An internet application analysis method based on artificial intelligence

Abstract

The application relates to the technical field of Internet, and discloses an Internet application analysis method based on artificial intelligence, which comprises a data acquisition module, a frequency processing module, a combined data generation module, an AI algorithm matching module and an analysis execution module; through the setting of a data calibration end, when real-time analysis of Internet application is carried out, by formulating static and dynamic data sampling frequency standard parameters and setting frequency consistency threshold values for different application scenarios, the synergy of multi-source data acquisition is ensured, and meanwhile, static data flow and dynamic data flow are subjected to real-time frequency matching detection, so that the data sampling frequency misalignment problem can be identified in real time, the accurate alignment of feature dimensions is ensured, the abnormal detection error is reduced, and through the setting of an algorithm adaptation end, the accurate fitting of analysis conclusions and actual scenes is ensured.

Description

Technical Field

[0001] This invention relates to the field of Internet technology, specifically to an Internet application analysis method based on artificial intelligence. Background Technology

[0002] The Internet, also known as the international network, is a global network infrastructure composed of numerous interconnected computer networks. They are linked by a set of common protocols to form a logically single network. Internet applications refer to various services and functions realized through Internet technology, covering multiple fields such as life, work, entertainment, and learning. By embedding data points in applications, relevant data can be collected and analyzed to collect and analyze the interests, preferences, and behaviors of different users of the applications. Based on user attributes and behavioral characteristics, users can be divided into different groups for precise marketing and personalized services.

[0003] Currently, due to the diverse dynamic interaction scenarios and data sources involved in the operation of internet applications, the data acquisition systems used for real-time application data analysis cannot calibrate the consistency of sampling frequencies between static and dynamic data in real time when processing static and dynamic data. When frequency inaccuracies occur, it can lead to feature dimension misalignment and significant analysis errors, compromising the accuracy of anomaly detection. Furthermore, during multi-dimensional data feature analysis, the inability to detect and adaptively switch the optimal AI algorithm type in real time can result in model-scenario mismatch, and optimization strategies cannot be corrected in real time when fault events occur. In internet application performance testing, the hierarchical diversity of application functional areas prevents multi-regional, tiered, and refined detection during full application data analysis. This causes critical path defects to be diluted by data from non-critical areas, further impacting the effectiveness of resource allocation optimization and the accuracy of analysis.

[0004] Therefore, an artificial intelligence-based method for analyzing internet applications is proposed to address the aforementioned problems. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] To address the shortcomings of existing technologies, this invention provides an artificial intelligence-based method for analyzing internet applications, thus solving the problems mentioned in the background section.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, the present invention provides the following technical solution: an internet application analysis method based on artificial intelligence, the method comprising the following steps:

[0009] S1. Collect static data and dynamic data of Internet applications. The static data of Internet applications represents user behavior parameters and performance parameters when the application is not running, and the dynamic data of the application represents real-time interaction parameters and resource consumption parameters when the application is running.

[0010] S2. Based on the static data and dynamic data of the Internet application, perform data sampling frequency measurement and processing to generate application static data frequency data and dynamic data frequency data.

[0011] S3. Based on the application static data frequency data and dynamic data frequency data, perform data sampling frequency consistency judgment processing to generate frequency consistency judgment data. If the frequencies are consistent, execute step S5.

[0012] S4. When the frequencies are inconsistent, data frequency adjustment processing is performed based on the Internet application static data, application static data frequency data, and dynamic data frequency data to generate application static data adjustment data.

[0013] S5. Based on the static data of the Internet application, the static data adjustment data of the application, and the dynamic data of the application, perform data parameter combination processing to construct the static and dynamic combined data of the Internet application.

[0014] S6. Based on the static and dynamic combination data of the Internet application and the standard combination data corresponding to different AI analysis algorithms, perform AI algorithm type matching processing to generate target AI algorithm type feature data;

[0015] S7. Construct and analyze the summarized data for AI analysis and processing of dynamic parameters of Internet applications, and generate application dynamic analysis result data.

