Internet consumption intelligent monitoring system and method based on big data

Abstract

The invention discloses an internet consumption intelligent monitoring system and method based on big data, and relates to the technical field of big data user behavior monitoring and intelligent recommendation systems, and the system comprises a behavior statistics module, a significance drift judgment module, and an update control module. According to the system disclosed by the invention, 'user interest fluctuation 'and'systematic behavior offset' are effectively distinguished, a more stable user behavior monitoring mechanism with a lower misjudgment rate is constructed, continuous dynamic description of a user behavior state is realized, a more accurate and more stable input basis is provided for subsequent offset identification, and on the premise that the robustness of the system is kept, the user behavior monitoring accuracy is improved. Important behavior changes possibly needing model response are sensitively recognized, the method is particularly suitable for user behavior collective migration scenes caused by rapid evolution of crowds or emergencies, and high-efficiency, low-cost and differentiated trigger type intelligent updating of recommendation models is achieved; and the adaptive capacity and the personalized accuracy of the recommendation system to the user interest migration are obviously improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology for internet consumption based on big data, specifically to an intelligent monitoring system and method for internet consumption based on big data. Background Technology

[0002] With the widespread adoption of internet applications, big data-based intelligent analytics has been extensively applied in e-commerce platforms, social media, and advertising recommendations. Especially in the consumer sector, user behavior data (such as clicks, browsing, and purchases) has gradually become a core resource for platforms to optimize recommendation systems and predictive models. Early recommendation systems were primarily based on simple collaborative filtering and content recommendation algorithms. However, with the increasing volume of user behavior data, how to dynamically analyze real-time data streams and promptly capture changes in user preferences has become a crucial direction for technological development. Currently, many e-commerce platforms and social networks have begun to introduce real-time data stream processing technology, utilizing sliding window technology to monitor and analyze user behavior in real time, thereby updating recommended content and improving user experience. However, most existing technologies rely on static analysis-based behavior prediction, lacking adaptability and struggling to cope with rapidly changing consumption trends and short-term fluctuations.

[0003] While existing technologies have improved the accuracy of user behavior prediction and recommendation to some extent, several significant shortcomings remain. Most existing systems rely on accumulated historical data and static recommendation models for user behavior prediction, failing to adequately consider real-time changes in user behavior. In traditional sliding window models, while smoothing historical data can capture certain trends, it often fails to respond promptly to significant shifts occurring in the short term (such as promotional activities or sudden surges in popularity of trending products). This lag prevents recommendation systems from making effective adjustments in the early stages of changes in user interests, thus impacting the platform's marketing effectiveness and user experience. More seriously, existing behavior monitoring systems often rely on simple rule thresholds to determine user behavior drift. These rules lack dynamic adaptability to data fluctuations, easily misinterpreting normal fluctuations as drift, leading to frequent model updates, which in turn cause system performance fluctuations and increase unnecessary computational burden. Existing technologies lack an effective saliency detection mechanism in detecting user behavior drift. Many methods rely solely on simple mean comparisons and fixed thresholds to determine drift occurrences, without considering the complexity and dynamism of user behavior. With the increasing volume of data and the diversification of consumption trends, simple statistical analysis methods are insufficient to capture subtle changes in user needs and interests. Therefore, when dealing with complex behavioral changes, systems are prone to misjudgments and inefficient responses, especially within short periods of rapid shifts in user needs. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by this invention is that existing user behavior analysis and recommendation system methods lack a dynamic adaptive drift detection mechanism and how to accurately trigger intelligent updates of the recommendation model when user behavior deviates significantly.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an intelligent monitoring system for internet consumption based on big data, comprising a behavior statistics module, a significant drift judgment module, and an update control module; the behavior statistics module is used to perform behavioral indicator statistics and volatility assessment on the click data of internet platform users within a sliding time window; the significant drift judgment module is used to determine whether the current user behavior significantly deviates from the historical click trend using a statistical significance test method; the update control module is used to trigger an update to the recommendation or prediction model when the user behavior drifts significantly and conforms to the platform's strategy control rhythm; determining whether the current user behavior significantly deviates from the historical click trend includes... The average user clicks in the current time window are standardized and compared with the historical average clicks updated via sliding. Combined with the degree of volatility and sample size, a statistic suitable for small sample scenarios is constructed to calculate the significance p-value and assess the significance of user behavior changes. Significant user behavior drift that conforms to the platform's strategy control rhythm includes generating adaptive behavior change thresholds based on consumption data fluctuations and performing a two-factor drift judgment based on confidence levels. Drift is considered significant only when the change exceeds the volatility tolerance range and the confidence level meets the standard. A minimum time interval threshold is set to control the update frequency of the recommendation or prediction model, triggering an update only when behavior drift is significant and the cooldown interval is met.

