Intelligent classification and accurate recommendation system and method for education resources based on artificial intelligence

Abstract

The application relates to the technical field of classified data processing, and particularly discloses an intelligent classification and accurate recommendation system and method for educational resources based on artificial intelligence, which extracts resource features by artificial intelligence driving, monitors abnormalities in real time and intelligently adjusts in the process, ensures high-precision and stable classification results, and then the classified data enters a storage module, the module monitors the storage process, actively discovers and optimizes storage abnormalities, and guarantees that educational resources are efficiently and completely stored in a database; finally, a recommendation module accurately matches high-quality pre-classified resources based on user behavior characteristics, and realizes highly personalized intelligent recommendation by relying on the structured data basis provided by the previous module; the whole process operates in a closed loop and is dynamically optimized, which significantly improves resource management efficiency, data quality and user experience.

Description

Technical Field

[0001] This invention relates to the field of classification data processing technology, specifically to an intelligent classification and precise recommendation system and method for educational resources based on artificial intelligence. Background Technology

[0002] The AI-powered educational resource classification system preprocesses educational resources, utilizing AI technology to extract deep features and identify multi-dimensional characteristics such as content themes, knowledge domains, difficulty levels, skill types, emotional tendencies, and media elements. Based on these features, the system applies an intelligent classification model to automatically, precisely, and with multiple labels classify the resources, constructing a structured knowledge resource graph. The system also builds user profiles, continuously analyzing users' learning foundations, interests, knowledge mastery, learning styles, goal progress, and potential weaknesses through explicit and implicit data. Combining the resource feature graph and dynamically updated user profiles, the system employs a hybrid recommendation algorithm for precise matching.

[0003] For example, Chinese invention patent CN119807486A discloses an artificial intelligence-based medical education guidance method and system. The method includes the following steps: setting up a medical education resource library in a cloud data center and constructing an artificial intelligence model; extracting and classifying features from several medical education resource data in the library; generating a user profile based on the user's real-time basic information and real-time behavioral data; matching the real-time user profile with several real-time resource classification results; providing medical education guidance based on the real-time user profile and the features of several matched real-time resource data; sending the real-time medical education strategy and several target medical education resource data to the user terminal; and optimizing the parameters of the artificial intelligence model based on the real-time feedback data sent by the user terminal.

[0004] For example, Chinese invention patent CN116028501A discloses a blockchain-based artificial intelligence data sharing method and system, which relates to the field of data processing technology. It addresses the problem that existing educational resource sharing systems fail to classify and store shared data, leading to server malfunctions as the space occupied by shared data and the number of users increasing. Furthermore, these systems cannot intelligently display shared data to users. This method analyzes shared data to obtain shared values, uses these values ​​to divide the data, and then classifies and stores the divided data. This ensures full utilization of storage space, prevents interference between shared data, reduces the probability of data corruption, maintains sufficient storage space, and displays shared data based on shared values ​​or display coefficients.

[0005] The aforementioned technologies suffer from at least the following technical problems: Existing storage systems exhibit severely insufficient performance and scalability, leading to I / O bottlenecks, response latency, or potential failure risks under high loads. When facing the unique challenges of massive, multi-source, high-concurrency, and real-time data requirements inherent in intelligent classification and recommendation systems for educational resources, these technologies have become key bottlenecks restricting overall system performance, user experience, service reliability, and scalability, resulting in significant operational complexity and potential risks. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides an intelligent classification and precise recommendation system and method for educational resources based on artificial intelligence, which can effectively solve the problems mentioned in the background technology.

[0007] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of the present invention provides an intelligent classification and precise recommendation system for educational resources based on artificial intelligence, comprising: an educational resource classification module, used to extract features from educational resources based on artificial intelligence, monitor the feature extraction process, collect and analyze abnormal parameters of the feature extraction process, determine whether to adjust the feature extraction process of educational resources, thereby intelligently classifying and storing the educational resources in the education area of ​​the database; an educational resource storage module, used to monitor the storage process of educational resources classified and stored in the education area of ​​the database, acquire and parse abnormal parameters of the classification storage process of educational resources, thereby determining whether to optimize the classification storage process of educational resources, until all educational resources are classified and stored in the storage area of ​​the database; and an educational resource recommendation module, used to extract feature information of user behavior, compare the feature information with the classification data of each category in the educational resources, thereby making precise recommendations for educational resources.