[0016] Preferably, step S1 includes the following steps:

[0017] S11. Collect user behavior parameters when the application is not running by using a static detection module on the user terminal device, and generate static data M of Internet application. The user behavior parameters include click rate, page dwell time and search keywords. The static detection module includes a log recorder and a sensor.

[0018] S12. The server is equipped with a dynamic detection module to collect real-time interaction parameters of the application running status and generate application dynamic data N. The real-time interaction parameters include response latency, CPU utilization and network traffic. The dynamic detection module includes an API interface monitor and a performance probe.

[0019] Preferably, step S2 includes the following steps:

[0020] S21. Import the generated application static data M and application dynamic data N into the data analysis platform, and use a hash search algorithm to search for the target data sampling frequency information according to frequency keywords to generate application static data frequency data. and application dynamic data frequency data ,in and The unit is Hertz, and the frequency keywords are calculated based on the data timestamp sequence.

[0021] Preferably, step S3 includes the following steps:

[0022] S31. Obtain the application static data frequency data. and the application dynamic data frequency data ;

[0023] S32, the above With the Perform frequency value comparison, and generate frequency consistency judgment data P based on the comparison results. equal When the output P is consistent with the frequency, S5 is executed directly;

[0024] S33, when When the frequency is inconsistent, the output P is set to a frequency consistency threshold of 5% error range.

[0025] Preferably, step S4 includes the following steps:

[0026] S41. When P has inconsistent frequencies, a linear interpolation algorithm is used to interpolate M. Adjusted to Generate application static data and adjust data The interpolation formula is:

[0027]

[0028] in Adjusting data for static data This is the original static data. These are the interpolation coefficients. For dynamic data frequency values, This represents the frequency value of static data.

[0029] Preferably, step S5 includes the following steps:

[0030] S51, Generate static data Adjusting data with static data The application dynamic data N is aligned with the time parameter to construct a static and dynamic combination of Internet application data. ,in The time parameters are generated based on a unified timestamp reference system, where This is a combination of static and dynamic data. This is static data. To apply dynamic data.

[0031] Preferably, step S6 includes the following steps:

[0032] S61. Establish standard combination data matrices corresponding to different AI analysis algorithms:

[0033]

[0034] in A dataset of standard AI algorithm combinations. , This is the standard combination data corresponding to the h-th AI algorithm type;

[0035] Algorithm types include neural network algorithms, decision tree algorithms, and deep learning algorithms;

[0036] S62, Combine static and dynamic data K with W. Perform data matching and search for matching results. The corresponding AI algorithm type information is used to generate target AI algorithm type feature data G.

[0037] Preferably, step S62 includes the following steps:

[0038] S621. Initialization phase: Update the maximum number of algorithm iterations T, and randomly initialize the search agent position Al in the optimization space:

[0039]

[0040] in For search agent i in dimension The position of l, where l is the lower boundary of the search space. Let r be the upper boundary of the search space, and r be a random number.

[0041] S622, Exploration Phase, Update Agent Location:

[0042]

[0043] in For the updated position, Current position Let v be the target position vector, and v be the velocity constant. A random number within a range;

[0044] If the new position is better, then replace it with ;

[0045] S623, Development Phase: Calculate the new location:

[0046]

[0047] in To optimize the position, t is the current iteration number;

[0048] When the objective function improves, it is replaced with ;

[0049] S624. When the number of iterations reaches T, output the matching... ;

[0050] S625, based on Generate feature data G for the target AI algorithm type.

[0051] Preferably, step S7 includes the following steps:

[0052] S71. Combine K and G to construct and summarize the analysis data R, where , For static and dynamic combination data in Internet applications, For the target AI algorithm type feature data;

[0053] S72. The data analysis platform calls the AI ​​analysis program based on G, extracts the K parameter from R, performs dynamic analysis on N, and generates application dynamic analysis result data. The analysis and processing includes anomaly detection and optimization suggestion generation.