[0007] As a preferred embodiment of the big data-based intelligent monitoring system for internet consumption described in this invention, the behavioral indicator statistics and volatility assessment include dividing the real-time collected user click behavior log stream into sliding windows of fixed duration, and within each time window, statistically analyzing all users' click behavior on products or content, measuring the overall user consumption activity by calculating the exponentially smoothed mean of historical click behavior within the window, and assessing the dispersion of click behavior by using the standard deviation; analyzing the dynamic trend of users' historical interests, and using an exponential weighting method to fuse the click mean of the current and previous windows to construct a smooth click baseline that reflects the inertia of consumption behavior.

[0008] As a preferred embodiment of the big data-based intelligent monitoring system for internet consumption described in this invention, the step of determining whether the current user behavior significantly deviates from the historical click trend includes: standardizing and comparing the average user clicks in the current time window with the historical average clicks updated by sliding; combining the fluctuation of the current click behavior and the sample size to construct a statistic suitable for small sample scenarios to measure the significance of changes in user behavior; and obtaining a confidence p-value by finding a standard t-distribution to determine whether the current user click behavior belongs to a systematic shift.

[0009] As a preferred embodiment of the big data-based intelligent monitoring system for internet consumption described in this invention, the determination of whether the current user behavior significantly deviates from the historical click trend further includes classifying the deviation between the confidence p-value of the user click behavior change and the exponentially smoothed mean of the historical click behavior.

[0010] As a preferred embodiment of the big data-based intelligent monitoring system for internet consumption described in this invention, the significant user behavior drift that conforms to the platform's strategy control rhythm includes adaptively generating a platform behavior change threshold based on the fluctuation level of current consumption behavior data, and constructing a two-factor drift identification system in conjunction with a confidence judgment mechanism; the platform only determines behavior drift when the change in user click behavior exceeds the current fluctuation tolerance range and the confidence level is higher than the set standard, thereby triggering a potential model response operation.

[0011] As a preferred embodiment of the big data-based intelligent monitoring system for internet consumption described in this invention, the significant drift of user behavior that conforms to the platform's strategy control rhythm further includes setting a minimum time interval threshold to control the update frequency of the recommendation or prediction model; the update process of the recommendation or prediction model is only triggered when the user behavior is determined to be drifting again within a continuous window, and the time window since the last update of the recommendation or prediction model exceeds the set cooling interval.

[0012] As a preferred embodiment of the big data-based intelligent monitoring system for internet consumption described in this invention, the triggering of updating the recommendation or prediction model includes executing an update process after confirming that the conditions for updating the recommendation or prediction model are met, while recording the current time window number as the latest update point, and the platform restarting the data collection and behavior analysis task for the next sliding window cycle.

[0013] Another objective of this invention is to provide a big data-based intelligent monitoring method for internet consumption. This method can trigger updates to recommendation or prediction models when user behavior drifts significantly and conforms to the platform's policy control rhythm. This solves the problems of current recommendation system model update techniques, such as single update strategies, frequent updates leading to wasted system resources, and model overfitting.

[0014] As a preferred embodiment of the big data-based intelligent monitoring method for internet consumption described in this invention, the method includes: performing behavioral indicator statistics and volatility assessment on the click data of internet platform users within a sliding time window; using a statistical significance test method to determine whether the current user behavior significantly deviates from the historical click trend; and triggering an update to the recommendation or prediction model when the user behavior drifts significantly and conforms to the platform's strategy control rhythm.

[0015] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program as steps to implement a big data-based intelligent monitoring method for internet consumption.

[0016] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of a big data-based intelligent monitoring method for internet consumption.

