A search engine based data updating method

Abstract

The application provides a search engine-based data updating method, which comprises the following steps: obtaining search behavior data of a user and extracting a personalized behavior feature vector of the user; predicting future search demand of the user according to the personalized behavior feature vector of the user, identifying a webpage that needs to be predicted and updated; predicting a behavior feature vector of the user at a future time; dynamically evaluating an updating priority of each webpage and generating a prediction strategy; designing a multi-level intelligent cache optimization strategy to optimize the prediction strategy of the search engine; designing an intelligent selective data updating mechanism according to the optimized prediction strategy and the updating priority of each webpage, specially optimizing a data updating scene in the search engine; and establishing an adaptive feedback loop to dynamically optimize the overall performance of the search engine. The application can effectively improve the updating efficiency of the search engine, reduce the waste of system resources, and enhance the search experience of the user.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of search engines, and particularly relates to a data updating method based on a search engine. BACKGROUND

[0002] With the explosive growth of Internet information, search engines have become an important tool for users to obtain information. The core function of a search engine is to quickly find the most relevant results in a vast amount of data, providing accurate and timely information for users. However, in order to achieve this, search engines need to continuously update their index databases to ensure the real-time and accuracy of search results. Traditional search engine data updating methods mainly rely on periodic crawling of web page content and updating of index libraries, but this approach has significant limitations.

[0003] Firstly, the traditional periodic crawling and updating strategy cannot effectively cope with the dynamic changes of network content. In the Internet environment, the frequency and range of web content updates have high uncertainty. If the search engine adopts a fixed time interval to update the data, it may lead to insufficient updates for some frequently changing web pages and excessive updates for some less frequently changing web pages, which not only wastes a lot of computing resources and network bandwidth, but also may lead to a decrease in the real-time and relevance of search results. Especially in the case of hot events, users want to quickly obtain the latest information, and the traditional method often fails to meet this real-time demand.

[0004] Secondly, the traditional data updating method usually ignores the dynamic changes of user behavior and search intent. Users' search behavior and interests are personalized and dynamic, and a fixed updating strategy cannot optimize for each user's search demand. This "one-size-fits-all" approach makes it difficult for search engines to accurately capture users' current search intent and cannot intelligently prefetch and update data according to users' actual needs, affecting user experience.

[0005] In addition, search engines face huge data volume and storage space challenges during data updating. With the continuous growth of the number of web pages, the index data that search engines need to manage also increases continuously. How to ensure the updating and maintenance of index library data without increasing system burden has become a problem to be solved. The traditional updating strategy cannot effectively utilize system resources, and when dealing with large-scale data, often causes resource waste, system overload and other problems. Especially for the cache management of search engines, how to effectively utilize cache space, ensure fast access to important data, and reduce unnecessary data redundancy and storage overhead is a difficult problem that needs to be solved.

[0006] To solve the above problems, the existing solutions such as the update method based on content change detection and the optimization of crawling strategy using web access logs, although to some extent, improve the efficiency and accuracy of search engine data update, but still have obvious shortcomings. These methods often lack in-depth analysis of user behavior, cannot achieve personalized real-time data update, and cannot dynamically balance resource consumption and data update quality in the update strategy. In addition, these methods lack effective cache management and data selection mechanism when facing large-scale data, resulting in waste of system resources and low update efficiency. Therefore, the current search engine data update technology still has many shortcomings, and an intelligent, dynamic and efficient update method is urgently needed. SUMMARY

[0007] The purpose of the present application is to design a search engine-based data update method, which introduces intelligent prediction and dynamic cache management multi-dimensional method, to solve the problems of poor dynamic response, insufficient personalization and low resource management efficiency in the prior art, and provides a new search engine data update solution. This solution can effectively improve the update efficiency of the search engine, reduce the waste of system resources, and enhance the user's search experience.

[0008] To achieve the above purpose, the present application provides a search engine-based data update method, characterized in that the method comprises the following steps:

[0009] S1, obtaining the search behavior data of the user and extracting the personalized behavior feature vector of the user, comprising:

[0010] The search behavior data of the user is integrated into the user behavior feature vector, representing the user's search preference at a specific time, and a sliding window mechanism is used to dynamically update the user's behavior data;

[0011] S2, predicting the user's future search demand according to the user's personalized behavior feature vector, and identifying the web pages that need to be predicted and updated in combination with the change of web page content, comprising:

[0012] According to the change of web page content, the long short-term memory network predicts the user's behavior feature vector at the future time, and according to the change of web page content and the predicted user's behavior feature vector at the future time, an adaptive prediction strategy is designed, which is set for the web pages with frequent content changes and strong user demand correlation;

[0013] S3, dynamically evaluating the update priority of each web page in combination with the personalized behavior feature vector and the change of web page content, and generating a prediction strategy, comprising:

[0014] According to the user future time behavior feature vector and the change of the web page content, a web page importance scoring model is designed, the web page importance scoring model includes user relevance score and content change score to obtain a comprehensive web page score, and according to the comprehensive web page score, the web page priority is sorted to generate a specific prediction strategy;

[0015] S4, a multi-level intelligent cache optimization strategy is designed to optimize the prediction strategy of the search engine, including:

[0016] A three-level cache system is designed, which is hot cache, warm cache and cold cache. According to the comprehensive web page score, the highest priority web page is placed in the hot cache, and the lowest priority web page is placed in the cold cache, which is shown as follows:

[0017]

[0018] Wherein, L c represents the cache level of the web page c, which determines which cache the web page should be placed in; U c represents the comprehensive web page score of the web page c, which measures the importance of the web page; θ1, θ2 represents the cache level threshold, which is dynamically calculated by the web page priority. θ1 is used to determine whether the web page should enter the hot cache, and θ2 is used to determine whether the web page should enter the warm cache or the cold cache;

[0019] A regularization term based on web page access frequency and timeliness is designed to dynamically adjust the cache strategy, so that the cache system can adapt to the change of user behavior, and prevent low-frequency web pages from occupying cache resources for a long time. It is shown as follows:

[0020]

[0021] Wherein, represents the comprehensive loss function of the web page c placed in the cache level, L c is the cache level of the web page c; U c represents the comprehensive web page score of the web page c; f c represents the access frequency of the web page c in a certain time window, which is counted through log data; Δt c represents the time interval between the last update time of the web page c and the current time; Α and Β represent the weights of the control of the influence of the web page access frequency and the update time on the cache level allocation;