[0008] The second aspect of this invention provides an artificial intelligence-based method for intelligent classification and precise recommendation of educational resources, comprising: Step 1, extracting features from educational resources based on artificial intelligence, monitoring the feature extraction process, collecting and analyzing abnormal parameters in the feature extraction process, determining whether to adjust the feature extraction process of educational resources, thereby intelligently classifying the educational resources and storing them in the education area of ​​the database; Step 2, monitoring the storage process of educational resources in the education area of ​​the database, acquiring and parsing abnormal parameters in the classification storage process of educational resources, thereby determining whether to optimize the classification storage process of educational resources, until all educational resources are classified and stored in the storage area of ​​the database; Step 3, extracting feature information of user behavior, comparing the feature information with the classification data of each category in the educational resources, thereby making precise recommendations for educational resources.

[0009] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects:

[0010] (1) This invention provides an intelligent classification and precise recommendation system and method for educational resources based on artificial intelligence. This system uses artificial intelligence to drive resource feature extraction (classification module), monitoring anomalies in real time and making intelligent adjustments to ensure high-precision and stable classification results. This classified data then enters a storage module, which monitors the data entry process, proactively identifying and optimizing storage anomalies to ensure efficient and complete storage of educational resources in the database. Finally, the recommendation module accurately matches pre-classified high-quality resources based on user behavior characteristics, leveraging the structured data provided by the preceding modules to achieve highly personalized intelligent recommendations. The entire process operates in a closed loop and is dynamically optimized, significantly improving resource management efficiency, data quality, and user experience.

[0011] (2) The core advantage of the educational resource classification module lies in its use of artificial intelligence for automated feature extraction, and its proactive collection and analysis of abnormal parameters through real-time monitoring of the feature extraction process. Its key value lies in its ability to intelligently judge and dynamically adjust the extraction process, thereby significantly improving the accuracy and stability of educational resource feature recognition. Ultimately, it produces high-quality, structured classification data, laying a solid and reliable foundation for subsequent stages of the entire system.

[0012] (3) The core advantage of the educational resource storage module lies in its continuous monitoring of the process of classifying and storing educational resources. It can proactively collect abnormal parameters during the storage process and intelligently determine whether and how to optimize the storage. Its key value lies in ensuring that massive amounts of educational resources can be stored in the database efficiently, securely, and completely. Through this closed-loop anomaly management and optimization mechanism, problems such as data loss, misalignment, or low storage efficiency are effectively prevented, ensuring the reliability, accessibility, and data integrity of the educational area of ​​the database.

[0013] (4) The core advantage of the educational resource recommendation module lies in its ability to accurately extract and analyze the characteristic information of user behavior and intelligently compare and match it with the structured educational resource data produced by the classification module. Its key value lies in achieving highly personalized and accurate recommendations of educational resources. By utilizing the high-quality and organized data foundation provided by the preceding modules (classification and storage), this module breaks through the limitations of traditional recommendations, can more accurately understand user needs and match the most suitable resources, thereby significantly improving user experience, resource utilization efficiency and user satisfaction. Attached Figure Description

[0014] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0015] Figure 1This is a schematic diagram of the system module connections of the present invention.

[0016] Figure 2 This is a schematic diagram of the method steps of the present invention.

[0017] Figure 3 This is a flowchart illustrating the classification of educational resources according to the present invention.

[0018] Figure 4 This is a flowchart of the educational resource storage process of the present invention.

[0019] Figure 5 This is a flowchart illustrating the educational resource recommendation process of the present invention. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0021] Reference Figure 1 As shown, the first aspect of the present invention provides an intelligent classification and precise recommendation system for educational resources based on artificial intelligence, including: an educational resource classification module, an educational resource storage module, an educational resource recommendation module, and a database.

[0022] The database is used to store parameters involved in the AI-based intelligent classification and precise recommendation system for educational resources.

[0023] The educational resource classification module is connected to the educational resource storage module, which in turn is connected to the educational resource recommendation module. All three modules are connected to the database.

