertisement recommendation path for users who are not pushed the advertisement, thereby reducing "information overload" and improving the browsing experience.

[0013] Furthermore, the similarity is cosine similarity.

[0014] Furthermore, the final similarity satisfies: ; In the formula, For the Article and The final similarity of the ads, For the Article and The initial similarity of the ads, and Respectively Article and The total number of purchases recorded in the behavioral data of users who recognized the advertisement, and Respectively Article and The average age of users who recognized the ads, is a hyperparameter, is the standard normalization function.

[0015] The beneficial effects are: by combining the similarity of advertising content, purchase history and user age, the accuracy of advertising similarity calculation is improved, the interests and needs of users are better met, and the relevance of advertising is improved; the most suitable advertisements for users are effectively screened out, the click-through rate and conversion rate are increased, and the efficiency of advertising delivery is improved through user behavior and feature data.

[0016] Furthermore, the recommended prediction value satisfies: ; In the formula, For the Ads were not pushed to users The recommended prediction value for each user, For the The number of ads delivered to users, For the Article and The final similarity of the ads, For the Users responded to the first The rating value of the ad.

[0017] The beneficial effects are: by combining ad similarity and user ratings, accurate predictions are made for users who have not been pushed the ad, thereby improving the personalization and relevance of the push; by comprehensively considering ad similarity and user feedback, the push is more in line with user preferences, improving ad click-through rate and conversion rate; based on rating data of user historical behavior, the push is more in line with user interests and needs; the weighting mechanism allows the system to adjust the weights of similarity and ratings according to actual needs, thereby improving the flexibility and adaptability of the push.

[0018] In a second aspect, the present invention provides an advertising push marketing system based on big data, which adopts the following technical solution: A big data-based advertising push marketing system comprises: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned big data-based advertising push marketing method is implemented.

[0019] By adopting the above technical solution, the above-mentioned advertising push marketing method based on big data is generated into a computer program and stored in a memory to be loaded and executed by a processor, so that a terminal device is made according to the memory and the processor for easy use.

[0020] The present invention has the following technical effects: The present invention no longer relies solely on user interaction data, but determines the similarity of advertisements by combining the recognized user behavior data of advertisements and factors such as the average age of recognized users, fully considering the differences in different advertising audiences, that is, the objective characteristics of advertisements, avoiding the problem of reduced push accuracy caused by ignoring these factors, and can more accurately characterize the relationship between advertisements, laying the foundation for subsequent accurate push; based on the final similarity of advertisements determined by the above comprehensive considerations, further combined with the target user's rating value of the pushed advertisement, to determine the recommendation prediction value of the target advertisement to the target user, this multi-dimensional data fusion method comprehensively and meticulously analyzes the relationship between users and advertisements, and compared with traditional collaborative filtering algorithms, it can more accurately predict the target user's interest in the target advertisement, thereby performing advertisement push marketing according to the order of the recommended prediction value, greatly improving the accuracy of advertisement push, and making it more likely that the advertisement will reach users who are truly interested; by obtaining the behavior data of each advertisement when it is pushed to each user in the recent period, comprehensively collecting and utilizing big data information, and deeply exploring the potential needs and preferences behind user behavior, refined advertisement push marketing based on big data is realized, and the effect and efficiency of advertisement marketing are effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-restrictive manner, and the same or corresponding numbers are the same or corresponding parts.

[0022] Figure 1 It is a method flow chart of an advertising push marketing method based on big data in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] 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 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0024] It should be understood that when the terms "first", "second", etc. are used in the claims, descriptions, and drawings of the present invention, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the description and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their collections.

[0025] The embodiment of the present invention discloses an advertising push marketing method based on big data, referring to Figure 1 , comprising steps S1 to S7: S1: Obtain the behavior data of each advertisement when it is pushed to each user in the recent period and the age of each user.

[0026] Big data technology is used to obtain the behavioral data of each advertisement pushed to each user in the recent period (for example, within one month), including explicit rating values, viewing time and purchase records (including whether and how many times the advertisement was purchased), as well as the age of each user.

[0027] S2: For a target advertisement in each advertisement, a rating value of the target advertisement by any user to which the advertisement has been pushed is determined according to the behavior record of the user to which the advertisement has been pushed.