[0017] The beneficial effects of this invention are as follows: The internet consumption intelligent monitoring system based on big data provided by this invention achieves continuous and dynamic characterization of user behavior by statistically analyzing behavioral indicators and assessing volatility of user click data on internet platforms within a sliding time window. It uses statistical significance testing to determine whether current user behavior significantly deviates from historical click trends, and can sensitively identify important behavioral changes that may require model response. It is suitable for scenarios involving rapid population evolution or collective migration of user behavior triggered by sudden events. Under conditions where user behavior drift is significant and conforms to the platform's strategy control rhythm, it triggers intelligent updates to recommendation or prediction models, achieving efficient, low-cost, and differentiated triggered intelligent updates to recommendation models. This invention achieves better results in terms of dynamic accuracy of user behavior monitoring, judgment accuracy of behavior drift identification, and intelligent strategy for recommendation model updates. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is an overall module diagram of an Internet consumer intelligence monitoring system based on big data, provided in Embodiment 1 of the present invention.

[0020] Figure 2 The above is an overall flowchart of an Internet consumption intelligent monitoring method based on big data, provided in Embodiment 2 of the present invention. Detailed Implementation

[0021] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0022] Example 1, referring to Figure 1 As an embodiment of the present invention, an Internet consumption intelligent monitoring system based on big data is provided, including a behavior statistics module 100, a significance drift judgment module 200, and an update control module 300.

[0023] The behavior statistics module 100 is used to perform behavioral indicator statistics and volatility assessment on the click data of Internet platform users within a sliding time window.

[0024] Furthermore, the behavioral indicator statistics and volatility assessment include dividing the real-time collected user click behavior log stream into sliding windows of fixed duration, and within each time window, statistically analyzing all users' click behavior on products or content. The overall user consumption activity is measured by calculating the exponentially smoothed mean of historical click behavior within the window, and the dispersion of click behavior is assessed by the standard deviation. The dynamic trend of users' historical interests is analyzed, and the click mean of the current and previous windows is merged using an exponential weighting method to construct a smooth click baseline that reflects the inertia of consumption behavior.

[0025] It should also be noted that a preferred scheme for statistical analysis and volatility assessment of user click data on internet platforms specifically includes dividing the real-time collected user click behavior data into fixed-length sliding windows. Each window contains a set number of click sample data. The arithmetic mean of all click data within each window is calculated as the click mean for the current window. Based on the sample variance, the standard deviation of click behavior in the current window is calculated, and the click mean is calculated, expressed as:

[0026]

[0027] in, This represents the number of clicks by the i-th user in window τ, where N is the number of samples in each window; These are the mean clicks and standard deviation of the current window's clicks, expressed as:

[0028]

[0029] in, The standard deviation of click behavior within window τ reflects the volatility of user click behavior within that window. Greater volatility indicates more unstable behavior. The historical click mean is expressed as:

[0030]

[0031] in, It is the exponentially smoothed mean of historical click behavior. This represents the historical average click value inherited from the previous time window, and α∈(0,1) is the smoothing coefficient.

[0032] It should also be noted that by setting a sliding time window and performing real-time slicing of the user click log stream, high-frequency sampling and structured organization of platform user behavior data are achieved. Within each window, the system not only calculates the average value of click behavior to dynamically assess the current consumption activity of the platform, but also calculates the standard deviation index to capture the dispersion of behavior and reflect the concentration and stability of user interests. In addition, by introducing an exponential weighting mechanism to integrate current and historical window data, a smooth click baseline is constructed, and user behavior inertia features are extracted from the time dimension. This constructs a behavior monitoring model that takes into account both short-term sensitivity and long-term trends, which is significantly better than the traditional static average statistical method. It has significant improvements in real-time performance, dynamic responsiveness, and adaptability to periodic fluctuations. The platform can achieve continuous dynamic characterization of user behavior status, providing more accurate, basic, and stable input for subsequent offset identification, thereby improving the response quality of the entire system.

[0033] The significance drift judgment module 200 is used to determine whether the current user behavior significantly deviates from the historical click trend using statistical significance test methods.

[0034] Furthermore, determining whether current user behavior significantly deviates from historical click trends involves standardizing and comparing the average user clicks in the current time window with the historical average clicks updated via sliding. This is combined with the degree of fluctuation in current click behavior and the sample size to construct a statistic suitable for small-sample scenarios, measuring the significance of changes in user behavior. A confidence p-value is obtained by finding a standard t-distribution to determine whether the current user click behavior belongs to a systematic shift.

[0035] It should also be noted that determining whether current user behavior significantly deviates from historical click trends also involves classifying the deviation between the confidence p-value of the change in user click behavior and the exponentially smoothed mean of historical click behavior.