[0022] S5, according to the optimized prediction strategy and the update priority of each web page, an intelligent selective data update mechanism is designed, which is specially optimized for the data update scene in the search engine, including:

[0023] A page update scoring mechanism is designed to evaluate whether each page should be updated, wherein the page update scoring mechanism comprises a comprehensive page score of the page c, a content change score and page cache level information, and is expressed as follows:

[0024]

[0025] wherein P c represents a comprehensive update score of the page c, and is used to determine whether data update is performed; ι, κ, ν respectively represent weights of priority score U c , content change rate C c and cache level contribution, represents a balance system resource consumption regularization term; represents a cache level contribution function, which is used to represent the influence of the cache level on the update decision, and is defined as:

[0026]

[0027] A dynamic update decision threshold θ u is introduced at the same time, the update strategy is flexibly adjusted, the frequency of page update is adaptively controlled, and finally whether the page is updated is determined according to the comprehensive update score of the page c and the update decision threshold, and finally the update decision U update of the page is obtained:

[0028] If P c ≥ θ u , the page c is selected to be updated, and data crawling, content parsing and index updating will be performed on the page;

[0029] If P c < θ u , the page is delayed for update or is not updated, and waits for subsequent evaluation;

[0030] S6, an adaptive feedback loop is established to dynamically optimize the overall performance of the search engine.

[0031] Further, the search behavior data of the user comprises a query word vector of the user at time t, a page number clicked by the user, a time of the user staying on the page and a historical query frequency of the user;

[0032] The user behavior feature vector is expressed as follows:

[0033] X t = α1·Q t + α2·log(1+T t ) + α3·F t

[0034] wherein X tdenotes the user's comprehensive behavior feature vector at time t, representing the user's search preference at that moment; Q t denotes the user's query word vector at time t, describing the semantic information of the user's search content; T t denotes the time the user stays on the clicked webpage, which is logarithmically scaled to avoid weight deviation caused by large differences in stay time; F t denotes the frequency of the user's past search behavior, reflecting the user's long-term interest in certain information; α1, α2, α3 denote weight coefficients for balancing the influence of different features on the final feature vector.

[0035] Further, the adaptive prediction strategy controls the prediction decision of the webpage by introducing a dynamic threshold θ, which is automatically adjusted according to the relevance of historical webpage update frequency and current prediction, and can dynamically determine which webpages need to be updated in priority according to the current prediction state, as shown below:

[0036]

[0037] wherein S c denotes the decision variable, indicating whether the webpage c should be predicted to update, if S c ≥ θ, the webpage is selected for prediction; denotes the predicted user future behavior feature vector, reflecting the user's future interest point; W c denotes the weight of the change in webpage content, reflecting the response of the change in webpage content to the user's demand; θ denotes the adaptive threshold, which is automatically adjusted according to the historical webpage update frequency and the current prediction result, and is used to control which webpages need to be updated in priority.

[0038] Further, the user relevance score is calculated based on the user's behavior feature vector at the future moment, and the content change score is the weight of the change in webpage content.

[0039] Further, the user relevance score and the content change score are combined to generate a comprehensive webpage score, as shown below:

[0040] U c = λ · R c + (1- λ) · C c

[0041] wherein U c denotes the comprehensive priority score of the webpage c, used to determine the order of webpage update; R c denotes the user relevance score, reflecting the matching degree of the webpage content and the user's future demand; C crepresents the content change score, reflecting the dynamic change of the web page content; λ represents the weight adjustment coefficient, controlling the influence of user demand and web page change on the final priority score;

[0042] According to the comprehensive web page score of each web page, the web page priority is ranked, and the web pages with scores higher than a set threshold θ p are selected for data prediction and update, represented as follows:

[0043]

[0044] wherein S c represents the prediction decision variable of the web page c, S c = 1 indicates that the web page will be predicted and updated, and S c = 0 indicates that the web page will not be updated; θ p represents the threshold value of the prediction strategy, used to control the number of predicted web pages and ensure that system resources are not excessively consumed.

[0045] Further, in the S4, the regularization term -α·log(f c ) in the loss function is used to reduce the possibility of high-frequency access web pages occupying low-level caches, and -β·Δt c is used to avoid outdated content occupying the cache for a long time.

[0046] Further, the S4 further comprises:

[0047] According to the changes in the priority of the web page and the access frequency, the cache data is periodically migrated; when the cache space is close to full, a dynamic elimination mechanism is introduced, and the elimination is performed according to the priority score of the web page and the current utilization rate of the cache level, and the data of the web page with low priority and long time not accessed will be removed from the cache, wherein a time decay model is used to control the cache retention time of the web page, represented as follows:

[0048]

[0049] wherein T c represents the storage time of the web page c in the cache, which is prolonged with the increase of the priority U c ; T max represents the maximum cache storage time, indicating the longest retention time of the web page data; ∈ represents the prevention of division by zero error when the priority is zero.

[0050] Further, the dynamic update decision threshold θ u is adaptively controlled according to the load and historical update of the system, and the frequency of web page update is controlled, represented as follows:

[0051] θ u= θ0·(1+η·LoadFactor)

[0052] where θ u represents the dynamic threshold for updating decision, controlling whether a web page should be updated; θ0represents the base updating threshold, reflecting the initial value under normal load condition; η represents the adjustment coefficient, used to control the degree of influence of the load factor on the threshold; LoadFactor represents the current load factor, reflecting the system resource usage, and the calculation formula is as follows:

[0053]

[0054] When the system load increases, the threshold θ u will be increased accordingly, reducing unnecessary update operations; when the load decreases, θ u will be reduced, allowing more web pages to be updated;

[0055] According to the comprehensive update score P c and the update decision threshold θ u , whether the web page is updated is decided, and the specific steps are as follows:

[0056] If P c ≥ θ u , the web page c is selected for updating, and data crawling, content analysis and index updating will be performed on the web page;

[0057] If P c < θ u , the web page is delayed or not updated, waiting for subsequent evaluation.

