Intelligent advertisement pushing method based on Internet advertisement platform

Abstract

The invention relates to the technical field of internet advertisement services, and particularly discloses an intelligent advertisement pushing method based on an internet advertisement platform, which comprises the following steps: S1, taking the time when a user logs in the platform as a starting point, obtaining the stay duration of the user, and calculating the attention value of the user; s2, obtaining the advertisement pushing times of the platform in the monitoring period, and calculating the quality of the advertisement; s3, obtaining the number of times that the advertisement watching duration of the user is smaller than a preset value, correcting the quality to obtain comprehensive quality, calculating the mean value of the comprehensive quality of the advertisement, and setting an initial coefficient of the advertisement; and S4, setting an initial attenuation factor based on the comprehensive quality and the mean value, sorting and pushing the advertisements, updating the initial coefficient of the advertisements, and re-sorting and pushing the advertisements. Advertisement pushing is carried out on the user in a polling mode, the current situation that a head platform monopolizes flow and data is reduced, small and medium-sized enterprises obtain fair display opportunities, the Mortai effect is weakened, and innovation vitality is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet advertising services, and in particular to an intelligent advertising push method based on an Internet advertising platform. Background Art

[0002] Push advertising is an integral part of modern commerce and the digital ecosystem. It not only serves business growth needs, but also influences consumer decision-making experiences, driving advancements in marketing technology as a whole. For businesses, push advertising is a highly effective tool for driving sales, building brands, and capturing user feedback. Through targeted delivery, it delivers product information to potential customers, shortens the purchase process, and optimizes marketing strategies through data. For consumers, effective push advertising can reduce information search costs and quickly match their needs. For example, local service ads help users discover nearby deals, while interest-based recommendations provide content more tailored to their preferences.

[0003] While push advertising has become a core tool in digital marketing, it still faces numerous challenges in practice, particularly structural issues in the advertising market. The monopoly of traffic and data by leading platforms makes it difficult for small and medium-sized enterprises and emerging brands to obtain fair exposure. This Matthew effect further solidifies the market structure and stifles innovation. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent advertising push method based on an Internet advertising platform to solve the above technical problems.

[0005] The purpose of the present invention can be achieved through the following technical solutions: An intelligent advertising push method based on an Internet advertising platform comprises the following steps: S1: Taking the time when the user logs into the platform as the starting point, obtain the user's current stay time T and calculate the user's attention value ,in, Represents the average length of stay of the user; S2: Get the number of times N the platform pushes the same brand advertisement within the preset monitoring period and calculate the quality of the advertisement ,in, represents the total playing time of the ad during the monitoring period, t represents the single playing time of the ad, and S represents the number of times users enter the product interface through the ad; S3: Obtain the number of times P that users watch ads for less than 1 second and correct the quality Q. The corrected quality Q is recorded as the comprehensive quality ,in, Represents the preset first and second coefficients and , calculate the mean of the comprehensive quality of advertisements , and set the initial coefficient of the advertisement ; S4: Based on comprehensive quality and mean Set the initial decay factor Y and perform the following operations on the ads in the platform: Sort the ads according to the initial coefficient X from low to high, let the platform push the ad that is currently ranked first, and update the initial coefficient X of all ads except the current first, and record the updated initial coefficient X as the coefficient ,in, Represents the single playback duration of the pushed ad, and μ represents the preset one-unit correction coefficient; When this push ends, follow the above steps and re-sort from low to high according to the coefficient x, push the advertisement with the highest ranking, and repeat the push until the user exits the platform.

[0006] As a further solution of the present invention: in the step S4, it includes: When the top-ranked ad is pushed, its initial coefficient X is updated and recorded as the waiting coefficient And the waiting coefficient , let the waiting coefficient Dx re-participate in the sorting according to the above steps until the user exits the platform.

[0007] As a further solution of the present invention: in the step S1, if the user stays for a period of time T less than the minimum duration of a single advertisement playback, , then stop pushing ads to the user. When the user stays for Start pushing advertisements.

