An advertisement optimization method and system based on big data processing

Abstract

This invention discloses an advertising optimization method and system based on big data processing, belonging to the field of big data processing and advertising optimization technology. The method acquires historical advertising interaction data of users in the current session, determines type weight scores based on advertising type, determines behavior weight scores based on operational behavior, and accumulates these to obtain the cumulative score for the current session. When the cumulative score exceeds a preset threshold, the display of commercial advertisements in subsequent ad slots is paused, and platform-recommended content is displayed instead. This invention also proposes optimization mechanisms such as determining additional weight scores based on playback duration, correcting behavior weight scores based on the time difference of continuous operational behavior, and dynamically restoring ad display based on user browsing depth. This invention effectively solves the problems of insufficient perception of in-session advertising fatigue effect, low efficiency of advertising resource allocation, and decreased user experience in existing technologies, achieving dual optimization of advertising effectiveness and user experience.

Description

Technical Field

[0001] This application relates to the fields of big data processing and advertising optimization technology, and more specifically, to an advertising optimization method and system based on big data processing. Background Technology

[0002] The core of internet advertising lies in accurately reaching potential users to improve efficiency. However, in high-density scenarios such as short videos and live streaming, users are exposed to multiple ads in succession, which leads to a gradual decline in attention, click-through rate, and conversion probability of subsequent ads, creating an "in-session ad fatigue" effect.

[0003] Existing optimization methods primarily predict the independent value of individual ad impressions, assuming that each impression is relatively independent. However, when users are exposed to multiple ads consecutively within a short period, their interest in subsequent ads decreases as the number of views increases. Current methods fail to adequately capture this dynamic effect and may still predict a high independent value for each ad placement in a series, ignoring that the previous ad has already consumed some of the user's attention.

[0004] This neglect directly leads to inefficient resource allocation: high-value ads may be mistakenly placed in subsequent positions where users are already fatigued, resulting in actual effects far below expectations. Simultaneously, the rapid scrolling nature of short videos and live streams demands extremely high response times; when deciding to display the next ad, the system often lacks sufficient time to collect and process real-time user feedback on previous ads. This feedback delay makes it difficult for the system to dynamically adjust subsequent ad selections based on previous user reactions within the same usage session.

[0005] The inherent flaws in the system ultimately lead to unstable advertising performance, making it difficult for advertisers to predict and control campaign results, with significant fluctuations in conversion and click-through rates. The platform also faces the risk of a decline in user experience, as users may experience reduced overall satisfaction due to frequent and inefficient ad interruptions. The system relies excessively on predicting the value of individual ads, lacking the ability to perceive and manage the dynamic changes in user attention within a continuous ad sequence.

[0006] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0007] This application discloses an advertising optimization method and system based on big data processing, aiming to solve the in-session advertising fatigue effect in existing internet advertising, and the resulting problems such as unstable advertising performance and decreased user experience. The technical solution of this application is as follows: In a first aspect, this application discloses an advertising optimization method based on big data processing, the method comprising: Obtain the user's historical ad interaction data in the current session, which includes the ad types of ads that have been displayed and the user's actions on the displayed ads; The type weight score is determined based on the type of the advertisement, and the behavior weight score is determined based on the behavior. The type weight score and behavior weight score of each displayed advertisement are added together to obtain the cumulative score of the current session. Determine whether the accumulated score exceeds a preset score threshold; If the accumulated score exceeds the preset threshold, the display of commercial ads will be paused in subsequent ad slots of the current session, and platform-recommended content will be displayed instead; otherwise, commercial ads will continue to be displayed.

[0008] Furthermore, this application also discloses an advertising optimization method based on big data processing, wherein obtaining historical advertising interaction data of users in the current session includes: capturing user operation behavior on displayed advertisements through a local collection module deployed on the user's device, the operation behavior including swiping to skip, clicking to close, and clicking the "not interested" button within a preset time window after the advertisement is displayed; The ad type and action are sent to the server, which then adds the type weight score to the action weight score to obtain the cumulative score for the current session.

[0009] Based on the above, this application further proposes that determining the corresponding type weight score according to the type of advertisement includes: determining a basic type score based on the type of the displayed advertisement; and obtaining the playback duration of the displayed advertisement and determining whether the playback duration exceeds a preset baseline duration; if it does not exceed the baseline duration, the basic type score is used as the type weight score; if it exceeds the baseline duration, the difference between the playback duration and the baseline duration is calculated, and an additional weight score is determined based on the difference. The basic type score and the additional weight score are added together to obtain the type weight score; wherein, the additional weight score is used to characterize the additional fatigue experienced by the user due to watching an excessively long advertisement.

[0010] More specifically, in some implementation schemes, a corresponding behavior weight score is determined based on the operation behavior, including: monitoring the user's operation behavior within a preset time window after the advertisement is displayed; if the user performs a swipe to skip or click to close within the preset time window, a first behavior score is determined; if the user clicks the "not interested" button, a second behavior score is determined, which is greater than the first behavior score.

[0011] Preferably, determining the corresponding behavior weight score based on the operation behavior further includes: obtaining the operation behaviors of multiple consecutive displayed advertisements in the current session, and the execution time corresponding to each operation behavior; sequentially calculating the time difference between every two adjacent operation behaviors to obtain a time difference sequence; sequentially determining whether the later time difference in the time difference sequence is less than the previous time difference; if so, multiplying the first behavior score or the second behavior score corresponding to the later operation behavior corresponding to the later time difference by a preset multiple to obtain the corrected behavior weight score, and using the corrected behavior weight score to participate in the calculation of the cumulative score.

[0012] Based on the above, this application further proposes that if the cumulative score exceeds the preset threshold, after pausing the display of commercial advertisements in subsequent ad slots of the current session, the application further includes: obtaining the start time of pausing the display of commercial advertisements; starting from the start time, subtracting a preset score reduction amount from the cumulative score of the current session every preset time interval; determining in real time whether the cumulative score after subtracting the score reduction amount is lower than a preset recovery threshold; if it is lower than the recovery threshold, then resuming the display of commercial advertisements.

[0013] Furthermore, subtracting a preset score reduction from the current session's cumulative score further includes: During the pause of commercial advertising display, the number of complete views and interaction actions of users on the platform's recommended content are counted. The number of complete views is the number of times the user plays the platform's recommended content without performing a swipe to switch, and the number of interaction actions is the sum of the number of times the user clicks on details, posts comments, and likes. The number of complete views is multiplied by a first coefficient to obtain a viewing contribution score, and the number of interaction actions is multiplied by a second coefficient to obtain an interaction contribution score. The viewing contribution score and the interaction contribution score are added together to obtain a cumulative score. The cumulative score is first subtracted from the current session's cumulative score, and then a preset score reduction is subtracted every preset time interval.