[0024] The education resource classification module is used to extract features from education resources based on artificial intelligence, monitor the feature extraction process, collect and analyze abnormal parameters in the feature extraction process, determine whether to adjust the feature extraction process, and thus intelligently classify and store the education resources in the education area of ​​the database.

[0025] Specifically, the process of determining whether to adjust the feature extraction process of educational resources involves comparing the feature extraction anomaly index with the feature extraction anomaly threshold stored in the database. If the feature extraction anomaly index is less than or equal to the feature extraction anomaly threshold, it is determined that no adjustment will be made to the processing device to which the feature extraction of educational resources belongs, thereby intelligently classifying the educational resources and storing them in the education area of ​​the database.

[0026] Figure 3 The flowchart for the educational resource classification of this invention utilizes artificial intelligence technology to automatically extract features from educational resources and monitors relevant parameters in real time during the extraction process. By collecting and analyzing anomalies in these parameters, it is determined whether the feature extraction process needs to be dynamically adjusted. Based on the adjusted features, the educational resources are intelligently classified and stored in the corresponding classification areas of the database.

[0027] If the feature extraction anomaly index is greater than the feature extraction anomaly threshold, the educational resources are stored in the temporary storage area of ​​the database, and the initial learning rate of the AI ​​model is reduced based on the feature extraction anomaly index, thereby optimizing the learning behavior of the AI ​​model itself.

[0028] The educational resources in the temporary storage area of ​​the database are re-extracted using the optimized AI model to obtain a secondary feature extraction anomaly index. The secondary feature extraction anomaly index is compared with the feature extraction anomaly threshold. If the secondary feature extraction anomaly index is less than or equal to the feature extraction anomaly threshold, the educational resources are intelligently classified and stored in the education area of ​​the database.

[0029] If the secondary feature extraction anomaly index is greater than the feature extraction anomaly threshold, it is determined that the feature extraction process of educational resources should be adjusted a second time.

[0030] The feature extraction anomaly threshold is used to characterize the upper limit of the feature extraction anomaly index.

[0031] The initial learning rate of the AI ​​model is reduced based on the feature extraction anomaly index. In the database, the feature extraction anomaly index is divided into different gradients, and each gradient corresponds to a different amount of learning rate reduction. The current learning rate is subtracted from the learning rate reduction to obtain the next learning rate.

[0032] Addressing initial feature extraction anomalies is crucial. A high feature extraction anomaly index indicates significant problems with the model's handling of current educational resources; this is typically related to unstable learning behavior or maladaptation to the current data. Adjusting the learning rate optimizes the model's learning behavior, making it more stable and convergent during secondary processing of temporary educational resources. This significantly improves the success rate of secondary feature extraction. The increased success rate directly reduces repeated transmission of educational resources between the processing device and the database temporary storage area due to multiple extraction failures, saving network bandwidth. Simultaneously, it reduces CPU resource consumption caused by task failures, retry scheduling, and invalid computations, optimizing CPU utilization and ultimately lowering the feature extraction anomaly index.

[0033] This mechanism significantly improves the efficiency of educational resource processing and the accuracy of classification through intelligent anomaly detection and dynamic adjustment. First, it accurately identifies feature extraction anomalies using preset thresholds, avoiding unnecessary equipment adjustments and ensuring that qualified resources are efficiently classified and stored. For anomalous resources, they are temporarily stored while the AI ​​model's learning behavior is proactively optimized. This targeted model adjustment enhances the robustness of subsequent feature extraction. The optimized model then performs secondary processing on the anomalous resources, successfully reducing the anomaly index and achieving effective classification. The entire process forms a closed loop of detection, optimization, reprocessing, and verification. This not only minimizes processing delays for qualified resources and ensures data quality in core educational areas, but more importantly, it drives model self-optimization through intelligent feedback, effectively improving the system's ability to handle complex or anomalous educational resources. This enhances the adaptability and reliability of the overall classification system, ultimately ensuring the accuracy and usability of the features of the stored educational resources.

[0034] Specifically, the judgment is to make secondary adjustments to the feature extraction process of educational resources. The specific adjustment process is as follows: educational resources are classified and stored in the temporary storage area of ​​the database. Then, based on the anomaly index of secondary feature extraction, the initial learning rate of the AI ​​model is reduced and the noise intensity of the input data is increased.