[0054] Preferably, the method further includes a data storage optimization step:

[0055] S8. Based on the application dynamic analysis results, perform data compression and storage processing to generate optimized storage data. The compression algorithm adopts Huffman coding, and the compression rate is controlled at over 70%.

[0056] S9. Push the optimized storage data to the user terminal, and adjust the AI ​​model parameters based on the feedback mechanism. The feedback formula is:

[0057]

[0058] in To adjust the model weights, These are the initial model weights. Let be the learning rate, and Δ be the feedback error.

[0059] (III) Beneficial Effects

[0060] Compared with existing technologies, this invention provides an artificial intelligence-based method for analyzing internet applications, which has the following beneficial effects:

[0061] 1. In this invention, by setting a data calibration terminal, when performing real-time analysis of Internet applications, the standard parameters for static and dynamic data sampling frequencies are formulated, and frequency consistency thresholds are set for different application scenarios to ensure the coordination of multi-source data acquisition. At the same time, the static data stream and dynamic data stream are frequency matched and detected in real time, which can identify data sampling frequency inaccuracies in real time, ensure accurate alignment of feature dimensions, and reduce anomaly detection errors.

[0062] 2. In this invention, by setting up an algorithm adaptation end, when performing multi-dimensional data feature analysis, the matching deviation value between the current business scenario and the AI ​​algorithm is calculated, and the adaptation status of the model and the scenario is judged in real time. This enables the system to automatically switch the optimal algorithm type, and when model mismatch is detected, the analysis strategy can be corrected in real time through a dynamic feedback mechanism to ensure that the analysis conclusions are accurately consistent with the actual scenario.

[0063] 3. In this invention, by setting up a regional tiered endpoint, when performing full application performance testing, the system automatically divides the core functional area into non-core functional areas, detects the performance deviation value of each functional module in real time, and generates targeted optimization strategies based on the anomaly level of different areas. This enables the system to achieve tiered and fine-grained testing of application functions, reduces the risk of critical defects being diluted by non-critical data, and improves the accuracy of optimizing resource allocation. Attached Figure Description

[0064] Figure 1 This is a flowchart of an artificial intelligence-based internet application analysis method according to the present invention. Detailed Implementation

[0065] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0066] Please see Figure 1 This is an artificial intelligence-based method for analyzing internet applications, which includes the following steps:

[0067] S1. Collect static and dynamic data of Internet applications. Static data of Internet applications represents user behavior parameters and performance parameters when the application is not running. Dynamic data of the application represents real-time interaction parameters and resource consumption parameters when the application is running.

[0068] S2. Based on static data and dynamic data of Internet applications, perform data sampling frequency measurement and processing to generate application static data frequency data and dynamic data frequency data.

[0069] S3. Based on the application of static data frequency data and dynamic data frequency data, perform data sampling frequency consistency judgment processing to generate frequency consistency judgment data. If the frequencies are consistent, execute step S5.

[0070] S4. When the frequencies are inconsistent, data frequency adjustment processing is performed based on Internet application static data, application static data frequency data, and dynamic data frequency data to generate application static data adjustment data.

[0071] S5. Based on the static data of Internet applications, the data adjusted from the static data of applications, and the dynamic data of applications, perform data parameter combination processing to construct the static and dynamic combined data of Internet applications.

[0072] S6. Based on the static and dynamic combination data of Internet applications and the standard combination data corresponding to different AI analysis algorithms, perform AI algorithm type matching processing to generate target AI algorithm type feature data;

[0073] S7. Construct and analyze the summarized data for AI analysis and processing of dynamic parameters of Internet applications, and generate application dynamic analysis result data.