[0014] In some preferred embodiments, at preset time intervals, the cumulative score of the current session is subtracted by a preset score reduction amount, including: querying the application running status recorded by the operating system through the application client to detect whether the application client is running in the foreground; when it is detected that the application client has switched to running in the background, start a background timer; if the background timer exceeds the preset session retention time, reset the cumulative score of the current session and end the paused display of commercial advertisements; if the application client resumes foreground operation within the session retention time, continue to subtract the preset score reduction amount at preset time intervals.

[0015] Based on this, this application also proposes to continue subtracting a preset score reduction every preset time interval, including: collecting the user's viewing time and number of interaction operations on the platform's recommended content, including clicking on details, posting comments, and liking; weighting the viewing time and the number of interaction operations to obtain the user's browsing depth index; determining whether the user's browsing depth index has reached a preset depth threshold, and if so, continuing to subtract the preset score reduction every preset time interval; if not, pausing the subtraction of the preset score reduction.

[0016] Secondly, this application also discloses an advertising optimization system based on big data processing, such as... Figure 3 As shown, the system includes: The acquisition module is used to acquire the user's historical advertising interaction data in the current session. This historical advertising interaction data includes the types of ads that have been displayed and the user's actions on the displayed ads. The calculation module is used to determine the corresponding type weight score based on the ad type, determine the corresponding behavior weight score based on the operation behavior, and accumulate the type weight score and behavior weight score of each displayed ad to obtain the cumulative score of the current session. The judgment module is used to determine whether the accumulated score exceeds a preset score threshold; The execution module is configured to pause the display of commercial advertisements in subsequent ad slots of the current session and display platform-recommended content if the accumulated score exceeds the preset threshold. Beneficial effects

[0017] This application discloses an advertising optimization method based on big data processing. It acquires historical advertising interaction data of users in the current session in real time, including the types of displayed ads and user actions, and calculates type weight scores and behavior weight scores accordingly, accumulating these to obtain a cumulative score for the current session. When this cumulative score exceeds a preset threshold, the system intelligently pauses the display of commercial ads in subsequent ad slots and instead displays platform-recommended content. This innovative solution effectively solves the problem of insufficient perception of in-session advertising fatigue effects in existing technologies and overcomes the limitation of traditional advertising optimization methods in dynamically capturing user attention decay. By assessing users' degree of ad fatigue in real time, this application avoids incorrectly allocating high-value ads to ad slots where users are already fatigued, thereby significantly improving the efficiency of ad resource allocation and the actual effectiveness of ad delivery. Simultaneously, by pausing commercial ads in a timely manner and providing platform-recommended content, it greatly improves the user experience, reduces user annoyance caused by frequent and inefficient ad interruptions, and thus increases overall user satisfaction with the platform. This method achieves stability and predictability of advertising effects, providing advertisers with more reliable delivery guarantees and fostering a healthier user ecosystem for the platform. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the steps of the advertising optimization method based on big data processing disclosed in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the steps of the method for determining the corresponding behavior weight score based on the operation behavior disclosed in an embodiment of the present invention. Figure 3 This is a schematic diagram of the advertising optimization system structure based on big data processing disclosed in an embodiment of the present invention. Detailed Implementation

[0019] The implementation details of the technical solution in this embodiment are described in detail below: Traditional internet advertising methods primarily focus on predicting the independent value of individual ad impressions, assuming each ad impression is a relatively independent event. These methods excel at assessing the match between an ad and a user at a specific moment, but their predictive accuracy drops significantly when multiple ads are displayed consecutively within a very short period. This leads to inefficient allocation of advertising resources, unstable advertising performance, and potentially degraded user experience. The system relies too heavily on predicting the value of individual ad impressions, lacking the ability to perceive and manage the dynamic changes in user attention throughout a continuous ad sequence.

[0020] In this regard, such as Figure 1 As shown, this application proposes an advertising optimization method based on big data processing, the method comprising: S101, Obtain the user's historical advertising interaction data in the current session, the historical advertising interaction data including the types of advertisements that have been displayed and the user's operation behavior on the displayed advertisements; S102, determine the corresponding type weight score according to the ad type, determine the corresponding behavior weight score according to the operation behavior, and add the type weight score and behavior weight score of each displayed ad to obtain the cumulative score of the current session. S103, determine whether the accumulated score exceeds a preset score threshold; S104, if the cumulative score exceeds the score threshold, then the display of commercial advertisements in subsequent ad slots of the current session is paused, and platform-recommended content is displayed instead; otherwise, the display of commercial advertisements continues.

[0021] This application acquires users' historical ad interaction data in real time and dynamically calculates a cumulative score based on ad type and user behavior, thereby effectively sensing user ad fatigue during a session. When the cumulative score reaches a preset threshold, the system pauses the display of commercial ads and instead recommends platform content, thereby alleviating user fatigue, improving user experience, and optimizing ad delivery strategies.

[0022] To better understand the technical solutions proposed in this application, it is necessary to explain some key terms involved. Current session refers to a continuous usage process of a user within a specific application or platform, such as a user opening a short video application and continuously browsing content until exiting. Historical ad interaction data refers to all interaction information between the user and displayed commercial ads during the current session, including the category attributes of the ads and the user's actions towards these ads. Ad type can refer to the industry category to which the ad belongs (e.g., games, e-commerce, education, etc.), content theme (e.g., beauty, technology, food, etc.), or ad format (e.g., image ads, video ads, interactive ads, etc.). User behavior refers to the user's reaction after an ad is displayed, such as swiping to skip, clicking to close, clicking "not interested," clicking the ad details page, or viewing time. Type weight score is a score that quantifies the impact of ad type on user fatigue; different types of ads may cause different degrees of user fatigue. Behavioral weight score is a score that quantifies the impact of user's operational behavior on user fatigue; for example, actively closing an ad may reflect fatigue more strongly than simply swiping to skip it.

[0023] The cumulative score is the sum of the type weight score and behavior weight score of all ads displayed in the current session, and is used to comprehensively reflect the user's degree of ad fatigue in the current session.

[0024] The preset threshold is a pre-defined score limit. When the accumulated score exceeds this threshold, the system determines that the user has reached a high level of advertising fatigue.

[0025] Platform-recommended content refers to non-commercial content, such as short videos, live streams, and articles that users may be interested in. This content aims to improve user experience rather than generate direct profit.

[0026] The method proposed in this application can be implemented in various computing environments, such as mobile application clients, server-side systems, or cloud systems. Specifically, the local data acquisition module on the user's device can be responsible for capturing the user's operational behavior and sending the relevant data to the server for processing and calculation. The server is responsible for receiving data, performing weighted score calculation, cumulative score determination, and adjusting the advertising display strategy.