[0035] After the second adjustment is completed, the educational resources in the temporary storage area of ​​the database are re-extracted using the optimized AI model. The parameters of the feature extraction process are re-monitored, and the three feature extraction anomaly indices are compared with the feature extraction anomaly threshold. If the three feature extraction anomaly indices are less than or equal to the feature extraction anomaly threshold, the educational resources are intelligently classified and stored in the education area of ​​the database.

[0036] If the abnormal index of three feature extractions exceeds the abnormal threshold of feature extraction, an early warning will be issued and manual intervention will be carried out.

[0037] The core purpose of reducing the initial learning rate and increasing the noise intensity of the input data is to optimize the efficiency and resource consumption of the feature extraction process to cope with the high anomaly index of the initial feature extraction. Reducing the learning rate stabilizes the model learning process, reduces oscillations and invalid iterations, thereby shortening the total feature extraction time and reducing CPU utilization. Increasing the noise improves the model's robustness, prevents overfitting, and enables it to learn more essential features. This also helps to accelerate convergence, reduce the additional computational overhead caused by data sensitivity (reducing CPU utilization), and indirectly reduce network bandwidth consumption caused by retries or data transmission. This synergistic adjustment ultimately improves the stability, efficiency, and success rate of feature extraction, ensuring that educational resources can be efficiently and accurately classified and stored, thereby reducing the anomaly index of the third feature extraction.

[0038] The core advantage of categorized storage lies in its introduction of category-level management granularity. This not only provides robust traceability and problem isolation capabilities in the event of anomalies, but more importantly, it lays a solid foundation for implementing precise and differentiated AI model tuning strategies and subsequent processing procedures. This significantly improves the efficiency, accuracy, and controllability of the entire educational resource feature extraction, quality control, and intelligent classification process.

[0039] The anomalous feature extraction index reduces the initial learning rate of the AI ​​model and increases the noise intensity of the input data. In the database, the anomalous feature extraction index is divided into different gradients, each gradient corresponding to a different reduction in the learning rate. The current learning rate minus the reduction in the learning rate yields the learning rate for the next step. Similarly, the anomalous feature extraction index is divided into different gradients, each gradient corresponding to a different increase in noise intensity. The current noise intensity plus the increase in noise intensity yields the noise intensity for the next step.

[0040] This process significantly improves the robustness and stability of feature extraction by introducing a dynamic adjustment mechanism based on a secondary feature extraction anomaly index, effectively suppressing abnormal fluctuations. Subsequently, the optimized model is used to re-extract features in the temporary storage area and undergo rigorous three-stage anomaly index threshold comparison, ensuring the quality and reliability of the final database features. This closed-loop, step-by-step optimization and verification design not only greatly improves the accuracy and reliability of intelligent classification of educational resources and reduces the risk of misclassification, but also, through a clear early warning mechanism, precisely targets manual intervention to truly problematic resources. This significantly enhances the efficiency and intelligence of the entire educational resource database entry and management process, ensuring the high quality and reliability of the educational content in the database.

[0041] Furthermore, abnormal parameters of the feature extraction process are collected and analyzed. Specifically, the abnormal parameters of the feature extraction process include the total duration of the feature extraction process, the network bandwidth utilization of the processing device to which the feature extraction is performed, and the CPU utilization of the processing device to which the feature extraction is performed.

[0042] The feature extraction anomaly index is used to quantitatively assess the degree of abnormality in equipment resource consumption during the execution of feature extraction tasks.

[0043] The total duration of the feature extraction process is measured. High-precision timestamps are recorded at task start and end (using `time.perf_counter()`), and the time difference is calculated to obtain the complete processing time. For the network bandwidth utilization of the processing device involved in feature extraction, the initial number of bytes sent and received by the network interface is obtained using the system tool (psutil) before the task starts, and the cumulative number of bytes is read again after the task ends. Combined with the measured total duration, the average bandwidth is calculated as ((Δ bytes × 8) / (duration × 10)). 6The utilization percentage is obtained by dividing the maximum bandwidth of the interface by the CPU utilization of the feature extraction processing device. The CPU utilization of the processing device is obtained by sampling the CPU utilization of the processing device at fixed intervals (such as per second) during the feature extraction execution (using psutil.cpu_percent()), and finally taking the arithmetic mean of all sampled values ​​as the overall utilization.