[0058] Further, the S6 is specifically analyzed as follows:

[0059] According to the update decision variable of the web page c, a feedback optimization model is designed through a feedback loop to dynamically adjust the threshold θ u ;

[0060] A feedback optimization function is designed to dynamically adjust the weights ι, κ, ν of the priority score U c , the content change rate C c and the cache level contribution, so that the system focuses on resource usage under high load, and focuses on content change rate and cache level under low load, which is represented as follows:

[0061]

[0062] wherein, represents the feedback optimization function, used to adjust the weight parameters ι, κ, ν, affecting the calculation of the web page update score; represents the current system query response time; represents the click rate data of users, reflecting the relevance and satisfaction of users to query results; T max represents the maximum allowed response time of the system as a performance target; C target represents the predetermined target click rate, reflecting the user click rate expected to be achieved by the system; ω1, ω2 represent coefficients for balancing the importance of response time and click rate in the weight optimization process;

[0063] A feedback correction mechanism based on historical data is constructed to correct the regularization term in the updated score, wherein the feedback correction mechanism based on historical data corrects the regularization term according to historical update results and user behavior feedback, ensures that low-priority, high-frequency access web pages are not updated excessively, but are optimized according to actual needs, and is represented as follows:

[0064]

[0065] wherein λ new represents the updated regularization parameter for controlling the update frequency of high-frequency access web pages; λ old represents the original regularization parameter; μ represents an adjustment coefficient for controlling the influence degree of the regularization parameter when the click rate deviates from the target; represents the actual user click rate; C target represents the target click rate.

[0066] Further, the feedback optimization model is constructed as follows:

[0067] LoadFactor max is defined as the ratio of the current resource usage of the system to the maximum resource, and is represented as follows:

[0068]

[0069] wherein, represents the current CPU usage of the system. represents the current memory usage; and represent the maximum CPU and memory capacity of the system;

[0070] According to the load factor LoadFactor max , the threshold value θ u is dynamically adjusted to ensure that the update operation is reduced when the load is high, and vice versa:

[0071] θ u = θ0· (1 + η· LoadFactor max )

[0072] wherein θ​u represents the dynamic threshold for updating decisions, controlling whether the webpage should be updated; represents the base update threshold, reflecting the initial value under normal load conditions; represents the adjustment coefficient, used to control the degree of influence of the load factor on the threshold.

[0073] The beneficial technical effects of the present application are at least the following points:

[0074] The present application proposes a search engine-based data updating method, aiming to solve the shortcomings of traditional methods in dealing with dynamically changing network content, personalized user behavior, and large-scale data management. This method effectively improves the data updating efficiency and accuracy of the search engine through the collaborative work of multiple modules.

[0075] Firstly, the present application introduces a user behavior prediction and intelligent prefetching module. By analyzing the user's search history and behavior characteristics, it uses deep learning algorithms to predict the user's future search demand and pre-fetch relevant webpage data. This method overcomes the shortcomings of traditional updating strategies that cannot dynamically respond to changes in user search intentions, significantly improving the real-time performance and personalized experience of data updating.

[0076] Secondly, the present application designs a multi-level cache management module based on fuzzy logic. This module combines multiple factors such as webpage access frequency, content change rate, and user click rate to calculate the priority of data updating, and optimizes data storage through multi-level cache strategy. This innovation solves the problem of improper resource utilization in traditional cache management methods, ensuring that high-priority data can be quickly accessed while reducing the storage overhead of low-priority data, effectively improving the resource utilization efficiency of the system.

[0077] In summary, the present application introduces intelligent prediction and dynamic cache management methods to address the poor dynamic response, lack of personalization, and low resource management efficiency in existing technologies, providing a new search engine data updating solution. This solution can effectively improve the updating efficiency of the search engine, reduce system resource waste, and enhance the user's search experience. BRIEF DESCRIPTION OF DRAWINGS

[0078] The present application is further illustrated by the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present application. For ordinary skilled in the art, other drawings can be obtained without creative labor based on the following drawings.

[0079] Figure 1 A search engine-based data updating method flowchart for the embodiments of the present application. DETAILED DESCRIPTION

[0080] Embodiments of the present application are described below in the detailed description and illustrated in the accompanying drawings by way of examples which, as will become apparent to those skilled in the art upon reading the present document, can be implemented in various ways. The embodiments described below are examples only, and are not to be construed as limiting the present application.

[0081] In one or more embodiments, as shown in Figure 1 a search engine-based data updating method is disclosed, the method comprising steps S1-S6, comprising:

[0082] S1, obtaining the search behavior data of the user, and extracting the personalized behavior feature vector of the user.

[0083] Specifically, the behavior data of the user is obtained in real time from the search engine, including but not limited to:

[0084] Q t represents the query word vector of the user at time t (the word vector can be processed by a pre-trained NLP model such as Word2Vec or BERT);

[0085] C t represents the web page number (unique identifier) clicked by the user;

[0086] T t represents the time the user stays on the web page;

[0087] F t represents the historical query frequency of the user.

[0088] These data constitute the input data set of user behavior, reflecting the search behavior of the user in a specific time period.

[0089] Further, the multi-dimensional user behavior data is integrated into the user behavior feature vector X t , which is used to describe the search preference of the user at a specific time t. Formula:

[0090] X t = α1·Q t + α2·log(1+T t )+ α3·F t

[0091] Where X t represents the comprehensive behavior feature vector of the user at time t, representing the search preference of the user at that moment. Q t represents the query word vector of the user at t, describing the semantic information of the user's search content. T t represents the time the user stays on the clicked web page, which is scaled by logarithm to avoid weight deviation due to large difference in stay time. Ft represents the frequency of the user's past search behavior, reflecting the user's long-term interest in a certain type of information. α1, α2, α3 represent weight coefficients for balancing the influence of different features on the final feature vector. This formula ensures that behavior data of different dimensions are integrated into a feature vector with clear physical meaning, and the balance between different sources of data is ensured through reasonable weight adjustment.

[0092] Further, the user's behavior feature vector is dynamic and needs to be updated over time and with changes in user behavior. The present application uses a sliding window mechanism to dynamically update the user's historical behavior, ensuring the real-time nature of the feature vector. The update mechanism is:

[0093] X t+1 = β · X t + (1 - β) · X new

[0094] where X t+1 represents the updated behavior feature vector of the user at time t+1. X new represents the newly generated behavior feature. β represents the update coefficient, which controls the weight of historical behavior and new behavior. When β is close to 1, it means that the historical behavior feature is more inclined to be maintained, while when β is close to 0, it means that the latest behavior is more inclined to be adopted. This dynamic updating mechanism can effectively balance historical behavior and recent behavior, ensuring that the feature vector can reflect the user's long-term interest and current demand.

[0095] Finally, after processing the behavior data of each user at different times, the user behavior feature vector X t is obtained, which will be used as input for subsequent steps for user future behavior prediction and data update priority evaluation. Through this series of steps, the user's personalized behavior feature vector is successfully constructed. This feature vector accurately captures the user's search preferences and can dynamically change over time, serving subsequent user behavior prediction and data update strategy formulation.