[0027] The core of the advertising optimization method based on big data processing proposed in this application lies in the real-time perception and dynamic adjustment of user advertising fatigue during a session.

[0028] In practical implementation, the first step is to acquire historical ad interaction data from users during the current session. This data forms the basis for assessing user ad fatigue. For example, a data collection module can be deployed on the user's device to monitor user interaction with ads in real time while browsing content. When a commercial ad is displayed to a user, the system records the ad type; for example, the ad might be a game promotion. Simultaneously, the system also records the user's actions related to the ad; for instance, the user might skip it halfway through playback or click the close button. This data is collected and stored as the basis for subsequent calculations.

[0029] Next, based on the acquired ad type and user behavior, the system determines the corresponding type weight score and behavior weight score. For example, for game promotion ads, a basic type weight score, such as 5 points, can be preset. This is because some types of ads may be more likely to cause user fatigue. For the user's swipe-to-skip action, a behavior weight score, such as 3 points, can be set; while for the click-to-close action, a higher behavior weight score, such as 8 points, can be set, because clicking to close usually indicates a higher level of user annoyance with the ad. These weight scores can be dynamically adjusted and optimized based on historical data analysis, expert experience, or machine learning models. After acquiring the type weight score and behavior weight score for each displayed ad, the system accumulates these scores to obtain the cumulative score for the current session. For example, if a user watches two ads, the first ad has a type weight score of 5 points and a behavior weight score of 3 points; the second ad has a type weight score of 6 points and a behavior weight score of 8 points, then the cumulative score for the current session is 5 + 3 + 6 + 8 = 22 points.

[0030] The system then determines whether the accumulated score exceeds a preset threshold. This preset threshold is a key indicator for measuring user advertising fatigue. For example, the preset threshold can be set to 20 points. If the accumulated score reaches or exceeds 20 points, it indicates that the user may have developed a high degree of fatigue with commercial advertisements.

[0031] Finally, the corresponding ad display strategy is executed based on the judgment result. If the accumulated score exceeds the score threshold, commercial ads will be paused in subsequent ad slots of the current session, and platform-recommended content will be displayed instead. For example, if the accumulated score is 22 points, exceeding the preset threshold of 20 points, then in subsequent ad slots, the system will no longer display commercial ads, but will recommend short videos, articles, or other non-commercial content that the user may be interested in. This can effectively alleviate user ad fatigue and improve user experience. Conversely, if the accumulated score does not exceed the preset threshold, commercial ads will continue to be displayed. For example, if the accumulated score is 15 points, which does not reach the preset threshold, the system will consider that the user has not yet reached a fatigue state, and therefore will continue to display commercial ads in subsequent ad slots.

[0032] The advertising optimization method proposed in this application, based on big data processing, can intelligently adjust advertising display strategies by monitoring user advertising interaction data in real time and dynamically calculating advertising fatigue. The core innovation of this method lies in its ability to overcome the limitations of traditional advertising optimization methods that focus solely on the independent value of a single advertisement, instead taking into account the dynamic feedback of users to continuous advertisements throughout the entire session.

[0033] Compared to existing technologies, this application has the advantage of addressing the in-session ad fatigue effect that often occurs when users are exposed to multiple ads consecutively within a short period. This can lead to high-value ads being incorrectly assigned to subsequent ad slots where users are already experiencing ad fatigue, thus reducing ad effectiveness and user experience. This application, by introducing the concept of a cumulative score and dynamically calculating this score based on ad type and user behavior, can reflect the degree of ad fatigue in real time and quantitatively. When the cumulative score reaches a preset threshold, the system can promptly pause the display of commercial ads and instead provide platform-recommended content, effectively alleviating user fatigue and preventing user churn due to excessive interruption. This dynamic adjustment mechanism not only improves the accuracy and efficiency of ad delivery, ensuring that high-value ads are displayed when users are focused, but also significantly improves user experience, increasing user satisfaction and retention rates. In this way, this application achieves dual optimization of ad effectiveness and user experience, providing the internet advertising industry with a smarter and more user-friendly solution.

[0034] In some of the embodiments described above in this application, in order to accurately obtain users' real feedback on advertisements and thus assess users' fatigue with advertisements, it is necessary to obtain users' historical advertisement interaction data in the current session.

[0035] In response, this application further proposes the above-mentioned step S101, which involves obtaining the user's historical advertising interaction data in the current session, including: capturing the user's operational behavior towards displayed advertisements through a local collection module deployed on the user's device, wherein the operational behavior includes swiping to skip, clicking to close, and clicking the "not interested" button within a preset time window after the advertisement is displayed; sending the advertisement type and operational behavior to the server, whereby the server accumulates the type weight score and the behavior weight score to obtain the cumulative score for the current session.

[0036] Specifically, the local data collection module can be understood as a functional module integrated into the application client or operating system of the user's device (such as a smartphone, tablet, or personal computer). Its main responsibility is to monitor and record the user's interaction with displayed advertisements in real time while browsing content. This module is designed to accurately capture the user's actions within a preset time window after the advertisement is displayed. The preset time window is designed to focus on the user's immediate reaction to the advertisement; for example, the time window can be set to 3 seconds, 5 seconds, or 10 seconds after the advertisement is displayed, to ensure that the captured actions effectively reflect the user's direct feelings about the current advertisement.

[0037] Specifically, the aforementioned actions refer to specific behaviors performed by users within a preset time window after an advertisement is displayed, such as swiping to skip, clicking to close, and clicking the "not interested" button. Swiping to skip typically refers to users quickly skipping the current advertisement by swiping the screen; clicking to close refers to users actively clicking the close button on the advertisement interface to stop the advertisement from displaying; clicking the "not interested" button indicates that users explicitly express their disinterest in the current advertisement content. These actions are all considered direct manifestations of negative emotions or fatigue towards the advertisement from the user.

[0038] In practical applications, after capturing user actions, the local data collection module sends this action data, along with the ad types of the displayed ads, to the server. Upon receiving this data, the server assigns a type weight score to each ad type and a behavior weight score to each action based on preset rules. Subsequently, the server sums the type weight scores and behavior weight scores of each displayed ad to obtain the cumulative score for the current session. This process ensures the timeliness and accuracy of user interaction data, providing a reliable basis for subsequent ad optimization decisions.