[0008] As a further solution of the present invention: in the step S1, the time interval of the user logging into the advertising platform is obtained. If the time interval is less than a preset judgment threshold, it is recorded as a login behavior.

[0009] As a further solution of the present invention: comprising: The ads complained by users will be marked, and when ads are pushed to the user in the future, they will be excluded from participating in the subsequent steps.

[0010] As a further solution of the present invention: in the step S2, when the quality Q of the advertisement is 0, it is recorded as a useless advertisement, and the useless advertisement is removed and not pushed.

[0011] As a further solution of the present invention: in step S4, based on the comprehensive quality and mean Set the initial decay factor .

[0012] As a further solution of the present invention: in the step S4, if there are advertisements with the same initial coefficient X, the advertisement with the fewer total push times is placed at the front of the ranking.

[0013] The beneficial effects of the present invention are as follows: first, based on the time when the user logs into the platform as the starting point, the user's current stay time is obtained, and the user's attention value is calculated. This value is used to preliminarily judge the time and energy the user is willing to spend on this. When pushing advertisements subsequently, the initial coefficient of the advertisement can be set according to the attention value here. The higher the attention value, the larger the initial coefficient of the advertisement set subsequently, and the longer the push interval. This can ensure the rationality of the advertisement push. The larger the attention value, the stronger the user's tolerance, and there is enough time to push advertisements for them.

[0014] The purpose of setting up a monitoring cycle is to obtain the content and quality of advertisements that users are exposed to in a short period of time, so as to infer the feedback effect of a brand's advertisements during this period. If you want to understand the specific situation, you need to calculate the quality of the advertisement. Since the quality of an advertisement is not a quantifiable value, and different users have different acceptance of different advertisements, the quality calculated here is only for a single user and is not a reference for the whole user. The quality of the advertisement can be roughly seen by calculating the ratio of the total time users watch the advertisement to the total playback time, plus the number of times users click on the advertisement. From another perspective, it represents the user's interest in the advertisement.

[0015] Then the quality of the advertisement needs to be corrected. After watching the advertisement many times, the user will have resistance. At this time, if the quality of the advertisement is not enough to meet the user's requirements, the user will not be willing to stop and watch the advertisement. Therefore, it can be judged that when the user watches the advertisement for less than a certain time, it can be determined that the user has developed resistance to the advertisement. At this time, it is necessary to count the number of times the user watches the advertisement for less than 1 second, and correct the quality through a weighted formula, and record the corrected quality as the comprehensive quality.

[0016] Then set the initial coefficient for the advertisement, which is equivalent to a countdown. The smaller the value, the faster the advertisement is pushed. According to the formula, in order to prevent the head platform from monopolizing most of the traffic and data, the more advertisements are pushed during the monitoring period, the larger the corresponding value will be. This will reduce the push frequency. The higher the quality of the advertisement, the smaller the initial coefficient, and the faster the advertisement push frequency will be.

[0017] Then the initial attenuation factor is calculated. According to the formula, it can be seen that the value of the initial attenuation factor is related to the comprehensive quality of the advertisement itself and the mean quality. When the comprehensive quality is greater than the mean quality, the attenuation factor is greater. The role of the attenuation factor is to subtract the attenuation factor from the initial coefficient after each advertisement is pushed, and then select an advertisement with the smallest initial coefficient from all advertisements for push. The advantages of doing this are, first, it can take into account the user's evaluation of the quality of the advertisement and make a rough judgment. Second, through this polling-like method, while ensuring that high-quality advertisements have more push opportunities, it ensures that there will be no monopoly effect caused by the traffic impact of the head platform, so that small and medium-sized enterprises can obtain fair display opportunities, weaken the Matthew effect and enhance innovation vitality. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The present invention will be further described below with reference to the accompanying drawings.