[0096] S2, according to the user's personalized behavior feature vector, predict the user's future search demand, and identify the web pages that need to be predicted and updated in combination with the changes in web page content.

[0097] Specifically, the present application designs an adaptive time series model that combines changes in web page content. Traditional time series models can only capture the temporal dependence of user behavior, and cannot effectively combine the dynamic changes of web page content to predict the user's future demand. Therefore, the present application improves the LSTM model to enable it to process information W c about changes in web page content, and use it as a control input to enable the model to automatically adjust the prediction according to changes in web page content.

[0098] The model not only relies on user behavior features X t but also combines the weight of web content changes W c so that the changes in web content can directly affect the prediction of future user behavior. This design solves the problem of integrating the continuous changes in user behavior and the sudden changes in web content, ensuring that the model can accurately predict in complex search scenarios. Formula:

[0099]

[0100] where, represents the predicted user behavior feature vector at the future time, reflecting the user's possible future search interest. LSTM represents a long short-term memory network that processes time series data and adapts to the prediction of web content changes. X t represents the user's behavior feature input at time t. W c represents the weight of web content changes, reflecting the response of web content changes to user demand. This design ensures that the model can capture sudden changes in web content when user behavior changes, thereby generating more accurate prediction results.

[0101] It can be understood that according to the predicted user future behavior feature and the weight of web content changes W c , the present application designs an adaptive prefetching strategy, especially for web pages with frequent content changes and strong user demand correlation. For this purpose, the present application introduces a dynamic threshold θ to control the prefetching decision of the web page. This threshold is not fixed, but is automatically adjusted according to the historical web update frequency and the current prediction relevance, so that the system can dynamically determine which web pages need to be updated in priority according to the current prediction state. Formula:

[0102]

[0103] where, S c represents the decision variable, indicating whether the web page c should be prefetched and updated, if S c ≥ θ, the web page is selected for prefetching. represents the predicted user future behavior feature vector, reflecting the user's future interest point. W c represents the weight of web content changes, reflecting the response of web content changes to user demand. θ represents the adaptive threshold, which is automatically adjusted according to the historical web update frequency and the current prediction result, and is used to control which web pages need to be updated in priority.

[0104] This design solves the limitations of traditional fixed threshold strategies in handling mismatch between web content and user demand changes. With adaptive thresholds, the system can flexibly handle web prefetching needs in different situations, ensuring that highly relevant pages are updated in time, while low-frequency change pages do not consume excessive system resources.

[0105] Finally, this step outputs two important results: represents the predicted user future behavior feature vector, which is used for the next step of dynamic priority evaluation. c represents the prefetching strategy, which is used to determine which pages will be updated first.

[0106] These outputs ensure that the search engine can respond to user demand changes in time and dynamically adjust the data update strategy, thereby improving the response speed and accuracy of the search engine.

[0107] S3, combining the personalized behavior feature vector and the change of web content, dynamically evaluates the update priority of each web page and generates a prediction strategy.

[0108] Specifically, to effectively evaluate the priority of web pages, the present application designs a web importance scoring model that combines user future behavior features with web content changes to quantify the priority of web pages.

[0109] The score is composed of two main factors:

[0110] User relevance score: based on predicted user behavior features The present application calculates the relevance of each web page to user future demand. Assuming the content of web page c is represented by feature vector V c , the user relevance can be calculated by the following formula:

[0111]

[0112] where R c represents the user relevance score of web page c, and represents the matching degree of web content to user future demand. represents the predicted behavior feature vector of the user at the future time. V c represents the content feature vector of web page c, and represents the core information of the web page.

[0113] Content change score: to reflect the timeliness of web content, the present application introduces the weight W c of web content change to measure the dynamic change speed of web content. The greater the change of the web page, the higher the necessity of its update. Combined with the change rate of the web page, the present application defines the content change score as:

[0114] C c = Wc

[0115] where C c represents the content change score of webpage c, and R c represents the content update rate of webpage c. W c represents the weight of webpage content change, which is calculated in the previous step.

[0116] Further, the user relevance score and the content change score are combined to generate the comprehensive priority score of the webpage. To balance the influence of user demand and content change, the present application introduces an adjustment coefficient λ to determine the relative importance of user behavior and content change to webpage update. Formula:

[0117] U c = λ·R c + (1-λ)·C c

[0118] where U c represents the comprehensive priority score of webpage c, which is used to determine the order of webpage update. R c represents the user relevance score, reflecting the matching degree of webpage content and future user demand. C c represents the content change score, reflecting the dynamic change of webpage content. λ represents the weight adjustment coefficient, controlling the influence of user demand and webpage change on the final priority score.

[0119] Further, webpage content change and user demand change are often out of sync in time, so the traditional priority evaluation method cannot effectively handle this complexity. This scheme introduces an adaptive adjustment coefficient λ, which can flexibly adjust the relative weight of user demand and webpage change in different scenarios, ensuring optimal update decisions in different update scenarios.

[0120] Further, according to the comprehensive priority score U c of each webpage, a specific prefetching strategy is generated. The present application sorts the webpage priority and selects webpages with a score higher than a set threshold θ p for data prefetching and updating. Formula:

[0121]

[0122] where S c represents the prefetching decision variable of webpage c, S c = 1 indicates that the webpage will be prefetched and updated, and S c = 0 indicates that the webpage will not be updated. U c represents the comprehensive priority score of the webpage. θ p represents the threshold value of the prefetching strategy, which is used to control the number of prefetched webpages, ensuring that system resources are not excessively consumed.

[0123] The application designs a dynamic prefetch threshold θ p based on historical queries and content change density, ensuring quick response when data changes frequently and reducing system load when data changes slowly.

[0124] Output: The results of this step include:

[0125] S c represents the webpage prefetch strategy, which determines which webpages will be updated first.

[0126] U c represents the comprehensive priority score list, indicating the order of webpage updates.

[0127] These outputs provide clear guidance for the actual update operation of the system, ensuring that the system can prioritize updating those webpages that are most valuable to users and change most quickly, thereby improving the response speed of search engines and the timeliness of data.

[0128] S4, design a multi-level intelligent cache optimization strategy to optimize the prediction strategy of the search engine.