[0039] This application's solution achieves real-time and accurate capture of users' historical advertising interaction data by deploying a local data collection module on the user's device. Specifically, the local data collection module can directly monitor user actions such as swiping to skip, clicking to close, and clicking "not interested" within a preset time window after an ad is displayed. These actions directly reflect the user's advertising fatigue or discomfort. By sending these detailed user actions along with the ad type to the server, the server can calculate more refined type weight scores and behavior weight scores based on this raw data, and then accumulate them to obtain a cumulative score that accurately reflects the user's current advertising fatigue state. This combination of distributed data collection and centralized weight calculation ensures the timeliness of data acquisition and the efficiency of processing, providing a solid data foundation for subsequent judgments on whether users have reached the advertising fatigue threshold.

[0040] Furthermore, this application proposes a more refined method for determining type weight scores, which more accurately assesses user fatigue by comprehensively considering both ad type and actual playback duration. This application further proposes the step in S102 above, which involves determining the corresponding type weight score based on the ad type, specifically including: Based on the type of the displayed advertisement, determine the basic type score; and obtain the playback duration of the displayed advertisement, and determine whether the playback duration exceeds the preset baseline duration. If the score does not exceed the limit, the basic type score will be used as the type weight score. If the duration exceeds the baseline duration, the difference between the playback duration and the baseline duration is calculated. An additional weight score is determined based on the difference. The basic type score is added to the additional weight score to obtain the type weight score. The additional weight score is used to characterize the extra fatigue experienced by the user due to watching the excessively long advertisement.

[0041] Specifically, the base type score is an initial score pre-set based on the inherent type of the advertisement (e.g., game ads, e-commerce ads, news ads, etc.). This score reflects the user acceptance or potential fatigue level of that type of advertisement under normal circumstances. For example, some highly intrusive ad types may have higher base type scores, while ad types with generally high user acceptance have lower base type scores. Playback duration refers to the actual length of time a user watches the displayed advertisement. The baseline duration can be understood as the generally accepted, acceptable duration for users to watch an advertisement, such as 15 seconds or 30 seconds, as recognized by the platform or industry. This baseline duration can be dynamically adjusted based on the ad format, user group characteristics, or historical data analysis. In practical applications, when the playback duration exceeds the baseline duration, the difference between the two needs to be calculated. This difference directly reflects the extent to which users watch excessively long advertisements. Based on this difference, an additional weight score can be determined. This additional weight score aims to quantify the additional negative experience or fatigue experienced by users due to watching advertisements exceeding the expected duration. The additional weight score is positively correlated with the difference; that is, the larger the difference, the higher the additional weight score. Finally, the basic type score and the additional weight score are added together to obtain a more comprehensive and accurate type weight score.

[0042] This application's solution addresses the limitations of solely relying on ad type to determine the type weight score by incorporating consideration of the displayed ad playback duration and dynamically adjusting the type weight score based on a comparison between the playback duration and a baseline duration. Since user fatigue with ads is not only related to ad type but also closely linked to actual viewing time, especially excessively long viewing times, this solution calculates the difference between the playback duration and the baseline duration and determines an additional weight score. This allows the type weight score to more accurately represent the additional fatigue experienced by users due to watching excessively long ads. Therefore, when calculating the cumulative score for the current session, it more realistically reflects the user's current ad acceptance and fatigue level, providing a more accurate basis for determining whether to pause the display of commercial ads.

[0043] In some preferred embodiments, a specific example is given below. Assume a platform's preset ad duration is 30 seconds. When a user watches an ad: Scenario 1: If the ad type is a game ad, its base type score is set to 5 points. If the user actually watches the game ad for 20 seconds, which does not exceed the baseline duration of 30 seconds, then the ad's type weight score is the base type score of 5 points.

[0044] Scenario 2: If the ad type is still a game ad, the base type score is also 5 points. However, if the user actually watches the game ad for 45 seconds, exceeding the base duration of 30 seconds, the difference between the playback duration and the base duration is 45 seconds - 30 seconds = 15 seconds. According to the preset rules, for example, an additional weighted score of 1 point is added for every 5 seconds exceeding the base duration, then the 15-second difference will correspond to an additional weighted score of 3 points. Therefore, the ad type weighted score will be calculated as the base type score of 5 points + the additional weighted score of 3 points = 8 points.

[0045] As can be seen from the above examples, when a user watches an extra-long ad, its type weight score will increase due to the extra playback time, which will cause the cumulative score of the current session to reach the preset threshold more quickly, prompting the system to pause the display of commercial ads earlier to alleviate the user's extra fatigue.

[0046] This application further proposes that the step of determining the corresponding behavior weight score based on the operation behavior in S102 above includes: monitoring the user's operation behavior within a preset time window after the advertisement is displayed; if the user performs a swipe to skip or click to close within the preset time window, then a first behavior score is determined; if the user clicks the "not interested" button, then a second behavior score is determined, wherein the second behavior score is greater than the first behavior score.

[0047] Specifically, the preset time window refers to a specific duration after the advertisement content begins to be displayed, such as 3 seconds, 5 seconds, or 10 seconds, with the aim of capturing the user's immediate reaction to the advertisement. Within this time window, the system continuously monitors the user's interactive behavior. "Slide to skip" refers to the user quickly switching to or closing the currently playing advertisement by swiping on the screen; "click to close" refers to the user stopping the advertisement by clicking the close button provided on the advertisement interface. When the user performs these two actions, it indicates that the user has a certain degree of dissatisfaction or disinterest in the current advertisement. At this time, a first behavior score is determined, which is used to quantify the user's advertisement fatigue caused by these actions. In practical applications, "clicking the uninterested button" refers to the user actively clicking the function button clearly marked on the advertisement interface, such as "uninterested" or "no longer recommending this type of advertisement." This action usually indicates that the user has a stronger and more explicit negative feedback on the advertisement content, and their aversion level is higher than that of simply sliding to skip or clicking to close. Therefore, when a user clicks the "not interested" button, a second line score is determined. This second line score is set to be greater than the first line score to more accurately reflect the user's deeper level of ad fatigue or discomfort.

[0048] This application's solution addresses the issue of inaccurate user fatigue assessment in basic solutions by monitoring and differentiating user actions within a preset time window after ad display and assigning different weights to these actions based on their nature. Specifically, swiping to skip and clicking to close are considered mild negative feedback and thus assigned a first-order score; while clicking the "not interested" button is considered severe negative feedback and assigned a higher second-order score. This differentiated weighting allows the system to more precisely capture users' true attitudes and fatigue levels towards ads. In this way, the system can more accurately quantify user discomfort caused by ads, providing more reliable data support for subsequent ad optimization decisions.