[0044] The initial number of bytes sent and received usually refers to the initial receive window size negotiated or announced by both parties in network communication, especially when establishing a connection using the TCP protocol.

[0045] This index calculates the deviation rates of the total feature extraction process time, the network bandwidth utilization rate of the feature extraction processing device, and the CPU utilization rate of the feature extraction processing device from preset reference total feature extraction process time, reference network bandwidth utilization rate, and reference CPU utilization rate in the database. Based on the importance of the total feature extraction process time, the network bandwidth utilization rate of the feature extraction processing device, and the CPU utilization rate of the feature extraction processing device to the anomaly, preset measurement ratios in the database are assigned to each deviation rate. Finally, the three summarized deviation rates are summed to obtain the feature extraction anomaly index.

[0046]

[0047] TYC is the feature extraction anomaly index, CLS is the total duration of the feature extraction process, TNZ is the network bandwidth utilization of the processing device to which the feature extraction is performed, CPL is the CPU utilization of the processing device to which the feature extraction is performed, CLS_L is the preset reference total duration of the feature extraction process in the database, TNZ_L is the preset reference network bandwidth utilization in the database, CPL_L is the preset reference CPU utilization in the database, A1 is the measurement ratio corresponding to the preset total duration in the database, A2 is the measurement ratio corresponding to the preset network bandwidth utilization in the database, and A3 is the measurement ratio corresponding to the preset CPU utilization in the database.

[0048] The reference feature extraction process total time is a reference value used to characterize the total time of the feature extraction process; the reference network bandwidth utilization rate is a reference value used to characterize the network bandwidth utilization rate of the processing device to which the feature extraction belongs; and the reference CPU utilization rate is a reference value used to characterize the CPU utilization rate of the processing device to which the feature extraction belongs.

[0049] The efficiency of the coordinated operation of the total feature extraction process duration, the network bandwidth utilization of the feature extraction processing device, and the CPU utilization of the feature extraction processing device is primarily reflected in the deviations of these parameters relative to reference values. Ideally, the reference network bandwidth utilization and reference CPU utilization should precisely match to minimize the total duration. However, deviations are common in practice: if the network bandwidth utilization of the feature extraction processing device is consistently higher than the reference network bandwidth utilization range, it indicates that transmission capacity has become a bottleneck, and the CPU utilization of the feature extraction processing device will fall below the reference CPU utilization due to waiting for data, leading to an increase in total duration. Conversely, if the deviation of the CPU utilization of the feature extraction processing device consistently approaches or exceeds the reference CPU utilization, it reveals insufficient computing power. In this case, the network bandwidth utilization of the feature extraction processing device will be lower than the reference network bandwidth utilization, similarly extending the total duration. Therefore, the degradation of total duration is essentially a direct result of resource idleness (one side underutilization) or capacity saturation (the other side overutilization) caused by the deviation of the network bandwidth utilization or CPU utilization of the feature extraction processing device from the reference network bandwidth utilization or reference CPU utilization.

[0050] The metric ratio corresponding to the total feature extraction time indicates the degree of influence of the relative deviation rate between the total feature extraction time and the reference total feature extraction time on the feature extraction anomaly index. This ratio quantifies the critical weight of processing efficiency anomalies in the overall evaluation; the greater the time delay, the more significant the contribution to the feature extraction anomaly index. The metric ratio corresponding to the network bandwidth utilization rate of the processing device involved in feature extraction indicates the strength of the effect of the relative deviation rate between the network bandwidth utilization rate of the processing device involved in feature extraction and the reference network bandwidth utilization rate on the feature extraction anomaly index. This value determines the sensitivity of network transmission resource anomalies in the comprehensive evaluation; bandwidth fluctuations exceeding the expected range will amplify the anomaly index. The metric ratio of the CPU utilization rate of the processing device involved in feature extraction indicates the weight of the influence of the relative deviation rate between the CPU utilization rate of the processing device involved in feature extraction and the reference CPU utilization rate on the feature extraction anomaly index. This ratio measures the importance of computational load anomalies in system stability evaluation; abnormal fluctuations in the CPU utilization rate of the processing device involved in feature extraction directly reflect the health status of the processing device.