[0036] It should also be noted that a preferred approach to determining whether current user behavior significantly deviates from historical click trends includes the following: simple numerical deviation does not necessarily represent a "behavioral change"; statistical inference methods must be used to verify its significance. A small-sample t-test is used to determine whether the mean of current window behavior differs statistically from historical behavior. The t-value is directly related to the sample mean, standard deviation, and sample size. The p-value quantifies the significance of the drift; the smaller the p-value, the higher the confidence that "current behavior deviates from history." A confidence threshold (e.g., 0.01) is typically set as the judgment standard. Since user behavior has a certain degree of volatility, frequent updates to the recommendation or prediction model should not be triggered due to small fluctuations. Therefore, a "dynamic threshold" is introduced in this step, adaptively adjusting the drift judgment threshold based on the fluctuation of the current window. Simultaneously, combined with the p-value judgment mechanism of the t-test, it is ensured that the system only truly recognizes a behavioral drift under the dual conditions of "sufficiently large offset + reliable significance," expressed as:

[0037]

[0038] Among them, T (τ) This is a statistic representing the mean clicks in the current window; the corresponding two-sided p-value is:

[0039]

[0040] Among them, P (τ) The p-value represents the significance of the current window's drift. A smaller p-value indicates that the current click behavior deviates more from the historical behavior. v represents the degrees of freedom, v = N-1. The symbol Γ(·) represents the Gamma function, and t represents the integral variable.

[0041] When P (τ) <0.01 and |T (τ) When |≥2, the drift probability is high and the drift is significant. For example, if users suddenly burst into clicking or purchasing behavior on a certain type of product (such as 3C digital products), it indicates that a significant interest migration has occurred; when 0.01≤P (τ) <0.05 and 1≤|T (τ) When |<2, it is considered that there is a moderate degree of deviation in user behavior, but not enough to completely confirm a behavior shift. This area is defined as the drift transition interval, which is common in short-term event-driven situations such as the initial stage of live-streaming e-commerce; when 0.05≤P (τ) And |T (τ) When |<1, it indicates that the fluctuation in the current user's click behavior may be caused by natural noise. The system determines that the behavior is in a stable state and it is not recommended to adjust the recommendation model or parameters.

[0042] It should also be noted that current recommendation or prediction models include one of the common recommendation models, matrix factorization models (such as SVD), which reveal the implicit association between users and items by decomposing the user-item rating matrix into multiple latent factor matrices. In matrix factorization models, the focus of updates is usually on the latent factor matrices of users and items. By introducing new user behavior data (such as clicks, purchases, etc.), these factor matrices are incrementally updated, thereby improving recommendation accuracy. One common prediction model is the regression model (such as linear regression or decision tree regression), which predicts numerical outputs (such as predicting sales) based on input features. In this model, the focus of updates is usually on feature weights or model parameters. These parameters are adjusted by introducing new data to ensure that the model can reflect the latest trends and data changes. In practical applications, recommendation model updates usually focus on changes in user behavior data, while prediction models pay more attention to the relationship between input features and historical data, ensuring that the model's predictive performance keeps pace with the times through periodic or incremental updates.

[0043] It should also be noted that by constructing statistics suitable for small sample scenarios, a standardized difference analysis is performed on the mean of current click behavior and the historical baseline of sliding updates. Combining sample volatility and sample size factors, a t-distribution is used to look up the table to obtain the significance confidence p-value, further supporting hierarchical judgment. By comprehensively considering both the magnitude of click changes and statistical significance, the problem of misjudging short-term fluctuations in user behavior is solved, and the risk of missing key behavioral changes due to excessive conservatism is avoided. It effectively distinguishes between "fluctuations in user interest" and "systematic behavioral shifts," constructing a more stable user behavior monitoring mechanism with a lower misjudgment rate. The platform can sensitively identify important behavioral changes that may require a response from recommendation or prediction models while maintaining system robustness. It is especially suitable for scenarios where the user population is rapidly evolving or where sudden events trigger a collective migration of user behavior.

[0044] The update control module 300 is used to trigger an update to the recommendation or prediction model when user behavior drifts significantly and conforms to the platform's policy control rhythm.

[0045] Furthermore, significant user behavior drift that aligns with the platform's strategy control rhythm includes adaptively generating a behavior change threshold based on the volatility of current consumer behavior data, and constructing a two-factor drift identification system in conjunction with a confidence judgment mechanism. The platform only determines behavior drift when the magnitude of user click behavior changes exceeds the current volatility tolerance range and the confidence level is higher than the set standard, thereby triggering potential model response operations.