[0049] In some preferred embodiments, a specific example is given below. Assume the preset time window is set to 5 seconds after the advertisement is displayed. When a user is watching video content, the system inserts a commercial advertisement. Scenario 1: Within 3 seconds of the advertisement being displayed, the user quickly skips the advertisement by swiping upwards on the screen. At this time, the system detects the swipe-to-skip operation and determines a first-row score based on preset rules, for example, 5 points. Scenario 2: Within 2 seconds of the advertisement being displayed, the user clicks the close button in the upper right corner of the advertisement interface. At this time, the system detects the click-to-close operation and similarly determines a first-row score, for example, 5 points.

[0050] Scenario 3: Within 4 seconds of an ad being displayed, a user clicks the "Not Interested" button provided at the bottom of the ad interface. The system detects this click and, based on preset rules, assigns a second-order score, for example, 15 points. As this example demonstrates, clicking the "Not Interested" button reflects a stronger negative emotion from the user, therefore its second-order score is significantly higher than the first-order score for swiping to skip or clicking to close. These weighted scores are then added to the current session's cumulative score, allowing the system to more sensitively detect user fatigue caused by advertising and decide whether to pause subsequent commercial ad displays accordingly.

[0051] This application further proposes a scheme to optimize the determination of behavior weight scores. By considering the time difference between consecutive operational behaviors, the negative emotions of users can be assessed more precisely, and the behavior weight scores can be adjusted accordingly to more accurately reflect the user's true feedback to the advertisement.

[0052] In S102 above, the corresponding behavior weight score is determined based on the operation behavior, such as... Figure 2 As shown, it also includes: S2001, obtain the operation behavior of multiple consecutive displayed advertisements in the current session, and the execution time of each operation behavior; S2002, calculate the time difference between every two adjacent operations in sequence to obtain the time difference sequence; S2003, sequentially determine whether the subsequent time difference in the time difference sequence is less than the previous time difference; S2004, if so, then the first or second action score corresponding to the next operation action corresponding to the next time difference is multiplied by a preset multiple to obtain the corrected action weight score, and the corrected action weight score is used in the calculation of the cumulative score.

[0053] Specifically, in this embodiment, the actions of displaying multiple consecutive advertisements refer to a series of operations performed by the user in the current session on different advertisements or different displays of the same advertisement, such as swiping to skip, clicking to close, or clicking the "not interested" button. The execution time refers to the specific point in time when the user performs each action, and its precision can be set according to actual needs, such as to milliseconds or seconds. By recording these execution times, basic data can be provided for subsequent time difference calculations.

[0054] The time difference sequence is obtained by subtracting the execution times of every two adjacent operations. This sequence reflects the speed at which the user performs consecutive operations. For example, if the user performs three operations consecutively in a short period of time, with execution times of T1, T2, and T3 respectively, the first time difference Δt1 = T2 - T1 and the second time difference Δt2 = T3 - T2 can be calculated.

[0055] In practical applications, determining whether each subsequent time difference in the time difference sequence is less than the previous time difference is a key condition for judging the increasing negative emotions of the user. If the user performs subsequent operations faster than before, i.e., the subsequent time difference is less than the previous time difference, it indicates that the user's impatience or fatigue is increasing.

[0056] In a preferred implementation, the preset multiplier is a coefficient greater than 1, used to amplify the behavior weight score. For example, it can be set to 1.2, 1.5, 2, etc., and the specific value can be adjusted according to the actual application scenario, user feedback data, and advertising optimization goals. By multiplying the first behavior score or the second behavior score by this preset multiplier, a corrected behavior weight score can be obtained. This corrected score can more strongly express the user's current negative emotions, and it is used in the calculation of the cumulative score, thus more accurately reflecting the user's true feedback to the advertisement.

[0057] This application's solution addresses the problem of traditional methods failing to effectively capture rapidly accumulating negative emotions in users by incorporating consideration of the time difference between consecutive actions. Specifically, after the system acquires a user's consecutive actions and their execution times, it can calculate the time difference between adjacent actions. By comparing these time differences, especially determining whether a later time difference is less than a previous one, the system can identify a trend of increasing speed at which the user executes negative actions. This acceleration trend directly reflects the user's growing impatience or fatigue. Once this trend is detected, the system multiplies the weight of subsequent actions by a preset factor, thereby amplifying the contribution of that action to the cumulative score. It is precisely because of this dynamic weight adjustment mechanism that when a user executes negative actions rapidly and consecutively within a short period, the cumulative score can reach the preset threshold more quickly, thus accelerating the decision to pause commercial advertisements.

[0058] In some preferred embodiments, a specific example is given below. Suppose a user is browsing a short video application and encounters multiple commercial advertisements in succession. First, the user watches the first advertisement A, and performs a swipe to skip it 3 seconds after the advertisement is displayed, at time T1.

[0059] Immediately after, the second advertisement B appears, and the user swipes to skip it 1 second after it is displayed, at time T2. Then, the third advertisement C appears, and the user clicks the "not interested" button 0.5 seconds after it is displayed, at time T3. According to the scheme of this application, the system will calculate the time difference between adjacent actions sequentially: the first time difference Δt1 = T2 - T1 = 1 second; the second time difference Δt2 = T3 - T2 = 0.5 seconds. Next, the system determines whether the subsequent time difference Δt2 is less than the previous time difference Δt1. In this example, 0.5 seconds is less than 1 second, so the result is yes. Therefore, the system will adjust the score of the second action corresponding to the next action (i.e., the user clicking the "not interested" button on advertisement C) corresponding to the second time difference Δt2. Assuming the preset multiplier is 1.5, and the score of the second action of clicking the "not interested" button is 10 points, the adjusted action weight score will become 10 * 1.5 = 15 points.

[0060] In this way, the contribution of negative user actions towards ad C to the cumulative score is significantly amplified, allowing the cumulative score for the current session to accumulate more quickly. For example, if the swipe-skip actions for ads A and B each contributed 5 points, the total score before the correction would be 5 + 5 + 10 = 20 points. After the correction, the total score becomes 5 + 5 + 15 = 25 points. This accelerated accumulation makes it easier for the cumulative score to exceed the preset threshold, thus triggering the mechanism to pause commercial ads earlier and instead display platform-recommended content, effectively alleviating ad fatigue caused by continuous and rapid negative actions.

[0061] This application further proposes a scheme that can dynamically adjust and restore the display of commercial advertisements after the aforementioned pause in displaying commercial advertisements. If the cumulative score exceeds the score threshold, after pausing the display of commercial advertisements in subsequent advertisement slots of the current session, the method further includes: obtaining the start time of pausing the display of commercial advertisements; starting from the start time, subtracting a preset score reduction amount from the cumulative score of the current session every preset time interval; determining in real time whether the cumulative score after subtracting the score reduction amount is lower than a preset recovery threshold; if it is lower than the recovery threshold, restoring the display of commercial advertisements.