[0019] Figure 1 It is a flow chart of an advertisement intelligent push method based on an Internet advertisement platform of the present invention. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making any creative efforts shall fall within the scope of protection of the present invention.

[0021] See also Figure 1 As shown, the present invention is an advertisement intelligent push method based on an Internet advertising platform, comprising the following steps: S1: Taking the time when the user logs into the platform as the starting point, obtain the user's current stay time T and calculate the user's attention value ,in, Represents the average length of stay of the user; S2: Get the number of times N the platform pushes the same brand advertisement within the preset monitoring period and calculate the quality of the advertisement ,in, represents the total playing time of the ad during the monitoring period, t represents the single playing time of the ad, and S represents the number of times users enter the product interface through the ad; S3: Obtain the number of times P that users watch ads for less than 1 second and correct the quality Q. The corrected quality Q is recorded as the comprehensive quality. ,in, Represents the preset first and second coefficients and , calculate the mean of the comprehensive quality of advertisements , and set the initial coefficient of the advertisement ; S4: Based on comprehensive quality and mean Set the initial decay factor Y and perform the following operations on the ads in the platform: Sort the ads according to the initial coefficient X from low to high, and let the platform push the ad that is currently ranked first, and update the initial coefficient X of all ads except the current first, and record the updated initial coefficient X as the coefficient ,in, Represents the single playback duration of the pushed ad, and μ represents the preset one-unit correction coefficient; When this push ends, follow the above steps and re-sort from low to high according to the coefficient x, push the advertisement with the highest ranking, and repeat the push until the user exits the platform.

[0022] It's important to note that a sound advertising strategy is crucial for improving user experience and realizing commercial value. Collecting and analyzing relevant data, starting with the time a user logs into the platform, is fundamental to building an effective advertising mechanism.

[0023] When a user logs into the platform, the system starts a timing function to track and record the length of time the user stays on the current page or in the app in real time. This dwell time carries behavioral information and is one of the important indicators for measuring user attention to content. Based on this dwell time, the user's attention value is further calculated. The calculation process of this attention value comprehensively considers the user's browsing speed between different sections, whether they frequently switch pages, the number of times they repeatedly view specific content, etc., converting these behavioral data into a quantified attention value. This attention value can provide initial insights into the amount of time and energy users are willing to invest in the platform. Longer dwell time and higher attention values ​​often mean that users have a high level of interest and engagement in the currently displayed content, and they are more willing to actively explore and gain a deeper understanding of relevant information.

[0024] When pushing ads, we set a personalized initial ad coefficient for each user based on the attention value calculated here. A higher attention value indicates a user is highly focused, and the initial ad coefficient will increase accordingly. This is because users with high attention values ​​are generally more receptive to information and are more likely to notice ads while browsing. At the same time, to minimize disruption to user experience, we appropriately extend the interval between ad pushes. This design ensures that ads effectively reach target users while avoiding user annoyance caused by overly frequent ad displays. A higher attention value indicates a more tolerant user; while focusing on the primary content, they are also willing to accept a certain level of advertising. This provides ample time and space for carefully selected, high-quality ads that align with users' interests. This approach not only improves ad conversion rates and effectiveness, but also allows users to naturally encounter valuable commercial information while enjoying high-quality content. By using user login time as a starting point to measure duration and calculate attention values, and then setting the initial ad coefficient and push interval strategy accordingly, we can improve overall platform operational efficiency and user satisfaction.

[0025] Next, a monitoring cycle is set. This measure aims to capture the specific content and quality of ads users are exposed to over a short period of time. In the digital marketing landscape, evaluating advertising effectiveness is far from static; rather, it requires dynamic tracking and real-time feedback. Monitoring cycles are a key means of achieving this goal. By setting a monitoring cycle, detailed data on user interactions with various types of ads during that timeframe can be collected. Next, ad quality metrics need to be calculated. It's important to understand that ad quality is a complex and highly subjective concept, difficult to precisely quantify. Different users have distinct cultural backgrounds, consumption habits, aesthetic preferences, and information processing methods, leading to significant differences in their receptiveness and perception of the same ad. Therefore, the ad quality calculated here is essentially a personalized assessment of a single user. It simply reflects that user's unique perception and evaluation of a particular ad and is not universally valuable for the entire user population.