[0046] It should also be noted that a preferred solution for user behavior drift that is significant and conforms to the platform's policy control rhythm specifically includes adaptive drift threshold setting, expressed as:

[0047]

[0048] Where θ0 is the base offset tolerance value (e.g., 1.5), representing the default tolerance for change when there is no behavioral fluctuation, λ is the dynamic adjustment factor, and Θ (τ) The threshold for determining the offset of the current window; the drift trigger condition is expressed as:

[0049]

[0050] Among them, D (τ) This is the drift determination output for window τ, with a value of 1 indicating drift and 0 indicating no drift. δ is the confidence significance threshold (e.g., 0.01, representing a 99% confidence requirement). D (τ) =1 indicates a significant drift within window τ, meaning the platform detects a significant shift in the magnitude and stability of user click behavior. For example, if a brand's product experiences a sudden surge in clicks due to an external hot topic, and D... (τ) =0 indicates that although the current window has behavioral fluctuations, they are not enough to reach the system's reaction threshold. The system judges it as a non-drift state, and the recommendation logic remains unchanged.

[0051] It should also be noted that significant user behavior drift that conforms to the platform's policy control rhythm also includes setting a minimum time interval threshold to control the update frequency of the recommendation or prediction model; the update process of the recommendation or prediction model is only triggered when user behavior is determined to be drift again within a continuous window, and the time window since the last update of the recommendation or prediction model exceeds the set cooling interval.

[0052] It should also be noted that a preferred approach to controlling the update frequency of recommendation or prediction models includes, even in the event of significant drift, frequent updates would lead to training costs and performance instability if the recommendation or prediction model has just been updated. Therefore, an "update cooldown period" mechanism is added. The recommendation or prediction model is only allowed to be updated after the cooldown period has expired and if the drift is confirmed. Based on the recommendation or prediction model update decision control mechanism, the decision of whether the recommendation or prediction model can be updated is controlled. To avoid high-frequency false triggers, a cooldown time mechanism is set, and a recommendation or prediction model update decision function U is constructed. (τ) , is represented as:

[0053]

[0054] Where, τ last Δ is the window number of the last recommendation or prediction model update. min This is the minimum time window interval threshold.

[0055] If the system confirms that the current behavior drift is valid, and the interval since the last update has exceeded the minimum interval Δ. minIf this happens, the recommendation or prediction model update operation is triggered, and the latest update window is recorded, with the current window number τ recorded as the new τ. last After the update, the recommendation or prediction model enters a new monitoring cycle. The system starts capturing behavioral trends again from the sliding window. If the cooling condition is not met, even if the drift is established, the recommendation or prediction model will not be updated immediately to keep the current recommendation structure stable in order to avoid the problem of "false triggering of short-term explosive points".

[0056] It should also be noted that triggering an update to the recommendation or prediction model involves executing the update process after confirming that the conditions for updating the recommendation or prediction model are met, while recording the current time window number as the latest update point, and the platform restarting the data collection and behavior analysis tasks for the next sliding window cycle.

[0057] It should also be noted that, based on the significant drift determination, platform strategy conditions are further introduced as control thresholds. These include logical judgments such as whether the magnitude of behavioral change exceeds the current fluctuation tolerance range, whether the significance confidence level exceeds the set standard, and whether the minimum time cooling interval is met. Only when all these conditions are met will the recommendation or prediction model update process be executed, and the update window be recorded, realizing platform-level model lifecycle management. This step constructs a "strategy-controllable + behavior-driven" recommendation or prediction model update mechanism, effectively solving the system jitter or overfitting problems caused by frequent updates in traditional recommendation or prediction model update strategies. At the same time, by setting cooling intervals and multi-factor thresholds, the platform can reduce resource consumption and improve business controllability while ensuring the timeliness of recommendation or prediction model response. The platform can achieve efficient, low-cost, and differentiated trigger-based intelligent updates of recommendation models, improving the adaptability of the recommendation system to user interest migration and personalization accuracy.

[0058] Example 2, refer to Figure 2 As an embodiment of the present invention, a method for intelligent monitoring of internet consumption based on big data is provided, comprising:

[0059] S1: Perform behavioral metric statistics and volatility assessment on click data of internet platform users within a sliding time window.

[0060] S2: Use statistical significance testing to determine whether the current user behavior significantly deviates from the historical click trend.

[0061] S3: When user behavior drifts significantly and conforms to the platform's policy control rhythm, trigger an update to the recommendation or prediction model.