[0062] Specifically, obtaining the start time of pausing the display of commercial advertisements refers to the precise point in time when the system records the cessation of advertisement display. This start time serves as the benchmark for subsequent timing and score decay. The preset time interval can be understood as the periodic cycle during which the system periodically performs score reduction operations, for example, it can be set every one minute or every five minutes. The preset score reduction refers to the fixed amount deducted from the current session's accumulated score in each score decay operation, its purpose being to simulate the natural recovery process of user fatigue. Real-time judgment of whether the accumulated score after deducting the score reduction is lower than the preset recovery threshold means that the system continuously monitors changes in the accumulated score; once it falls below the preset recovery threshold, it is considered that the user's acceptance of the commercial advertisement has recovered to a level where it can be displayed. The recovery threshold is usually lower than the preset threshold that triggers the pause, to ensure that the user has sufficient buffer time.

[0063] This application's solution effectively solves the problem of commercial advertisements failing to automatically resume once paused in the basic solution by introducing a dynamic decay mechanism for accumulated scores. Specifically, when a commercial advertisement is paused, the system records the start time of the pause and, from that time, periodically subtracts a preset score at preset time intervals. This process simulates the natural law that user fatigue with advertisements gradually decreases over time when there is no commercial advertisement interference. By continuously monitoring the accumulated score after decay and comparing it with a preset recovery threshold, once the accumulated score falls below the recovery threshold, it indicates that the user's advertising fatigue has been sufficiently alleviated, and the system can safely resume the display of commercial advertisements.

[0064] In some preferred embodiments, assuming a preset threshold of 100 points, a preset recovery threshold of 50 points, a preset time interval of 5 minutes, and a preset score reduction of 10 points, when a user's accumulated score in the current session reaches 120 points due to viewing or interacting with commercials, the system determines that it exceeds the preset threshold of 100 points, immediately pauses the display of commercials, and records the start time of the pause. From this start time, the system deducts 10 points from the accumulated score every 5 minutes. For example, after a 5-minute pause, the accumulated score becomes 110 points; after another 5 minutes, it becomes 100 points; and so on. When the accumulated score drops to 40 points after multiple reductions, the system determines that it has fallen below the preset recovery threshold of 50 points, at which point the display of commercials resumes. In this way, the system can intelligently determine the timing of resuming commercials based on the natural recovery of user fatigue, avoiding permanent pauses in advertising.

[0065] This application further proposes a step of subtracting a preset score reduction from the cumulative score of the current session, including: during the pause of displaying commercial advertisements, counting the number of complete views and interaction operations performed by the user on the platform's recommended content, wherein the number of complete views is the number of times the user plays the platform's recommended content without performing a swipe to switch, and the number of interaction operations is the sum of the number of times the user clicks on details, posts comments, and likes; multiplying the number of complete views by a first coefficient to obtain a viewing contribution score, and multiplying the number of interaction operations by a second coefficient to obtain an interaction contribution score; adding the viewing contribution score and the interaction contribution score to obtain a cumulative score; first subtracting the cumulative score from the cumulative score of the current session, and then subtracting the preset score reduction every preset time interval.

[0066] This application addresses the aforementioned problem by introducing statistics on the number of times users fully view and interact with platform-recommended content during periods when commercial ads are paused. These statistics are then converted into cumulative scores to participate in the calculation of cumulative score reductions. Specifically, when users actively and fully view platform-recommended content or interact with it, it indicates a high level of interest in the current content and that their ad fatigue may be effectively alleviating. By quantifying these positive behaviors as viewing and interaction contribution scores and accumulating them into a cumulative score, the cumulative score can receive additional, real-time reductions based on user behavior, on top of the original periodic reductions. This mechanism more accurately reflects the user's actual recovery status from ad fatigue, avoiding the lag or inaccuracy inherent in relying solely on fixed-time score reductions.

[0067] As a specific implementation method, assuming a user's accumulated score has reached and exceeded a preset threshold, the system pauses the display of commercial advertisements and begins displaying platform-recommended content. During the pause, the system detects that the user has watched two platform-recommended content items in their entirety and liked one of them. At this point, assuming the first coefficient is set to 5 and the second coefficient to 10, the calculated viewing contribution score is: 2 (number of complete views) × 5 (first coefficient) = 10 points. The interaction contribution score is: 1 (number of likes, belonging to the number of interaction operations) × 10 (second coefficient) = 10 points. The total score is: 10 (viewing contribution score) + 10 (interaction contribution score) = 20 points. In the subsequent process of reducing the accumulated score, the current session's accumulated score will first be subtracted from this 20-point total score. For example, if the current accumulated score is 120 points, after calculating the total score, the accumulated score will immediately become 100 points (120 - 20). Afterward, the system will continue to deduct a preset score at pre-defined time intervals, such as every 30 seconds (e.g., 5 points), until the accumulated score falls below a preset recovery threshold. In this way, the user's active participation in the platform's recommended content is quantified in real time and reflected in the rapid decline of the accumulated score, allowing the recovery of commercial advertisements to occur earlier and more accurately, thereby improving the responsiveness of the advertising optimization method and user satisfaction.

[0068] This application further proposes the following steps for subtracting a preset score reduction from the current session's cumulative score at preset time intervals: querying the application client's operating system records of the application's running status to detect whether the application client is running in the foreground; when the application client is detected to be switching to the background, starting a background timer; if the background timer exceeds a preset session retention time, resetting the current session's cumulative score and ending the paused display of commercial advertisements; if the application client resumes foreground operation within the session retention time, continuing to subtract the preset score reduction at preset time intervals.

[0069] Specifically, application clients can obtain their real-time running status, such as whether they are active in the foreground or suspended in the background, by calling API interfaces provided by the operating system or querying system logs. Detecting whether the application client is running in the foreground aims to determine whether the user is actively interacting with the application. When the system detects that the application client has switched from the foreground to the background, it immediately starts a background timer to record the continuous running time of the application. The preset session retention time is a configurable parameter, and its value can be set according to the actual application scenario and user behavior habits, for example, it can be set to several minutes to several hours. Its purpose is to define the boundary between the user temporarily leaving the application and completely disconnecting from it. If the time recorded by the background timer exceeds the preset session retention time, it indicates that the user may have been disconnected from the current application session for a long time. At this time, to ensure the accuracy of the advertising strategy and user experience, the accumulated score of the current session will be reset to the initial state, and the paused display of commercial advertisements will also end, preparing for a possible new session. Conversely, if the application client resumes its foreground operation before the preset session retention time has elapsed in the background, it indicates that the user has only temporarily left the application. The user's session context and ad fatigue status should still be retained. Therefore, the system will continue to deduct the preset score at preset time intervals to reflect the user's natural recovery of fatigue during the pause.