[0026] To measure the quality of ads for individual users as objectively as possible, we calculate the ratio of the total ad viewing time to the total playback time. This ratio intuitively reflects the proportion of time users are willing to actively spend watching ads, and to some extent, indicates the appeal of the ad content to users. Furthermore, the number of times users click on ads is also an important consideration. Combining these two key metrics—the ratio of the total ad viewing time to the total playback time plus the number of ad clicks—can provide a rough estimate of the ad's quality. By collecting and analyzing large amounts of individual user data, we can gradually outline the dissemination patterns and effectiveness characteristics of different types of ads among different user groups, providing strong data support for subsequent ad optimization and targeted delivery.

[0027] Afterwards, more refined and dynamic adjustments to the quality of the ads will be needed to ensure that the evaluation results are more realistic. In the actual advertising process, as users view the same ad multiple times, they are likely to gradually develop resistance. When the length of time a user views an ad is less than a pre-set threshold, it can be inferred that the user has developed a clear resistance to the ad. Special attention should be paid to situations where users view an ad for less than 1 second. The reason for choosing 1 second as the critical point is that in this extremely short period of time, users pay almost no substantial attention to the ad, often quickly swiping past it or skipping it directly. Counting the number of times users view an ad for less than 1 second is like collecting "votes" on the user's dissatisfaction with the ad. Each such brief exposure represents a negative feedback from the user about the current ad.

[0028] Based on these statistics, a weighted formula will be introduced to adjust ad quality. By assigning appropriate weights to different factors, the adjusted ad quality can be calculated more scientifically and recorded as the overall quality. Specifically, if users frequently view an ad for less than one second, it indicates that the ad is seriously lacking in attracting users' attention. In this case, the weighted formula will correspondingly lower the overall quality score. Conversely, if this situation occurs less frequently, it indicates that the ad still has a certain degree of appeal, and the overall quality score will be relatively high.

[0029] Furthermore, incorporating historical user behavior patterns is crucial. Some users tend to browse quickly, and even those who find content of interest may only glance briefly. For these user groups, we apply special considerations when applying the weighting formula to ensure that their personal habits don't misjudge the true quality of ads.

[0030] After a comprehensive assessment and dynamic correction of ad quality, an initial coefficient is set for each ad. This initial coefficient acts like a countdown timer, directly determining the cadence and frequency of ad pushes. Specifically, a lower value means more frequent pushes; conversely, a higher value increases the interval between pushes. The logic behind this design is to achieve a balanced traffic distribution through a clever mathematical formula. Of particular note, the system incorporates an anti-monopoly mechanism. During the monitoring period, if an ad is over-promoted (i.e., pushed too many times), its corresponding initial coefficient automatically increases, proactively reducing the frequency of subsequent pushes. This dynamic adjustment mechanism effectively prevents leading platforms from monopolizing traffic and data through their resource advantages, maintaining a fair competitive environment in the market ecosystem. The quality of the ad content itself also plays a crucial role. Higher-quality ads have faster countdown times, allowing high-quality creatives to spread more quickly, creating a virtuous cycle of "first come, first served."