[0129] Specifically, this step relies on the webpage prefetch decision S c and the webpage comprehensive priority score U c generated in the previous step. S c represents the webpage prefetch decision variable, S c = 1 indicates that the webpage needs to be updated first, and S c = 0 indicates that the webpage does not need to be updated. U c represents the comprehensive priority score of the webpage, which is evaluated by the user demand relevance and webpage content change speed.

[0130] Further, the application designs a three-level cache system, which includes:

[0131] Hot cache: specifically stores webpages with strong correlation to future user behavior and high priority score, ensuring fast response to user queries.

[0132] Warm cache: stores webpages with medium priority and moderate access frequency, ensuring certain response speed of content without consuming a large amount of cache resources.

[0133] Cold cache: stores low-priority and slow-changing webpage data, which may be occasionally accessed but does not need to be frequently updated.

[0134] The application uses priority score U cThe dynamic adjustment of the position of the webpage in different cache levels ensures that high-priority webpages are placed in the hot cache first, and low-priority webpages are placed in the cold cache. In order to deal with the complexity of the cache system, especially the characteristics of the webpage priority changing over time, the present application proposes an adaptive dynamic cache level allocation mechanism.

[0135] Further, the design of the cache strategy mainly considers the priority score U c and the prefetching demand S c of the webpage, and the present application designs a multi-dimensional scoring mechanism for each webpage, based on which the webpage will be allocated to the appropriate cache level. Formula:

[0136]

[0137] wherein L c represents the cache level of the webpage c, which determines in which cache the webpage should be placed. U c represents the comprehensive priority score of the webpage c, which measures the importance of the webpage. θ1, θ2 represent the cache level threshold, which is dynamically calculated based on the webpage priority. θ1 is used to determine whether the webpage should enter the hot cache, and θ2 is used to determine whether the webpage should enter the warm cache or the cold cache.

[0138] Further, in order to optimize the effect of cache level division, the present application designs a regularization term based on the access frequency and timeliness of the webpage for the cache frequency and priority adjustment problem in the search engine scenario. The regularization term is used to dynamically adjust the cache strategy, so that the cache system can adapt to the changes in user behavior, and prevent low-frequency webpages from occupying cache resources for a long time. Formula:

[0139]

[0140] wherein, represents the comprehensive loss function of the webpage c placed in the cache level, L c is the cache level decision. U c represents the priority score of the webpage c. f c represents the access frequency of the webpage c within a certain time window, which is counted through log data. Δt c represents the time interval between the last update time of the webpage c and the current time. A and B represent the weights controlling the influence of the access frequency and the update time of the webpage on the cache level allocation.

[0141] wherein the regularization term -α·log(f c ) in this loss function is used to reduce the possibility of high-frequency access webpages occupying low-level caches, and -β·Δt cTo avoid outdated content long-term occupation of cache, ensure timeliness. Through the formula, the application can dynamically adjust the cache strategy, and balance the relationship between webpage priority and cache space use.

[0142] Further, in order to ensure efficient use of cache space, avoid cache overflow, the application designs a cache eviction and migration mechanism based on timeliness and priority dynamic adjustment. The specific steps are as follows:

[0143] According to the change of webpage priority and access frequency, the cache data will be migrated regularly. For example, when the priority score U c of the webpage decreases or the access frequency decreases, the webpage will be migrated from hot cache to warm cache, or even cold cache. The operation is determined by the above formula , and adjusted through cache log;

[0144] When the cache space is close to full, the application introduces a dynamic eviction mechanism, which evicts according to the priority score of the webpage and the current utilization rate of the cache level. The webpage data with low priority and long time without access will be removed from the cache. The application uses a time decay model to control the cache retention time of the webpage:

[0145]

[0146] Where T c represents the storage time of webpage c in the cache, and the storage time is prolonged with the increase of priority U c . T max represents the maximum cache storage time, and represents the maximum retention time of webpage data. ∈ represents the prevention of zero division error when the priority is zero.

[0147] It can be understood that by introducing the regularization term of timeliness control, the application can ensure that the cache will not be occupied by low-frequency webpage data for too long, and high-frequency high-priority webpages can obtain more cache retention time. The model ensures that the cache space can be flexibly adjusted and reduces the risk of cache overflow through the access frequency f c of the webpage and the update time interval Δt c .

[0148] Further, the application monitors the execution of the cache strategy in real time, and designs a feedback loop to dynamically adjust the cache threshold θ1, θ2 through user query log and webpage access data, ensuring the adaptability of the system under high load.

[0149] These outputs ensure that the system can respond to changes in user behavior in real time, dynamically adjust cache space allocation, avoid response delay of the system under high load, and improve the intelligent level of cache management through the feedback mechanism.

[0150] S5, design an intelligent selective data update mechanism according to the optimized prediction strategy and the update priority of each webpage, and specially optimize the data update scene in the search engine.

[0151] Specifically, the goal of this step is to design an intelligent selective data update mechanism based on the cache strategy and webpage priority score of the previous step, and to specially optimize the data update scene in the search engine. In the search engine environment, due to the frequent changes of webpage content and the complex user needs, data must be selectively updated to avoid resource waste and delay caused by full update. The particularity of this scheme is to introduce additional regular terms, dynamic threshold adjustment mechanism and innovative update scoring mechanism to ensure efficient use of cache resources and maximize system response speed and update accuracy.

[0152] The input data includes the webpage cache level information L c and the webpage priority score U c generated in the previous step, which are used for intelligent selective update decision. Among them, L c represents the webpage cache level information, which determines where the webpage is stored (hot cache, warm cache or cold cache). U c represents the comprehensive priority score of the webpage, reflecting user demand, content relevance and change frequency.

[0153] Further, the application designs a special update scoring mechanism P c for evaluating whether each webpage should be updated. The scoring mechanism consists of the following three parts:

[0154] Priority score U c represents the importance of the webpage, which is calculated in combination with the user's query demand and the change frequency of the webpage content.

[0155] Content change rate C c represents the degree of change of the webpage content, which is obtained in the previous step through the weight W c of the webpage content.

[0156] Cache level contribution L c represents the storage situation of the webpage in different cache levels, and the webpage in the hot cache is updated first, and the webpage in the cold cache can be delayed.