[0070] This application's solution addresses the inaccuracies that can arise from purely time-based score deduction by introducing a mechanism to detect the application client's running status and a background timer. When the application client switches to background operation, the system no longer blindly deducts accumulated points but instead starts a background timer. This mechanism can distinguish whether the user is temporarily leaving the application or has completely exited the session. Through a preset session retention time, the system can intelligently determine the validity of the user's session. If the user returns to the application within the session retention time, points continue to be deducted, ensuring the logical continuity of ad recovery; if the user does not return for an extended period, the accumulated points are reset and the pause state ends, avoiding unnecessary point deduction operations when the user has exited the session, thus making the ad recovery strategy more aligned with the user's actual usage scenario.

[0071] In some preferred embodiments, a specific example is given below. Suppose a user is browsing a content platform application and triggers the ad pause mechanism due to viewing too many ads. At this point, the application starts displaying platform-recommended content, and the accumulated score begins to decrease by 10 points every minute. After viewing a segment of platform-recommended content, the user switches the application to the background to reply to a message. At this time, the system detects that the application client has switched to background operation and immediately starts a background timer. If the preset session duration is 5 minutes, and the user switches the application back to the foreground after 3 minutes, the system detects that the application client has resumed foreground operation. Since 3 minutes does not exceed the 5-minute session duration, the system will continue to deduct the accumulated score according to the rule of deducting 10 points every minute, starting from the moment the user switches back to the foreground. However, if a user reopens the app after switching to the background for more than 5 minutes (e.g., 10 minutes), the system will detect that the background timer has exceeded the preset session duration. At this point, the accumulated score of the current session will be reset to the initial value, and the paused display of commercial ads will also end. This means that the system will treat this reopening of the app as a new session, thus avoiding the unreasonable continuation of the fatigue state of the old session into the new session.

[0072] This application further proposes to combine the user's browsing depth of the platform's recommended content with the preset score reduction at each preset time interval to more accurately assess the user's fatigue recovery.

[0073] Specifically, the aforementioned process of subtracting a preset score at preset time intervals includes: collecting the user's viewing time and number of interactive operations on the platform's recommended content, including clicking on details, posting comments, and liking; weighting the viewing time and number of interactive operations to obtain a user browsing depth index; determining whether the user browsing depth index has reached a preset depth threshold; if it has, continuing to subtract the preset score at preset time intervals; if it has not, pausing the subtraction of the preset score.

[0074] Specifically, after the application client resumes foreground operation, the system continuously collects the viewing time of users on the currently displayed platform-recommended content and the number of interactive actions they take. Viewing time refers to the actual length of time a user stays on the recommended content page and watches it. Interactive actions encompass positive feedback behaviors shown by users towards the recommended content, such as clicking on details to view more information, posting comments to express opinions or participate in discussions, and liking to show approval or appreciation. These data are considered key indicators for measuring users' interest and engagement with the platform's recommended content.

[0075] The collected viewing time and number of interactions are weighted and calculated to comprehensively assess the depth of user engagement with the platform's recommended content. For example, different weighting coefficients can be assigned to viewing time and number of interactions. By multiplying viewing time by its corresponding weight and interaction number by its corresponding weight, and then adding the two together, a comprehensive user engagement depth index is obtained. This index aims to quantify the user's focus and participation when browsing the platform's recommended content; a higher value generally indicates greater user interest and deeper engagement with the content.

[0076] In practical applications, the system determines whether the calculated user browsing depth index reaches a preset depth threshold. This depth threshold is a pre-set benchmark value used to define whether the user's browsing depth and activity level with the platform's recommended content is sufficient. If the user's browsing depth index reaches or exceeds this threshold, it indicates that the user has been fully engaged with the platform's recommended content, and their ad fatigue may be effectively recovering. In this case, the system will continue to deduct points from the accumulated score at preset time intervals. Conversely, if the user's browsing depth index fails to reach the preset depth threshold, it may mean that the user's interest or engagement with the platform's recommended content is insufficient, and their ad fatigue recovery is not effective. In this case, the system will pause the deduction of preset points to avoid prematurely resuming commercial ad display due to misjudgment of user fatigue recovery.

[0077] This application's solution effectively addresses the limitation of judging user fatigue recovery solely based on whether the application client is running in the foreground by introducing a user browsing depth index as a criterion for further score reduction. When a user brings the application client from the background to the foreground, not all users will immediately or deeply browse the platform's recommended content. By collecting viewing time and interaction counts and calculating the user browsing depth index, this solution can more precisely identify the user's true interest and level of engagement with the platform's recommended content. Score reduction only continues when the user's browsing depth index reaches a preset depth threshold. This ensures that score deductions only occur when the user is truly engaged with the platform's recommended content and potentially alleviates ad fatigue, thus avoiding misjudgments of fatigue recovery based on superficial foreground activity.

[0078] Secondly, this application also discloses an advertising optimization system based on big data processing, such as... Figure 3 As shown, the system includes: The acquisition module 301 is used to acquire the user's historical advertising interaction data in the current session. The historical advertising interaction data includes the advertising types of the displayed advertisements and the user's operation behavior on the displayed advertisements. The calculation module 302 is used to determine the corresponding type weight score according to the advertisement type, determine the corresponding behavior weight score according to the operation behavior, and accumulate the type weight score and behavior weight score of each displayed advertisement to obtain the cumulative score of the current session. The judgment module 303 is used to determine whether the cumulative score exceeds a preset score threshold; The execution module 304 is configured to pause the display of commercial advertisements in subsequent ad slots of the current session and display platform-recommended content if the cumulative score exceeds the score threshold.

[0079] The advertising optimization system based on big data processing proposed in this application aims to solve the problem of user fatigue during traditional advertising sessions. By systematically integrating acquisition, calculation, judgment, and execution modules, the system can dynamically perceive the user's fatigue level with continuous advertising in real time and intelligently adjust the advertising display strategy accordingly. Specifically, the acquisition module is responsible for collecting user interaction data with advertisements, the calculation module performs quantitative analysis on this data to derive a cumulative score of user fatigue, and the judgment module guides the execution module to decide whether to continue displaying commercial advertisements or switch to providing platform-recommended content based on the comparison result of this score with a preset threshold. This collaborative working mechanism ensures a balance between the accuracy of advertising and user experience, effectively avoiding user churn and decreased advertising effectiveness caused by excessive advertising interruption.