[0031] Next, the initial attenuation factor needs to be calculated. As the formula shows, its value is closely tied to two key indicators: the overall quality score of the current ad and the average quality of all ads. When an ad's overall quality is significantly higher than the industry average, the system assigns it a larger attenuation factor. The attenuation factor works by automatically deducting the corresponding attenuation factor from the ad's initial coefficient after each ad push. Subsequently, among the candidate ads, the one with the lowest initial coefficient is prioritized for the next push. This dual guarantee mechanism offers multiple advantages: First, it fully incorporates real user feedback on ad quality, enabling the system to dynamically adjust based on actual performance. Second, this algorithm, similar to round-robin scheduling, ensures that high-quality ads receive more exposure while effectively countering the traffic siphoning effect of leading platforms. Through this refined traffic management, small and medium-sized enterprises are able to break through the barriers of major platforms and emerge on a level playing field. More importantly, this mechanism breaks the "Matthew effect" in the traditional Internet field and injects continuous innovative vitality into the market - emerging brands are no longer trapped in the start-up stage due to lack of resources, and high-quality small and medium-sized businesses can also gain growth space through excellent content creation.

[0032] In another preferred embodiment of the present invention, it includes: When the top-ranked ad is pushed, its initial coefficient X is updated and recorded as the waiting coefficient And the waiting coefficient , let the waiting coefficient Follow the above steps to re-participate in the ranking until the user exits the platform.

[0033] It's worth noting that to avoid the potential problem of repeatedly pushing the same ad in a short period of time, a more balanced and orderly ad push sequence is established by adjusting the ad display rhythm and extending the time period for ads to enter the loop. For highly popular but overexposed ads, this cool-down period will be extended accordingly; while for new or infrequently shown ads, the waiting time will be shortened to ensure they receive sufficient exposure.

[0034] In another preferred embodiment of the present invention, if the user's stay time T is less than the minimum value of the single advertisement playback time, , then stop pushing ads to the user. When the user stays for Start pushing advertisements.

[0035] It is understandable that a rule is needed to balance user experience and business needs. When the user's stay time T reaches or exceeds Threshold, the targeted advertising process officially begins. This is based on deep insights into user behavior patterns. A user's effective attention span on the platform is a limited resource. If a user only stays briefly, perhaps browsing quickly or about to leave the page, forcibly inserting an ad at this time will not only fail to effectively reach the user, but will also disrupt the user's browsing continuity and even trigger negative emotions. Therefore, the system monitors user interaction data in real time to accurately identify this type of shallow engagement and suspend ad delivery to maintain the user experience.

[0036] In another preferred embodiment of the present invention, the time interval of the user logging into the advertising platform is obtained, and if the time interval is less than a preset judgment threshold, it is recorded as one login behavior.

[0037] It's important to note that the system continuously collects the time difference between two user logins to the advertising platform and dynamically compares it with a pre-set threshold. When the time interval is detected to be less than a certain threshold, it is considered a valid login and recorded. From a data collection perspective, the system uses a high-precision timer to mark each user interaction at the millisecond level, ensuring the accuracy and continuity of timestamp recording.

[0038] In another preferred embodiment of the present invention, it includes: The ads complained by users will be marked, and when ads are pushed to the user in the future, they will be excluded from participating in the subsequent steps.

[0039] It's important to note that during the subsequent ad push decision-making process, whenever a personalized ad candidate set is generated for a user, the system performs a pre-filtering operation. All ads marked with a complaint are automatically removed from the queue, regardless of their original overall quality score. This rigid exclusion rule demonstrates the utmost respect for user preferences and ensures that content that previously caused a negative experience will not reappear in the user's view.

[0040] In another preferred embodiment of the present invention, when the quality Q of an advertisement is 0, it is recorded as a useless advertisement, and the useless advertisement is removed and not pushed.

[0041] Understandably, when the system detects an ad's quality score (Q) equal to zero, it will automatically be marked as "useless." Such ineffective content is immediately and permanently removed from the candidate pool, completely blocking its entry into the push pipeline. This zero-tolerance approach not only maintains a pure user experience but also encourages advertisers to continuously improve their creative standards and production precision, fostering a virtuous content ecosystem.

[0042] In another preferred embodiment of the present invention, based on the comprehensive quality and mean Set the initial decay factor .