[0157] Further, in order to ensure the flexibility of the update strategy, the application designs a comprehensive update scoring formula with regular terms:

[0158]

[0159] Among them, P crepresents the comprehensive update score of webpage c, which is used to determine whether to update data. c , content change rate C c and the weight of cache level contribution. represents the cache level contribution function, which is used to represent the influence of cache level on update decision, and is defined as:

[0160]

[0161] The webpage in the hot cache needs to be updated preferentially because it has a high cache level contribution due to high access frequency and high priority. represents a newly introduced regularization term, which is used to balance system resource consumption and avoid excessive updating of high-frequency and low-change webpages. The regularization term reflects the recent update frequency of the webpage, avoiding redundant operations caused by updating the webpage too frequently. The formula is defined as:

[0162]

[0163] wherein f c represents the access frequency of webpage c. Δt c represents the time interval between the last update of webpage c and the current time. This regularization term ensures that webpages with low-frequency updates and long time intervals are considered preferentially, while high-frequency update webpages need to meet higher standards to be updated again, preventing excessive resource waste.

[0164] It can be understood that by adding the cache level contribution and the regularization term, the system can intelligently determine which webpages should be updated preferentially, avoiding blind updates based only on webpage change rate or priority score. In particular, the introduction of the regularization term effectively controls the excessive update of high-access-frequency webpages, reducing unnecessary calculation and data traffic consumption.

[0165] Further, the present application introduces a dynamic update decision threshold θ u for flexible adjustment of the update strategy. The design purpose of the threshold θ u is to adaptively control the frequency of webpage updates according to the load and historical update of the system. Specifically, θ u is dynamically adjusted according to the following formula:

[0166] θ u = θ0·(1+κ·LoadFactor)

[0167] wherein θ uThe threshold value representing the current update decision. θ0represents the basic update threshold, which is set according to historical query data. κ represents the load influence coefficient, which is used to control the influence of load on the update threshold. LoadFactor represents the current load factor of the system, reflecting the resource usage rate of the system, and the calculation formula is as follows:

[0168]

[0169] When the system load increases, the threshold value θ u will be increased accordingly to reduce unnecessary update operations; when the load is low, θ u will be reduced to allow more web pages to be updated. This method of dynamically adjusting the threshold value ensures that the system can adaptively adjust the update strategy under high load and low load conditions.

[0170] Further, according to the comprehensive update score P c and the update decision threshold θ u , the present application makes a decision on whether to update the web page, and the specific steps are as follows:

[0171] If P c ≥ θ u , the web page c is selected for update, and the system will perform data crawling, content parsing and index updating on the web page.

[0172] If P c < θ u , the web page is delayed for update or not updated, waiting for subsequent evaluation.

[0173] The execution flow of the update operation can quickly respond to changes in user behavior and ensure that high-priority content is updated in a timely manner, reducing resource occupation of low-priority content. The execution process includes:

[0174] Data crawling: extracts content from the original web page and stores it in the cache or index.

[0175] Content parsing: uses a pre-set parser to perform semantic analysis on the web page content to generate new index data.

[0176] Index updating: replace the old index with the latest data to ensure that the latest web page information is displayed when querying.

[0177] Feedback mechanism and dynamic optimization: In order to ensure that the system can maintain optimal performance during long-term operation, the present application designs a feedback mechanism to optimize and adjust the weight parameters α, β, γ and the coefficient λ of the regularization term by monitoring query response time, user click rate and other data after the update operation. At the same time, the system load factor LoadFactor will also be dynamically updated to ensure load balancing.

[0178] Finally, output Uupdate The update decision of the webpage indicates which webpages need to be updated; the updated webpage data and index ensure that the search engine provides the latest content to the user; through monitoring of the system behavior, the parameter configuration of the update strategy is optimized, and an adaptive adjustment process is formed.

[0179] The scheme realizes an efficient and flexible selective data update mechanism by combining the webpage priority, content change rate, cache level contribution, and regularization control of the update frequency. In particular, through dynamic threshold adjustment and regularization terms, the system can adaptively update according to the load state and webpage access characteristics, optimize the use efficiency of system resources, and ensure timely updating of high-priority webpages.

[0180] S6, an adaptive feedback loop is established to dynamically optimize the overall performance of the search engine.

[0181] Specifically, the input of this step comes from the update decision variable U of the previous step update , and the feedback data R of the system sys . The input includes:

[0182] U update represents the update decision variable of the webpage, indicating which webpages are selected for updating (U update = 1) or delayed updating (U update = 0).

[0183] R sys represents the system feedback data, containing multiple dimensions of data, including system load, query response time, user click rate, and other real-time dynamic indicators. These data serve as feedback to guide the dynamic adjustment of system parameters.

[0184] Further, in order to realize adaptive optimization, the system needs to continuously adjust its parameters according to the feedback information in actual operation. The core optimization goal is to ensure that unnecessary update operations are reduced under high load, improving the utilization efficiency of system resources; and under low load, the update frequency is increased to improve the real-time performance of data. The present invention realizes this goal through the following three core optimization steps:

[0185] (1) The update decision of the previous step is based on the threshold value θ u to determine whether to update the webpage. The present invention dynamically adjusts this threshold value through the feedback loop according to the system load. The system load factor LoadFactor is defined as the ratio of the current resource usage to the maximum resource:

[0186]

[0187] wherein, represents the current CPU usage of the system, The current memory usage of the system. And The maximum CPU and memory capacity of the system.

[0188] (2) According to the load factor LoadFactor max , dynamically adjust the threshold value θ u , ensure that the system can reduce the update operation at high load, and vice versa at low load:

[0189] θ u = θ0·(1+η·LoadFactor max )

[0190] Where θ u represents the dynamic threshold value for update decision, controlling whether the webpage should be updated. θ0 represents the basic update threshold value, reflecting the initial value under normal load. η represents the adjustment coefficient, used to control the degree of influence of the load factor on the threshold value.

[0191] It can be understood that by introducing the system load factor, the system can automatically raise the threshold value during the peak load period, reduce unnecessary update operations, and thus relieve resource pressure; at low load, the system can reduce the threshold value, allowing more webpage content to be updated, ensuring the real-time nature of the data and user experience.

[0192] Further, in order to further improve the adaptive ability of the system, the present application designs a weight optimization mechanism for dynamically adjusting the weights α, β, γ of each factor in the update score. These weights affect the relative importance of the priority score U c , the content change rate C c and the cache level contribution L c in the webpage update decision. The system dynamically adjusts these weights through the feedback optimization function , so that the system pays more attention to resource usage at high load, and more considers the content change rate and cache level at low load. The specific optimization formula is as follows:

[0193]

[0194] Where, represents the feedback optimization function, used to adjust the weight parameters ι, κ, ν, affecting the calculation of the webpage update score. represents the current system query response time. represents the user click rate data, reflecting the relevance and satisfaction of the user query result. T max represents the maximum allowed response time of the system as a performance target. C targetω1, ω2 represent coefficients to balance the importance of response time and click rate in the weight optimization process.