[0051] The database stores preset evaluation benchmark parameters, including the total duration of the reference feature extraction process, reference network bandwidth utilization, and reference CPU utilization. These parameters are dynamically bound to the feature extraction process and the preset anomaly evaluation benchmarks through a structured weight configuration table, forming a complete feature extraction anomaly evaluation system. When it is necessary to calculate the feature extraction anomaly index, the system can retrieve the corresponding measurement ratio values ​​from the database in real time based on the total duration of the feature extraction process, the network bandwidth utilization of the processing device to which the feature extraction belongs, and the CPU utilization of the processing device to which the feature extraction belongs, using a relational query statement. Among them, the measurement ratio values ​​A1, A2, and A3 are weight coefficients, and their values ​​range from 0 to 1.

[0052] The educational resource storage module is used to monitor the storage process of educational resources classified and stored in the education area of ​​the database, obtain and parse abnormal parameters of the educational resource classification and storage process, and thus determine whether to optimize the educational resource classification and storage process until all educational resources are classified and stored in the storage area of ​​the database.

[0053] Specifically, the abnormal parameters of the classified storage process of educational resources are obtained and parsed. The specific parsing process is as follows: the abnormal parameters of the classified storage process of educational resources include the RAID utilization rate of the storage device to which the classified storage of educational resources belongs, the SSD reserved space of the storage device to which the classified storage of educational resources belongs, and the queue depth of the storage device to which the classified storage of educational resources belongs.

[0054] The measurement of abnormal parameters for the classified storage of educational resources should adopt a hierarchical approach: the RAID utilization rate of the storage devices belonging to the classified storage of educational resources is obtained directly from the physical space utilization rate of the underlying array through the storage management interface or controller command-line tools (prioritizing the file system tools of the operating system); the SSD reserved space of the storage devices belonging to the classified storage of educational resources relies on SMART data parsing, with a focus on monitoring the consumption status of the reserved space; the queue depth of the storage devices belonging to the classified storage of educational resources is mainly captured in real time by operating system-level performance tools to capture the number of pending I / O requests of the device, and can be verified through the storage system performance interface.

[0055] Obtain the final value of the feature extraction anomaly index and match it with the resource classification storage anomaly screening value from the database.

[0056] The final value of the feature extraction anomaly index includes the feature extraction anomaly index, the secondary feature extraction anomaly index, and the tertiary feature extraction anomaly index.

[0057] An anomaly index final value for the target resource is generated using a feature extraction algorithm. This final value is then used as a key query parameter to perform a matching operation in a pre-defined database anomaly rule table. The matching process locates the corresponding stored rule group based on the resource classification identifier, and finally retrieves the anomaly removal threshold associated with the final anomaly index value. The output result is the removal judgment value that meets the current resource type and anomaly level.

[0058] The Educational Resource Classification and Storage Anomaly Index is used to quantitatively assess the degree of anomalies in the educational resource classification and storage process.

[0059] The system retrieves preset reference RAID utilization, reference SSD reserved space, and reference queue depth from the database. It then monitors the RAID utilization, SSD reserved space, and queue depth of the storage devices belonging to the educational resource classification storage in real time. The system calculates the deviation rate between these three values ​​and the corresponding reference values, assigns each deviation rate to a preset metric ratio in the database, and summarizes the results. Finally, it filters out abnormal values ​​in the resource classification storage to obtain the educational resource classification storage anomaly index.

[0060]

[0061] JCZ is the anomaly index for classified storage of educational resources; RAL is the RAID utilization rate of the storage device to which the classified storage of educational resources belongs; SDL is the SSD reserved space of the storage device to which the classified storage of educational resources belongs; DSD is the queue depth of the storage device to which the classified storage of educational resources belongs; S is the anomaly screening value for classified storage of educational resources; RAL_L is the preset reference RAID utilization rate in the database; SDL_L is the preset reference SSD reserved space in the database; DSD_L is the preset reference queue depth in the database; B1 is the measurement ratio corresponding to the preset RAID utilization rate in the database; B2 is the measurement ratio corresponding to the preset SSD reserved space in the database; and B3 is the measurement ratio corresponding to the preset queue depth in the database.