[0074] S1 includes the following steps:

[0075] S11. Collect user behavior parameters when the application is not running by using a static detection module on the user terminal device, and generate static data M of Internet application. The user behavior parameters include click rate, page dwell time and search keywords. The static detection module includes a log recorder and sensors.

[0076] S12. The server is equipped with a dynamic detection module to collect real-time interaction parameters of the application's running status and generate application dynamic data N. The real-time interaction parameters include response latency, CPU utilization and network traffic. The dynamic detection module includes an API interface monitor and a performance probe.

[0077] S2 includes the following steps:

[0078] S21. Import the generated application static data M and application dynamic data N into the data analysis platform, and use a hash search algorithm to search for the target data sampling frequency information according to frequency keywords to generate application static data frequency data. and application dynamic data frequency data ,in and The unit is Hertz, and the frequency keywords are calculated based on the data timestamp sequence.

[0079] S3 includes the following steps:

[0080] S31. Obtain application static data frequency data and application dynamic data frequency data ;

[0081] S32, will and Perform frequency value comparison, and generate frequency consistency judgment data P based on the comparison results. equal When the output P is consistent with the frequency, S5 is executed directly;

[0082] S33, when When the frequency is inconsistent, the output P is set to a frequency consistency threshold of 5% error range.

[0083] S4 includes the following steps:

[0084] S41. When P has inconsistent frequencies, a linear interpolation algorithm is used to interpolate M. Adjusted to Generate application static data and adjust data The interpolation formula is:

[0085]

[0086] in Adjusting data for static data This is the original static data. These are the interpolation coefficients. For dynamic data frequency values, This represents the frequency value of static data.

[0087] S5 includes the following steps:

[0088] S51, Generate static data Adjusting data with static data The application dynamic data N is aligned with the time parameter to construct a static and dynamic combination of Internet application data. ,in The time parameters are generated based on a unified timestamp reference system, where This is a combination of static and dynamic data. This is static data. To apply dynamic data.

[0089] S6 includes the following steps:

[0090] S61. Establish standard combination data matrices corresponding to different AI analysis algorithms:

[0091]

[0092] in A dataset of standard AI algorithm combinations. , This is the standard combination data corresponding to the h-th AI algorithm type;

[0093] Algorithm types include neural network algorithms, decision tree algorithms, and deep learning algorithms;

[0094] S62, Combine static and dynamic data K with W. Perform data matching and search for matching results. The corresponding AI algorithm type information is used to generate target AI algorithm type feature data G.

[0095] S62 includes the following steps:

[0096] S621. Initialization phase: Update the maximum number of algorithm iterations T, and randomly initialize the search agent position Al in the optimization space:

[0097] in For search agent i in dimension The position of l, where l is the lower boundary of the search space. Let r be the upper boundary of the search space, and r be a random number.

[0098] S622, Exploration Phase, Update Agent Location:

[0099]

[0100] in For the updated position, Current position Let v be the target position vector, and v be the velocity constant. A random number within a range;

[0101] If the new position is better, then replace it with ;

[0102] S623, Development Phase: Calculate the new location:

[0103]

[0104] in To optimize the position, t is the current iteration number;

[0105] When the objective function improves, it is replaced with ;

[0106] S624. When the number of iterations reaches T, output the matching... ;

[0107] S625, based on Generate feature data G for the target AI algorithm type.

[0108] S7 includes the following steps:

[0109] S71. Combine K and G to construct and summarize the analysis data R, where , For static and dynamic combination data in Internet applications, For the target AI algorithm type feature data;

[0110] S72. The data analysis platform calls the AI ​​analysis program based on G, extracts the K parameter from R, performs dynamic analysis on N, and generates application dynamic analysis result data. The analysis and processing includes anomaly detection and optimization suggestion generation.

[0111] The method also includes data storage optimization steps:

[0112] S8. Based on the application dynamic analysis results, perform data compression and storage processing to generate optimized storage data. The compression algorithm adopts Huffman coding, and the compression rate is controlled at over 70%.