[0028] It should be noted that this step uses the explicit ratings, viewing time, and purchase records of different users for different advertisements to analyze and calculate the ratings of different users for different advertisements, so as to facilitate the subsequent calculation of the preliminary similarity between different advertisements based on the ratings of different users for different advertisement data. Therefore, different scores need to be assigned to different user behaviors, which are subsequently used to calculate the rating of a user for a pushed advertisement, wherein the normalized value of a user's explicit rating for an advertisement will be identified as the score contributed by a user's explicit rating behavior for an advertisement to the user's rating of the advertisement; for every increase in the viewing time of a user for an advertisement, the score contributed by a user's viewing time behavior for the user's rating of the advertisement will increase accordingly until the upper limit; if a user purchases an advertisement, the score contributed by a user's purchase behavior for an advertisement to the user's rating of the advertisement will be greater.

[0029] Specifically, the scoring value satisfies: ; In the formula, For the Sent to users The rating value of the advertisements. For the Sent to users The explicit rating value in the behavioral data of the advertisement, For the Sent to users The viewing time in the behavioral data of ads, For the Sent to users The contribution value of purchase records in the behavioral data of advertisements, For the preset contribution time, Contribute to the preset viewing time. Contribute to the preset viewing time limit. is the standard normalization function, is the minimum function, The floor symbol.

[0030] Among them, exemplary , , and Indicates Sent to users For every 5 seconds that the viewing time of an ad increases, the score contributed by the user to the viewing time of the ad increases by 0.1. Indicates the upper limit of the score contribution of a user's viewing time for an advertisement. , Indicates the first Sent to users The viewing time of an ad is the smaller value between the actual score contributed by the user to the ad rating and the upper limit of the score. This is to reduce the influence of other things on the user, which may cause the user to stay on an ad page for too long without actually viewing the ad. Indicates Sent to users The contribution value of the purchase record in the behavior data of the advertisement. For example, if a purchase behavior occurs, If no purchase occurs, .

[0031] S3: Determine the users who approve the target advertisement among the users to whom the target advertisement has been pushed, based on the ratings of all the users to whom the target advertisement has been pushed.

[0032] It should be noted that in order to more accurately quantify the similarity of advertisements with respect to target audiences, the users corresponding to the top 10% (which can be adjusted based on specific implementation circumstances) of the ratings given by users to each advertisement have been pushed to them are recognized users.

[0033] Specifically, the approved users are selected as approved users by sorting the ratings of the target advertisements by all the users who have pushed the advertisements, and selecting a preset percentage of users.

[0034] S4: For the target advertisement and any other advertisement, obtain the similarity of the rating values ​​of the two advertisements by all pushed users as the preliminary similarity of the two advertisements.

[0035] Specifically, the preliminary similarity satisfies: ; In the formula, For the Article and The initial similarity of the ads, is the number of users who watched the above two ads at the same time. and Respectively Sent to users Article and Since cosine similarity is a prior art, the logical principle of this formula will not be described in detail.

[0036] S5: Determine the final similarity of the two advertisements based on the preliminary similarity, the total number of purchases of the purchase records in the behavior data of the users who recognize the two advertisements, and the average age of the users who recognize the two advertisements.

[0037] It should be noted that, in order to take into account the different characteristics of the audiences of different advertisements, this step will analyze the age data and purchasing power data of the pushed users corresponding to different advertisements, obtain the similarity of the two advertisements with respect to the audiences, and correct the preliminary similarity between the two advertisements based on this indicator to obtain the final similarity between the two advertisements; when analyzing the similarity of the two advertisements with respect to the audiences, among the ratings of all users for the two advertisements, the smaller the difference in the total number of purchases made through the advertisements by the recognized users in the past month, the greater the similarity of the two advertisements with respect to the audiences; however, different advertisements may have similar appeal to users of different age groups, so only analyzing the difference in users' purchasing power cannot well distinguish the audiences of different advertisements; therefore, the smaller the difference in the average age of the recognized users, the greater the similarity of the two advertisements with respect to the audiences.

[0038] Specifically, the final similarity satisfies: ; In the formula, For the Article and The final similarity of the ads, For the Article and The initial similarity of the ads, and Respectively Article and The total number of purchases recorded in the behavioral data of users who recognized the advertisement, and Respectively Article and The average age of users who recognized the ads, is a hyperparameter, is the standard normalization function.

[0039] Implementers can set hyperparameters according to specific implementation conditions, for example, 0.001. The existence of hyperparameters is to prevent and A value of 0 makes the formula meaningless.

[0040] in, The larger the value, the greater the initial similarity between the two ads based on the analysis of user behavior data, and the greater the final similarity between the two ads. The smaller it is, the smaller the difference in the total number of times the approving users purchased through the advertisements in the past month is. This means that the audiences of the two advertisements are more similar, and the final similarity of the two advertisements will be greater. The smaller it is, the smaller the difference in the average age of all recognized users is. This means that the greater the similarity between the audiences of the two advertisements, the greater the credibility, and correspondingly, the greater the final similarity between the two advertisements will be.