[0062] Reference RAID utilization rate is a reference value used to characterize the RAID utilization rate of the storage devices belonging to the educational resource classification storage; Reference SSD reserved space is a reference value used to characterize the SSD reserved space of the storage devices belonging to the educational resource classification storage; Reference queue depth is a reference value used to characterize the queue depth of the storage devices belonging to the educational resource classification storage.

[0063] In the operation of an educational resource storage system, if the RAID utilization rate, SSD reserved space, and queue depth of the storage devices belonging to the educational resource classification storage significantly deviate from the reference ranges, it triggers a mutually reinforcing chain of performance degradation: A RAID utilization rate higher than the reference RAID utilization rate directly squeezes the SSD reserved space below the minimum level required to guarantee performance. The combined effect of these two factors leads to a significant increase in latency for I / O request processing, forcing the queue depth of the storage devices to be far higher than the reference queue depth range under normal load. Conversely, if the queue depth of the storage devices remains at a high deviation for an extended period, it further exacerbates resource contention and pressure within the devices, rapidly depleting the already insufficient SSD reserved space and amplifying the negative impact of RAID utilization on performance. Ultimately, this creates a vicious cycle that is difficult to break, threatening system stability and the smoothness of educational resource access.

[0064] The RAID utilization rate of the storage devices belonging to the educational resource classification storage represents the relative ratio of the absolute deviation between the RAID utilization rate of the educational resource classification storage devices and the reference RAID utilization rate, and its impact on the educational resource classification storage anomaly index. This ratio quantifies the contribution weight of abnormal core array load in the overall storage health assessment. The SSD reserved space metric of the storage devices belonging to the educational resource classification storage represents the relative ratio of the absolute deviation between the SSD reserved space of the educational resource classification storage devices and the reference SSD reserved space, and its impact on the educational resource classification storage anomaly index. This ratio determines the importance of insufficient or excessive SSD reserved space in the comprehensive assessment of storage performance and lifespan stability anomalies. The queue depth metric of the storage devices belonging to the educational resource classification storage represents the relative ratio of the absolute deviation between the queue depth of the educational resource classification storage devices and the reference queue depth, and its impact on the educational resource classification storage anomaly index. It is used to measure the sensitivity and impact weight of abnormal states such as saturated or idle I / O request processing capacity in the overall storage anomaly assessment.

[0065] The database stores preset benchmark parameters for evaluating the performance of educational resource storage, including reference RAID utilization, reference SSD reserved space, and reference queue depth. These benchmark parameters are dynamically associated with the specific storage devices and characteristics of each educational resource category through a structured metric ratio configuration table, forming the calculation model framework for the educational resource category storage anomaly index. When it is necessary to calculate the anomaly index of a specific educational resource category, the system can retrieve the corresponding reference value and preset metric ratio value from the database in real time through a relational query statement based on the unique identifier or characteristics of the device. Among them, the metric ratio values ​​B1, B2, and B3 serve as weighting coefficients reflecting the relative importance of each performance indicator, and their values ​​typically range from 0 to 1.

[0066] Specifically, the process of determining whether to optimize the classification and storage of educational resources involves comparing the educational resource classification and storage anomaly index with the storage anomaly threshold. If the educational resource classification and storage anomaly index is less than or equal to the storage anomaly threshold, then the educational resources in the educational region of the database are classified and stored in the storage area of ​​the database.

[0067] Figure 4 This is a flowchart of the educational resource storage process of the present invention, which monitors the storage process of writing educational resources to the database after classification; by acquiring and parsing abnormal parameters generated during the storage process, the efficiency and stability of the current storage process are evaluated; based on the analysis results, it is determined whether it is necessary to optimize or adjust the storage strategy or process.

[0068] If the educational resource classification storage anomaly index is greater than the storage anomaly threshold, the educational resources will be classified and stored in the secondary storage area of ​​the database, and the data page size during storage will be increased based on the educational resource classification storage anomaly index.

[0069] After adjustment, if the storage anomaly index of educational resource classification in the secondary storage area of ​​the database is less than or equal to the storage anomaly threshold, then the educational resource classification in the secondary storage area of ​​the database will be stored in the database's storage area.