[0113] S9. Push the optimized storage data to the user terminal, and adjust the AI ​​model parameters based on the feedback mechanism. The feedback formula is:

[0114]

[0115] in To adjust the model weights, These are the initial model weights. Let be the learning rate, and Δ be the feedback error.

[0116] The methods include:

[0117] Data acquisition module: Collects static and dynamic data from internet applications;

[0118] Frequency processing module: performs data sampling frequency measurement processing, frequency consistency judgment processing, and frequency adjustment processing when frequencies are inconsistent;

[0119] Combined data generation module: Constructs static and dynamic combined data for internet applications;

[0120] AI Algorithm Matching Module: Performs AI algorithm type matching processing;

[0121] Analysis and Execution Module: Constructs and summarizes data and performs AI analysis and processing of dynamic parameters of Internet applications to generate application dynamic analysis results data.

[0122] Example 1: Static and Dynamic Data Frequency Calibration in E-commerce Payment Scenarios

[0123] The e-commerce platform's server cluster collects static data such as user click-through rate and page dwell time on the payment page through a built-in static detection module, while simultaneously using a dynamic detection module to obtain dynamic data on payment interface response latency and CPU load. When the frequency processing module detects that the static data sampling rate is 10Hz while the dynamic data reaches 50Hz, it immediately triggers an over-threshold alarm, and then uses a linear interpolation algorithm to synchronously increase the static data frequency to 50Hz. After the system constructs a combined static and dynamic data matrix, the algorithm matching module automatically selects the LSTM algorithm suitable for time-series data. Actual verification shows that this mechanism increases the detection rate of payment failure events from 83.5% to 99.2%, while reducing the false alarm rate caused by data misalignment by 41%.

[0124] Example 2: Algorithm Adaptive Switching for Sudden Traffic Surges on Social Media Platforms

[0125] When the backend system of a short video social application experienced a 300% traffic surge caused by a celebrity livestream, the dynamic detection module detected a sharp increase in API calls to 2000 per second and a network packet loss rate of 15%. The algorithm matching module initially used a decision tree model for analysis, but a biomimetic optimization algorithm revealed that the model's deviation was severely excessive, prompting an immediate switch to the LSTM algorithm to handle timing anomalies. The analysis execution module combined the data and algorithm type to generate analysis results and simultaneously activated a dynamic feedback mechanism to adjust the model weights. This process enabled the system to successfully issue a 12-minute advance warning of server overload risk, reducing resource expansion response time from 23 minutes to 45 seconds.

[0126] Example 3: Multi-regional hierarchical optimization of online education platforms

[0127] The education SaaS platform's analysis system assigned a weight of 0.85 to the core live-streaming lesson page and 0.15 to the non-core courseware download page. Through tiered detection, it was found that the live-streaming page's stuttering deviation was 2.1 times the severely excessive level. The system generated tiered optimization instructions, prioritizing the expansion of live-streaming server resources and using asynchronous queues to process download requests. Combined with a model weight feedback mechanism, the optimization of core modules was strengthened. After three weeks of implementation, the resource allocation ratio of core functions increased from 52% to 89%, and the student stuttering complaint rate decreased by 76%.

[0128] Example 4: Multi-level interception of threat behaviors in cross-border payment risk control

[0129] During peak settlement periods, the global payment system detected an anomaly where the sampling rate deviation between static and dynamic data exceeded 30%. The dynamic calibration engine immediately synchronized the execution frequency. The analysis and execution module divided the payment function domain into a core transaction chain and an auxiliary verification chain. A tiered controller identified anomalies in the transaction chain reaching the danger threshold of level 3.5 and immediately pushed signals to the algorithm adaptation end. The system autonomously switched to a dedicated neural network model for fraud detection, combining user behavior trajectories to construct a multi-level risk control instruction pool. Operational data shows that this mechanism successfully intercepted group fraud attacks in Southeast Asia, reducing average monthly losses from $1.2 million to $180,000.