[0043] In another preferred embodiment of the present invention, if there are advertisements with the same initial coefficient X, the advertisement with the fewer total push times is placed at a higher position in the ranking.

[0044] It's worth noting that in the optimized design of the ad ranking algorithm, for the special case of identical initial coefficients X, when multiple ads share the same initial coefficient value, ads with fewer total pushes are prioritized at the top of the ranking queue. The core of this mechanism is to achieve fair competition by dynamically balancing exposure opportunities. Newer ads or long-tail content that haven't yet received sufficient exposure are given higher priority, thus preventing top ads from forming a monopoly due to their accumulated historical advantages. This ranking logic, which balances efficiency and fairness, essentially uses technology to tilt resources towards small and medium-sized advertisers, effectively curbing the spread of the Matthew Effect in digital marketing.

[0045] The above is a detailed description of an embodiment of the present invention. However, the content is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. An intelligent advertising push method based on an Internet advertising platform, characterized in that: The following steps are involved: S1: Taking the time when the user logs into the platform as the starting point, obtain the user's current stay time T and calculate the user's attention value ,in, Represents the average length of stay of the user; S2: Get the number of times N the platform pushes the same brand advertisement within the preset monitoring period and calculate the quality of the advertisement ,in, represents the total playing time of the ad during the monitoring period, t represents the single playing time of the ad, and S represents the number of times users enter the product interface through the ad; S3: Obtain the number of times P that users watch ads for less than 1 second and correct the quality Q. The corrected quality Q is recorded as the comprehensive quality ,in, Represents the preset first and second coefficients and , calculate the mean of the comprehensive quality of advertisements , and set the initial coefficient of the advertisement ; S4: Based on comprehensive quality Q s and mean Q ave Set the initial decay factor Y and perform the following operations on the ads in the platform: Sort the ads according to the initial coefficient X from low to high, let the platform push the ad that is currently ranked first, and update the initial coefficient X of all ads except the current first, and record the updated initial coefficient X as the coefficient , and x=X-(μYTX), where, Represents the single playback duration of the pushed ad, and μ represents the preset one-unit correction coefficient; When this push ends, follow the above steps and re-sort the ads from low to high according to the coefficient x, and push the top-ranked ad. Repeat the push until the user exits the platform.

2. The method for intelligently pushing advertisements based on an Internet advertising platform according to claim 1, characterized in that: In the step S4, it includes: When the top-ranked ad is pushed, its initial coefficient X is updated and recorded as the waiting coefficient And the waiting coefficient , let the waiting coefficient Follow the above steps to re-participate in the ranking until the user exits the platform.

3. The method for intelligently pushing advertisements based on an Internet advertising platform according to claim 1, characterized in that: In step S1, if the user stays for a certain time T less than the minimum duration of a single advertisement playback, , then stop pushing ads to the user. When the user stays for T≥T min Start pushing advertisements.

4. The method for intelligently pushing advertisements based on an Internet advertising platform according to claim 1, characterized in that: In the step S1, the time interval for the user to log into the advertising platform is obtained. If the time interval is less than a preset judgment threshold, it is recorded as a login behavior.

5. The method for intelligently pushing advertisements based on an Internet advertising platform according to claim 1, characterized in that: include: The ads complained by users will be marked, and when ads are pushed to the user in the future, they will be excluded from participating in the subsequent steps.

6. The method for intelligently pushing advertisements based on an Internet advertising platform according to claim 1, characterized in that: In the step S2, when the quality Q of the advertisement is 0, it is recorded as a useless advertisement, and the useless advertisement is removed and not pushed.

7. The method for intelligently pushing advertisements based on an Internet advertising platform according to claim 1, characterized in that: In step S4, based on the comprehensive quality and mean Set the initial decay factor .

8. The method for intelligently pushing advertisements based on an Internet advertising platform according to claim 1, characterized in that: In step S4, if there are advertisements with the same initial coefficient X, the advertisement with the fewer total push times is placed at a higher position in the ranking.

Images