[0195] Further, by introducing the feedback optimization function, the system can adaptively adjust the weights in the webpage update score, so that when the response time is longer, the system pays more attention to resource utilization, and when the click rate drops, the system considers the update frequency of the content more. This mechanism ensures that the system always maintains optimal performance under dynamic load.

[0196] Further, in addition to optimizing the update threshold and weight distribution, the system also introduces a feedback correction mechanism based on historical data to gradually optimize the regularization term in the comprehensive update score formula. Specifically, the system corrects the regularization term according to historical update results and user behavior feedback, ensuring that low-priority, high-frequency access webpages are not updated excessively, but are optimized according to actual needs. The optimization function of the regularization term is as follows:

[0197]

[0198] where λ new represents the updated regularization parameter, used to control the update frequency of high-frequency access webpages. λ old represents the original regularization parameter. μ represents the adjustment coefficient, used to control the influence of the click rate deviating from the target on the regularization parameter. represents the actual user click rate. C target represents the target click rate.

[0199] When the user click rate is lower than the target click rate, the regularization term will be adjusted to reduce the update priority of high-frequency access webpages, thereby better allocating system resources.

[0200] Further, the system continuously monitors the feedback data in R sys , including system load, response time, user click rate, etc. According to the load factor in the feedback, the update threshold θ u is adjusted in real time to reduce the update frequency at high load and increase the update at low load. Through the feedback optimization function dynamically adjust the weights α, β, γ in the update score, so that the system can balance the update frequency and performance in different scenarios. Combined with historical data and user behavior feedback, the regularization term is optimized to prevent low-priority webpages from being frequently updated.

[0201] It can be understood that in the patent scenario of the search engine, due to the need of the system to process a large number of user requests and maintain the real-time of data, the feedback mechanism is the key to ensure the adaptive optimization of the system. Through the feedback loop, the system can adjust the parameters in real time according to the actual running condition, and ensure the optimal response under different loads and user behaviors. Efficient feedback optimization not only reduces unnecessary resource consumption, but also improves the accuracy of user query and the overall performance of the system.

[0202] The scheme constructs a perfect adaptive feedback loop by dynamically adjusting the update threshold, weight optimization and regular term correction. The system can flexibly adjust the update strategy and cache management under different load and user behavior conditions, thereby realizing the best performance, and is especially suitable for complex high-concurrency scenarios such as search engines.

[0203] The above are some preferred embodiments of the present application, of course, cannot be limited by the scope of the present application, those skilled in the art can understand that the whole or part of the process of realizing the above embodiment, and the equivalent changes made by the claims of the present application, still belong to the scope covered by the application.

Claims

1. A data update method based on a search engine, characterized in that, The method includes the following steps: S1. Obtain user search behavior data and extract personalized behavioral feature vectors, including: User search behavior data is integrated into a user behavior feature vector, which represents the user's search preferences within a specific time period, and a sliding window mechanism is used to dynamically update the user behavior data. S2. Predict users' future search needs based on their personalized behavioral feature vectors, and identify web pages that need to be predicted and updated, including: A long short-term memory network is constructed based on changes in web page content to predict the user's future behavioral feature vector. At the same time, an adaptive prediction strategy is designed based on changes in web page content and the predicted user's future behavioral feature vector. The adaptive prediction strategy is set for web pages with frequent changes in content and strong relevance to user needs. S3. Dynamically evaluate the update priority of each webpage by combining personalized behavioral feature vectors and changes in webpage content, and generate a prediction strategy, including: A webpage importance scoring model is designed by combining the user's future behavioral feature vector with changes in webpage content. The webpage importance scoring model includes a user relevance score and a content change score to obtain a comprehensive webpage score. Based on the comprehensive webpage score, webpage priorities are ranked to generate specific prediction strategies. S4. Design a multi-level intelligent caching optimization strategy to optimize the search engine's prediction strategy, including: A three-level caching system was designed: hot cache, warm cache, and cold cache. Based on the comprehensive webpage score, the highest priority webpages are placed in the hot cache, while the lowest priority webpages are placed in the cold cache, as shown below: ； in, Represents a webpage The cache level determines which cache a webpage should be placed in; Represents a webpage A comprehensive webpage score measures the importance of a webpage; This represents the cache level threshold, which is dynamically calculated based on the page priority. Used to determine whether a webpage should be cached. Used to determine whether a webpage should be cached in a warm cache or a cold cache; The design incorporates a dynamic caching strategy based on regularization terms related to webpage access frequency and timeliness. This allows the caching system to adapt to changes in user behavior and prevents infrequently accessed webpages from occupying cache resources for extended periods. The strategy is illustrated below: ； in, Represents a webpage The comprehensive loss function placed at the cache level, It's a webpage. Cache hierarchy; Represents a webpage Overall webpage rating; Represents a webpage Access frequency within a certain time window is statistically analyzed using log data. Represents a webpage The time interval between the last update time and the current time; and This indicates the weight of the impact of controlling webpage access frequency and update time on cache hierarchy allocation; S5. Based on the optimized prediction strategy and the update priority of each webpage, design an intelligent selective data update mechanism, and perform special optimizations for data update scenarios in search engines, including: The design includes an update scoring mechanism to evaluate whether each webpage should be updated, wherein the update scoring mechanism includes webpages... The comprehensive webpage score, content change score, and webpage caching level information are represented as follows: ； in, Represents a webpage The overall update score is used to determine whether to update the data. The overall webpage ratings are as follows: Content Change Rating And the weight of the contribution of the cache level, This represents a regularization term to balance the resource consumption of the system; The cache level contribution function, used to represent the impact of the cache level on update decisions, is defined as follows: ； Simultaneously, a dynamically updated decision threshold is introduced. Flexibly adjust update strategies, adaptively control the frequency of webpage updates, and finally, based on the webpage... The system uses a comprehensive update score and update decision threshold to determine whether a webpage should be updated, ultimately resulting in the webpage update decision. : like Then the webpage If selected for updating, the webpage will undergo data scraping, content parsing, and index updates. like If so, the webpage will be delayed in updating or will not be updated, pending further evaluation; S6. Establish an adaptive feedback loop to dynamically optimize the overall performance of the search engine.