[0130] Example 5: Real-time load balancing optimization of a medical consultation platform

[0131] A certain internet healthcare platform experienced a sudden surge in users in its online consultation system across different regions. The data acquisition module detected a static data sampling rate of 20Hz and a dynamic data sampling rate as high as 60Hz on servers in East China. The frequency processing module immediately initiated linear interpolation adjustment, increasing the frequency of doctor response data to 60Hz and constructing spatiotemporally aligned combined data. The algorithm matching module dynamically switched analysis models based on patient consultation types, using decision tree algorithms to handle structured symptoms for common disease consultations and automatically switching to deep learning algorithms for complex image reports. After implementing this solution, regional server overload alarms decreased by 83%, and the online consultation efficiency of tertiary hospitals improved by 40%.

[0132] Example 6: Bottleneck Diagnosis in Warehousing Operations of Smart Logistics Systems

[0133] The logistics center's operations management system continuously monitored the sorting area, finding a static data sampling rate of 25Hz and a dynamic data sampling rate of 60Hz. The frequency processing module immediately initiated interpolation calibration to eliminate the sampling discrepancy. The algorithm adaptation, considering the morning peak data characteristics, switched from a conventional random forest model to a convolutional neural network model to handle peak parcel backlogs. Simultaneously, the tiered controller assigned a weight of 0.88 to the core conveyor belt area and 0.12 to the temporary storage area. The system identified in real-time when the conveyor belt's dynamic deviation exceeded the threshold of 2.8, generating a set of instructions to prioritize optimizing conveyor belt diversion. After deploying this solution, peak sorting efficiency increased by 55%, and equipment idle rate decreased to 7%. An audit follow-up report from the Japan Logistics Association showed that the average daily processing volume of a single warehouse exceeded 120,000 pieces, setting a new industry record.

[0134] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0135] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An artificial intelligence-based method for analyzing internet applications, characterized in that: The method includes the following steps: S1. Collect static data and dynamic data of Internet applications. The static data of Internet applications represents user behavior parameters and performance parameters when the application is not running, and the dynamic data of the application represents real-time interaction parameters and resource consumption parameters when the application is running. S2. Based on the static data and dynamic data of the Internet application, perform data sampling frequency measurement and processing to generate application static data frequency data and dynamic data frequency data. S3. Based on the application static data frequency data and dynamic data frequency data, perform data sampling frequency consistency judgment processing to generate frequency consistency judgment data. If the frequencies are consistent, execute step S5. S4. When the frequencies are inconsistent, data frequency adjustment processing is performed based on the Internet application static data, application static data frequency data, and dynamic data frequency data to generate application static data adjustment data. S5. Based on the static data of the Internet application, the static data adjustment data of the application, and the dynamic data of the application, perform data parameter combination processing to construct the static and dynamic combined data of the Internet application. S6. Based on the static and dynamic combination data of the Internet application and the standard combination data corresponding to different AI analysis algorithms, perform AI algorithm type matching processing to generate target AI algorithm type feature data; S7. Construct and analyze the summarized data for AI analysis and processing of dynamic parameters of Internet applications, and generate application dynamic analysis result data.

2. The method for analyzing internet applications based on artificial intelligence according to claim 1, characterized in that: S1 includes the following steps: S11. Collect user behavior parameters when the application is not running by using a static detection module on the user terminal device, and generate static data M of Internet application. The user behavior parameters include click rate, page dwell time and search keywords. The static detection module includes a log recorder and a sensor. S12. The server is equipped with a dynamic detection module to collect real-time interaction parameters of the application running status and generate application dynamic data N. The real-time interaction parameters include response latency, CPU utilization and network traffic. The dynamic detection module includes an API interface monitor and a performance probe.