[0041] S6: taking any user to whom the target advertisement has not been pushed as the target user, and determining the recommendation prediction value of the target advertisement for the target user according to the target user's rating value of the pushed advertisement and the final similarity between the target user's pushed advertisement and the target advertisement.

[0042] It should be noted that the final similarity between all advertisements is obtained through analysis and calculation in the above steps. In this step, the final similarity between an advertisement and the remaining advertisements that have been pushed to users to which the advertisement has not been pushed (the final similarity is calculated based on the remaining users to whom the advertisement has been pushed), and then the weighted summation of the ratings of the remaining advertisements that have been pushed to users to which the advertisement has not been pushed is combined to obtain the recommendation prediction value of the advertisement for the users to be pushed.

[0043] Specifically, the recommended prediction value satisfies: ; In the formula, For the Ads were not pushed to users The recommended prediction value for each user, For the The number of ads delivered to users, For the Article and The final similarity of the ads, For the Users responded to the first The rating value of the advertisements. For the of the ads sent to users Advertisements.

[0044] S7: Push advertising marketing to target users according to the order of the recommended prediction values.

[0045] Obtain the recommended predicted values ​​of all advertisements for target users, sort them according to the size of the recommended predicted values, and then push them to the target users in order to complete the advertising push marketing based on big data.

[0046] An embodiment of the present invention also discloses an advertising push marketing system based on big data, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an advertising push marketing method based on big data according to the present invention is implemented.

[0047] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0048] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, or any other medium that can be used to store the required information and can be accessed by an application, module, or both. Any such computer storage medium may be part of the device or accessible or connectable to the device.

[0049] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

[0050] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An advertising push marketing method based on big data, characterized in that: include: Obtain the behavior data of each advertisement when it is pushed to each user in the recent period and the age of each user; For a target advertisement in each advertisement, a rating value of the target advertisement by any user to which the advertisement has been pushed is determined according to the behavior record of the user to which the advertisement has been pushed; Determine the users who approve the target advertisement among the users who have been pushed the target advertisement according to the ratings of all the users who have been pushed the target advertisement; For the target ad and any other ad, obtain the similarity of the ratings of the two ads by all pushed users as the preliminary similarity of the two ads; Determine the final similarity of the two advertisements based on the preliminary similarity, the total number of purchases of the purchase records in the behavioral data of the users who recognized the two advertisements, and the average age of the users who recognized the two advertisements; Taking any user to whom the target advertisement has not been pushed as the target user, and determining the recommendation prediction value of the target advertisement for the target user according to the target user's rating value of the pushed advertisement and the final similarity between the target user's pushed advertisement and the target advertisement; Advertisement push marketing is performed to target users according to the order of the recommended prediction values.

2. The advertising push marketing method based on big data according to claim 1, characterized in that: The behavior data also includes display rating values ​​and browsing time.

3. The advertising push marketing method based on big data according to claim 2 is characterized in that: The rating value satisfies: ； In the formula, For the Sent to users The rating value of the advertisements. For the Sent to users The explicit rating value in the behavioral data of the advertisement, For the Sent to users The viewing time in the behavioral data of ads, For the Sent to users The contribution value of purchase records in the behavioral data of advertisements, For the preset contribution time, Contribute to the preset viewing time. Contribute to the preset viewing time limit. is the standard normalization function, is the minimum function, The floor symbol.

4. The advertising push marketing method based on big data according to claim 1, characterized in that: The approved users are selected as approved users by sorting the ratings of the target advertisements by all the users who have pushed the advertisements, and a preset percentage of users are selected.

5. The advertising push marketing method based on big data according to claim 1, characterized in that: The similarity is cosine similarity.

6. The advertising push marketing method based on big data according to claim 1, characterized in that: The final similarity satisfies: ； In the formula, For the Article and The final similarity of the ads, For the Article and The initial similarity of the ads, and Respectively Article and The total number of purchases recorded in the behavioral data of users who recognized the advertisement, and Respectively Article and The average age of users who recognized the ads, is a hyperparameter, is the standard normalization function.

7. The advertising push marketing method based on big data according to claim 1, characterized in that: The recommended prediction value satisfies: ； In the formula, For the Ads were not pushed to users The recommended prediction value for each user, For the The number of ads delivered to users, For the Article and The final similarity of the ads, For the Users responded to the first The rating value of the ad.

8. An advertising push marketing system based on big data, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an advertising push marketing method based on big data according to any one of claims 1-7 is implemented.

Images