[0062] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0063] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0064] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0065] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A big data-based intelligent monitoring system for internet consumption, characterized in that: include: Behavior statistics module (100), significance drift judgment module (200), update control module (300); The behavior statistics module (100) is used to perform behavior metric statistics and volatility assessment on the click data of Internet platform users within a sliding time window; The significance drift judgment module (200) is used to determine whether the current user behavior significantly deviates from the historical click trend using statistical significance test methods; The update control module (300) is used to trigger an update to the recommendation or prediction model when user behavior drifts significantly and conforms to the platform's policy control rhythm. Determining whether current user behavior significantly deviates from historical click trends involves standardizing and comparing the average user clicks in the current time window with the average historical clicks updated via sliding, and combining the degree of fluctuation and sample size to construct a statistic suitable for small sample scenarios, calculate the significance p-value, and assess the significance of changes in user behavior. Significant user behavior drift that aligns with the platform's strategy control rhythm includes generating adaptive behavior change thresholds based on consumption data fluctuations and using a two-factor drift assessment based on confidence levels. Significant drift is determined only when the magnitude of the change exceeds the fluctuation tolerance range and the confidence level meets the standard. A minimum time interval threshold is set to control the update frequency of the recommendation or prediction model, triggering an update only when the behavior drift is significant and the cooldown interval is met.

2. The Internet consumer intelligent monitoring system based on big data as described in claim 1, characterized in that: The behavioral indicator statistics and volatility assessment include dividing the real-time collected user click behavior log stream into sliding windows of fixed duration, and within each time window, counting all users' click behavior on products or content, measuring the overall user consumption activity by calculating the exponentially smoothed mean of historical click behavior within the window, and assessing the dispersion of click behavior by using the standard deviation. By analyzing the dynamic trends of users' historical interests, an exponential weighting method is used to merge the average clicks of the current and previous windows to construct a smooth click baseline that reflects the inertia of consumer behavior.

3. The Internet consumer intelligent monitoring system based on big data as described in claim 2, characterized in that: The method of determining whether the current user behavior significantly deviates from the historical click trend includes standardizing and comparing the average user clicks in the current time window with the average historical clicks updated by sliding. Combining the degree of fluctuation of the current click behavior and the sample size, a statistic suitable for small sample scenarios is constructed to measure the significance of changes in user behavior. By finding the standard t-distribution to obtain the confidence p-value, it can be determined whether the current user's click behavior belongs to a systematic shift.

4. The Internet consumer intelligent monitoring system based on big data as described in claim 3, characterized in that: The determination of whether the current user behavior deviates significantly from the historical click trend also includes classifying the deviation between the confidence p-value of the change in user click behavior and the exponentially smoothed mean of historical click behavior.

5. The Internet consumer intelligent monitoring system based on big data as described in claim 4, characterized in that: The significant user behavior drift that conforms to the platform's strategy control rhythm includes adaptively generating a threshold for platform behavior changes based on the fluctuation of current consumption behavior data, and constructing a two-factor drift identification system in conjunction with a confidence judgment mechanism. The platform only determines a user's click behavior to be a drift when the change exceeds the current fluctuation tolerance range and the confidence level is higher than the set standard, thereby triggering a potential model response operation.

6. The Internet consumer intelligent monitoring system based on big data as described in claim 5, characterized in that: The significant drift in user behavior that conforms to the platform's policy control rhythm also includes setting a minimum time interval threshold to control the update frequency of recommendation or prediction models. The update process for the recommendation or prediction model is triggered only when the user's behavior is determined to be drift again within a continuous window, and the time window since the last update of the recommendation or prediction model exceeds the set cooling interval.

7. The Internet consumer intelligent monitoring system based on big data as described in claim 6, characterized in that: The triggering of the recommendation or prediction model update includes executing the update process after confirming that the update conditions for the recommendation or prediction model are met, while recording the current time window number as the latest update point, and the platform restarting the data collection and behavior analysis task for the next sliding window cycle.

8. A method for intelligent monitoring of internet consumption based on big data, characterized in that: This includes statistical analysis of behavioral metrics and volatility assessment of user click data on internet platforms within a sliding time window; Statistical significance testing is used to determine whether the current user behavior significantly deviates from the historical click trend; When user behavior drifts significantly and aligns with the platform's policy control rhythm, an update to the recommendation or prediction model is triggered.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the Internet consumer intelligence monitoring method based on big data as described in claim 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the Internet consumer intelligence monitoring method based on big data as described in claim 8.

Images