2. The data update method based on a search engine according to claim 1, characterized in that, The user's search behavior data includes the user's search time. The query term vectors at the time, the page number clicked by the user, the time the user stayed on the page, and the user's historical query frequency; The user behavior feature vector is represented as follows: ； in, Indicates the user's time The comprehensive behavioral feature vector at that time represents the user's search preferences at that moment; Indicates that the user is The query term vectors at that time describe the semantic information of the user's search content; This represents the time a user spends on the webpage they click, and it has been logarithmically scaled to avoid weight shifts due to large differences in dwell time. It indicates the frequency of a user's past search behavior and reflects the user's long-term interest in a certain type of information; This represents the weighting coefficient, used to balance the influence of different features on the final feature vector.

3. The data update method based on a search engine according to claim 1, characterized in that, The adaptive prediction strategy introduces a dynamic threshold. The dynamic threshold controls the predictive decision-making of web pages. The system automatically adjusts based on the correlation between historical webpage update frequency and current prediction, dynamically determining which webpages need to be prioritized for updating based on the current prediction status, as shown below: ； in, Represents decision variables, referring to web pages Should it be predicted and updated? If so, the webpage is selected for prediction; This represents the predicted feature vector of the user's future behavior, reflecting the user's future interests. The weight of changes in webpage content reflects how the webpage content responds to changes in user needs. This represents a dynamic threshold that is automatically adjusted based on the historical webpage update frequency and the current prediction results. It is used to control which webpages need to be updated first.

4. The data update method based on a search engine according to claim 1, characterized in that, The user relevance score is calculated based on the user's future behavioral feature vector to determine the relevance of each webpage to the user's future needs; the content change score is the weight of webpage content changes.

5. The data update method based on a search engine according to claim 4, characterized in that, The user relevance score and content change score are combined to generate a comprehensive webpage score, as shown below: ； in, Represents a webpage A comprehensive webpage score is used to determine the order in which webpages are updated; This represents the user relevance score, reflecting the degree to which the webpage content matches the user's future needs; This indicates a content change score, reflecting the dynamic changes in the webpage content; This represents the weighting adjustment coefficient, which controls the impact of user needs and webpage changes on the overall webpage score. Web pages are prioritized based on their overall page score, and those with scores above a set threshold are selected. The webpage performs data prediction and updates, as shown below: ； in, Represents a webpage Decision variables, This indicates that the webpage will be predicted to be updated. This indicates that the webpage will not be updated; This represents the threshold of the prediction strategy, used to control the number of web pages predicted and ensure that system resources are not excessively consumed.

6. The data update method based on a search engine according to claim 1, characterized in that, In S4, the loss function Regularization term in This is used to reduce the likelihood of frequently accessed web pages consuming low-level cache. This is used to prevent outdated content from occupying the cache for a long time.

7. A data update method based on a search engine according to claim 6, characterized in that, The S4 further includes: Based on changes in webpage priority and access frequency, cached data is periodically migrated across different levels. When the cache space is nearing full capacity, a dynamic eviction mechanism is introduced, which evicts webpages based on their priority score and the current utilization rate of the cache level. Low-priority webpages that have not been accessed for a long time are removed from the cache. A time decay model is used to control the cache retention time of webpages, as shown below: ； in, Represents a webpage Storage time in the cache; This indicates the maximum cache storage time, representing the longest time webpage data can be retained. This prevents division by zero errors when the overall webpage score is zero.

8. A data update method based on a search engine according to claim 1, characterized in that, The dynamically updated decision threshold The frequency of webpage updates is adaptively controlled based on system load and historical update data, as shown below: ； in, This represents the dynamic update decision threshold used to update the decision. Control whether the webpage should be updated; This represents the basic update threshold, reflecting the initial value under normal load conditions; This represents the adjustment coefficient, used to control the degree of influence of the load factor on the threshold. The load factor represents the current load factor, reflecting the system resource utilization rate. The calculation formula is as follows: ； When system load increases, the threshold This will correspondingly increase efficiency and reduce update operations; when the load decreases, It will reduce the time required, allowing more web pages to be updated; Based on the comprehensive updated rating and dynamically update decision thresholds The specific steps to decide whether to update a webpage are as follows: like Then the webpage If selected for updating, the webpage will undergo data scraping, content parsing, and index updates. like If so, the webpage will be delayed in updating or will not be updated, pending further evaluation.

9. A data update method based on a search engine according to claim 1, characterized in that, The specific analysis of S6 is as follows: According to the webpage The decision variables are updated, and a feedback optimization model is designed through a feedback loop to dynamically adjust the threshold. ; The design feedback optimization function dynamically adjusts the overall webpage score. Content Change Rating Weight of contribution from cache level This allows the system to focus on resource usage under high load, while focusing on the rate of content change and cache level under low load, as shown below: ； in, This represents the feedback optimization function, used to adjust the weight parameters. This affects the calculation of webpage update scores; Indicates the current system query response time; This represents user click-through rate data, reflecting the relevance and satisfaction of user query results; This represents the system's maximum allowable response time, serving as a performance target. This indicates the predetermined target click-through rate, reflecting the user click-through rate that the system expects to achieve. This represents a coefficient used to balance the importance of response time and click-through rate in the weighting optimization process; A feedback correction mechanism based on historical data is constructed to correct the regularization term in the comprehensive update score. This mechanism adjusts the regularization term based on historical update results and user behavior feedback. The adjustments are made to ensure that low-priority, frequently accessed pages are not over-updated, but rather optimized according to actual needs, as shown below: ； in, This represents the updated regularization parameters, used to control the update frequency of frequently accessed web pages; Represents the original regularization parameters; This represents the adjustment coefficient, used to control the degree of influence of the regularization parameter when the click-through rate deviates from the target. This indicates the target click-through rate.

10. A data update method based on a search engine according to claim 9, characterized in that, The feedback optimization model is constructed as follows: Will Defined as the ratio of the current resource usage to the maximum resource usage, expressed as follows: ； in, This indicates the current CPU utilization rate of the system. Indicates the current memory usage; and This indicates the system's maximum CPU and memory capacity; according to Dynamically adjust threshold This ensures that update operations are reduced under high load and increased under low load: ； in, This represents the dynamic update decision threshold used for update decisions, controlling whether a webpage should be updated. This represents the basic update threshold, reflecting the initial value under normal load conditions; This represents the adjustment coefficient, used for control. The degree of influence on the threshold.

Images