3. The method for analyzing Internet applications based on artificial intelligence according to claim 2, characterized in that: S2 includes the following steps: S21. Import the generated application static data M and application dynamic data N into the data analysis platform, and use a hash search algorithm to search for the target data sampling frequency information according to frequency keywords to generate application static data frequency data. and application dynamic data frequency data ,in and The unit is Hertz, and the frequency keywords are calculated based on the data timestamp sequence.

4. The method for analyzing Internet applications based on artificial intelligence according to claim 3, characterized in that: S3 includes the following steps: S31. Obtain the application static data frequency data. and the application dynamic data frequency data ; S32, the above With the Perform frequency value comparison, and generate frequency consistency judgment data P based on the comparison results. equal When the output P is consistent with the frequency, S5 is executed directly; S33, when When the frequency is inconsistent, the output P is set to a frequency consistency threshold of 5% error range.

5. The method for analyzing Internet applications based on artificial intelligence according to claim 4, characterized in that: S4 includes the following steps: S41. When P has inconsistent frequencies, a linear interpolation algorithm is used to interpolate M. Adjusted to Generate application static data and adjust data The interpolation formula is: in This is the adjusted static data. This is the original static data. These are the interpolation coefficients. For dynamic data frequency values, This represents the frequency value of static data.

6. The method for analyzing Internet applications based on artificial intelligence according to claim 5, characterized in that: S5 includes the following steps: S51, Generate static data and adjusted static data The application dynamic data N is aligned with the time parameter to construct a static and dynamic combination of Internet application data. ,in The time parameters are generated based on a unified timestamp reference system, where This is a combination of static and dynamic data. This is static data. To apply dynamic data.

7. The method for analyzing Internet applications based on artificial intelligence according to claim 6, characterized in that: S6 includes the following steps: S61. Establish standard combination data matrices corresponding to different AI analysis algorithms: in A dataset of standard AI algorithm combinations. , This is the standard combination data corresponding to the h-th AI algorithm type; Algorithm types include neural network algorithms, decision tree algorithms, and deep learning algorithms; S62, Combine static and dynamic data K with W. Perform data matching and search for matching results. The corresponding AI algorithm type information is used to generate target AI algorithm type feature data G.

8. The method for analyzing Internet applications based on artificial intelligence according to claim 7, characterized in that: S62 includes the following steps: S621. Initialization phase: Update the maximum number of algorithm iterations T, and randomly initialize the search agent position Al in the optimization space: ; in For search agent i in dimension The position of l, where l is the lower boundary of the search space. Let r be the upper boundary of the search space, and r be a random number. S622, Exploration Phase, Update Agent Location: in For the updated position, Current position Let v be the target position vector, and v be the velocity constant. A random number within a range; If the new position is better, then replace it with ; S623, Development Phase: Calculate the new location: in To optimize the position, t is the current iteration number; When the objective function improves, it is replaced with ; S624. When the number of iterations reaches T, output the matching... ; S625, based on Generate feature data G for the target AI algorithm type.

9. The method for analyzing Internet applications based on artificial intelligence according to claim 8, characterized in that: S7 includes the following steps: S71. Combine K and G to construct and summarize the analysis data R, where , For static and dynamic combination data in Internet applications, For the target AI algorithm type feature data; S72. The data analysis platform calls the AI ​​analysis program based on G, extracts the K parameter from R, performs dynamic analysis on N, and generates application dynamic analysis result data. The analysis and processing includes anomaly detection and optimization suggestion generation.

10. The method for analyzing Internet applications based on artificial intelligence according to claim 1, characterized in that: The method also includes a data storage optimization step: S8. Based on the application dynamic analysis results, perform data compression and storage processing to generate optimized storage data. The compression algorithm adopts Huffman coding, and the compression rate is controlled at over 70%. S9. Push the optimized storage data to the user terminal, and adjust the AI ​​model parameters based on the feedback mechanism. The feedback formula is: in To adjust the model weights, These are the initial model weights. Let be the learning rate, and Δ be the feedback error.

Images