[0080] The core innovation of the big data-based advertising optimization system proposed in this application lies in its modular design, which enables real-time perception and dynamic management of user ad fatigue within a session. Compared with existing technologies, this system breaks through the limitations of traditional advertising optimization methods that only focus on the independent value of a single ad, instead taking into account the dynamic feedback of users to continuous ads throughout the session. Existing systems often struggle to capture the ad fatigue effect that occurs when users are exposed to multiple ads consecutively in a short period of time. This can lead to high-value ads being incorrectly assigned to subsequent ad slots where users are already experiencing fatigue, thereby reducing ad effectiveness and user experience. This system, through the collaborative work of acquisition, calculation, judgment, and execution modules, can reflect the user's ad fatigue level in real time and quantitatively. When the accumulated score reaches a preset threshold, the system can promptly pause the display of commercial ads and instead provide platform-recommended content, effectively alleviating user fatigue and preventing user churn due to excessive interruption. This dynamic adjustment mechanism not only improves the accuracy and efficiency of ad delivery, ensuring that high-value ads are displayed when users' attention is focused, but also significantly improves the user experience, increasing user satisfaction and retention rates. In this way, the system achieves dual optimization of advertising effectiveness and user experience, providing the internet advertising industry with a smarter and more user-friendly solution.

[0081] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. An advertising optimization method based on big data processing, characterized in that, The method includes: Obtain the user's historical advertising interaction data in the current session, including the types of ads that have been displayed and the user's actions on the displayed ads; The type weight score is determined according to the type of the advertisement, and the behavior weight score is determined according to the operation behavior. The type weight score and behavior weight score of each displayed advertisement are added together to obtain the cumulative score of the current session. Determine whether the accumulated score exceeds a preset score threshold; If the accumulated score exceeds the score threshold, the display of commercial advertisements will be paused in subsequent ad slots of the current session, and platform-recommended content will be displayed instead; otherwise, commercial advertisements will continue to be displayed.

2. The advertising optimization method based on big data processing according to claim 1, characterized in that, The step of obtaining the user's historical advertising interaction data in the current session includes: By using a local data collection module deployed on the user's device, the user's actions on the displayed advertisements are captured. These actions include swiping to skip, clicking to close, and clicking the "not interested" button within a preset time window after the advertisement is displayed. The advertisement type and operation behavior are sent to the server, which then adds the type weight score and the behavior weight score to obtain the cumulative score for the current session.

3. The advertising optimization method based on big data processing according to claim 1, characterized in that, The step of determining the corresponding type weight score based on the advertisement type also includes: Based on the type of the displayed advertisement, determine the basic type score; and obtain the playback duration of the displayed advertisement, and determine whether the playback duration exceeds the preset baseline duration. If the score does not exceed the limit, the basic type score will be used as the type weight score. If the duration exceeds the baseline duration, the difference between the playback duration and the baseline duration is calculated. An additional weight score is determined based on the difference. The basic type score is added to the additional weight score to obtain the type weight score. The additional weight score is used to characterize the extra fatigue experienced by the user due to watching the excessively long advertisement.

4. The advertising optimization method based on big data processing according to claim 1, characterized in that, The step of determining the corresponding behavior weight score based on the operation behavior includes: Monitor user behavior within a preset time window after an ad is displayed; If the user performs a swipe to skip or click to close within the preset time window, the first row is determined to be worth a score. If the user clicks the "Not interested" button, the second row is determined to have a score, which is greater than the score of the first row.

5. The advertising optimization method based on big data processing according to claim 4, characterized in that, The step of determining the corresponding behavior weight score based on the operation behavior also includes: Get the actions of multiple consecutive displayed ads in the current session, and the execution time of each action; Calculate the time difference between every two adjacent operations sequentially to obtain a time difference sequence; In the time difference sequence, determine whether each subsequent time difference is less than the previous time difference. If so, the first or second action score corresponding to the next operation action corresponding to the next time difference is multiplied by a preset multiple to obtain the corrected action weight score, and the corrected action weight score is used in the calculation of the cumulative score.

6. The advertising optimization method based on big data processing according to claim 1, characterized in that, The step of pausing the display of commercial advertisements in subsequent ad slots of the current session if the cumulative score exceeds the score threshold also includes: Obtain the start time of pausing the display of commercial advertisements; starting from the start time, every preset time interval, subtract a preset score reduction from the current session's cumulative score; The system determines in real time whether the cumulative score after deducting the score is lower than a preset recovery threshold; if it is lower than the recovery threshold, the commercial advertisement display is restored.

7. The advertising optimization method based on big data processing according to claim 6, characterized in that, The step of subtracting a preset score reduction from the current session's cumulative score further includes: During the period when commercial advertisements are suspended, the number of times users fully watch the platform's recommended content and the number of times they interact with it are counted. The number of times users fully watch the platform's recommended content without swiping to switch is counted. The number of times users interact with it is the sum of the number of times they click on details, post comments, and likes. Multiply the number of complete views by the first coefficient to get the viewing contribution score; multiply the number of interactive actions by the second coefficient to get the interaction contribution score; add the viewing contribution score and the interaction contribution score to get the total score. The cumulative score of the current session is first subtracted from the superimposed score, and then a preset score reduction is subtracted every preset time interval.

8. The advertising optimization method based on big data processing according to claim 6, characterized in that, The step of subtracting a preset score reduction from the current session's cumulative score after each preset time interval includes: The application client queries the operating system's recorded application running status to detect whether the application client is running in the foreground; when the application client is detected to switch to the background, a background timer is started. If the background timer exceeds the preset session duration, the accumulated score for the current session will be reset, and the paused display of commercial advertisements will be terminated. If the application client resumes foreground operation within the session persistence time, the preset score will continue to be deducted every preset time interval.

9. The advertising optimization method based on big data processing according to claim 8, characterized in that, The step of subtracting a preset score reduction every preset time interval includes: The system collects data on the duration of user viewing and the number of interactions with content recommended by the platform; these interactions include clicking on details, posting comments, and liking. The user browsing depth index is obtained by weighting the viewing time and the number of interactive operations. Determine whether the user's browsing depth index has reached a preset depth threshold. If it has, continue to deduct a preset score every preset time interval. If it has not reached the threshold, stop deducting the preset score.

10. An advertising optimization system based on big data processing, characterized in that, The system includes: The acquisition module is used to acquire the user's historical advertising interaction data in the current session. The historical advertising interaction data includes the types of advertisements that have been displayed and the user's operation behavior on the displayed advertisements. The calculation module is used to determine the corresponding type weight score according to the ad type, determine the corresponding behavior weight score according to the operation behavior, and accumulate the type weight score and behavior weight score of each displayed ad to obtain the cumulative score of the current session. The judgment module is used to determine whether the cumulative score exceeds a preset score threshold; The execution module is configured to pause the display of commercial advertisements in subsequent ad slots of the current session and display platform-recommended content if the accumulated score exceeds the score threshold.

Images