[0070] If the educational resource classification storage anomaly index in the secondary storage area of ​​the database is greater than the storage anomaly threshold, then the classified educational resources will be stored in the tertiary storage area of ​​the database, and it is determined that the educational resource classification storage process will be optimized a second time.

[0071] Increasing the data page size when the anomaly index of educational resource classification storage is high is essentially a space-for-time strategy that fundamentally reduces the I / O pressure on storage devices. This directly reduces the number of requests the device needs to process (optimizing queue depth), increases the effective load of a single I / O (optimizing RAID efficiency), and reduces the write burden on SSDs (optimizing reserved space consumption). These three factors work synergistically to improve the overall performance and stability of the storage subsystem, enabling it to cope with abnormal conditions and thus reducing the anomaly index of educational resource classification storage in the secondary storage area of ​​the database.

[0072] Storage anomaly threshold is the upper limit of the storage anomaly index for educational resource categories.

[0073] Based on the anomaly index of educational resource classification storage, the data page size during the storage process is increased. In the database, the educational resource classification storage anomaly index is divided into different gradients, each gradient corresponding to a different increment of data page size. Now, the data page size is added to the increment of the data page size to obtain the next data page size.

[0074] The core advantage of this educational resource classification and storage optimization mechanism lies in its intelligent hierarchical processing flow. The system makes decisions based on a dynamic comparison between the educational resource classification and storage anomaly index and the storage anomaly threshold: compliant resources directly enter the high-efficiency main storage area to ensure core performance and security; resources exceeding the threshold are isolated to the secondary storage area and undergo targeted optimizations such as increasing data page size to improve read / write bottlenecks. A secondary evaluation is then performed: successfully optimized resources are migrated back to the main storage area for maximum utilization; those failing optimization are downgraded to the tertiary storage area and trigger in-depth secondary optimization analysis. This design ensures main storage performance through strict storage anomaly threshold admission, intelligently repairs resources using isolation, optimization, and secondary opportunity mechanisms to avoid waste, and enhances system resilience by isolating risks through a hierarchical structure. The entire automated process significantly improves storage efficiency, resource utilization, stability, and scalability, providing strong support for the secure and efficient management of massive educational resources.

[0075] Furthermore, the decision is to perform a secondary optimization on the educational resource classification and storage process. The specific decision process is as follows: based on the educational resource classification and storage anomaly index in the secondary storage area of ​​the database, the data page size is increased and the TCP window size of the network transmission layer is increased.

[0076] If, after the second adjustment, the storage anomaly index of educational resources in the tertiary storage area of ​​the database is less than or equal to the storage anomaly threshold, then the classified educational resources will be stored in the database's storage area.

[0077] If the storage anomaly index of educational resources in the three storage areas of the database exceeds the storage anomaly threshold, an early warning will be issued and manual intervention will be carried out.

[0078] By increasing the data page size, the database system reduces the number of physical I / O operations required to interact with storage devices, promoting more efficient large-block sequential reads and writes, thereby improving RAID processing efficiency, optimizing SSD write patterns, and reducing the need for high storage queue depths. Simultaneously, increasing the TCP window size ensures that network transmission is no longer a bottleneck, enabling rapid delivery / retrieval of data that needs to be stored or retrieved, preventing network latency from causing storage request backlogs, and ensuring that optimized data pages can be efficiently transmitted across the network. These two adjustments work together to accelerate the entire classification, transmission, and storage process, reduce the anomaly index, and ultimately achieve the goal of efficiently and stably storing educational resources in the target storage area, thereby reducing the anomaly index for educational resource classification storage in the database's three storage areas.

[0079] Based on the abnormal storage index of educational resource classification in the secondary storage area of ​​the database, the data page size is increased twice, and the TCP window size of the network transport layer is increased. The abnormal storage index of educational resource classification in the secondary storage area of ​​the database is divided into different gradients, each gradient corresponding to a different increment of data page size. Now, the data page size is added to the increment of the data page size to obtain the next data page size. In the database, the abnormal storage index of educational resource classification in the secondary storage area of ​​the database is divided into different gradients, each gradient corresponding to a different increment of TCP window size. Now, the TCP window size is added to the increment of the TCP window size to obtain the next TCP window size.

[0080] The core advantage of this educational resource classification and storage optimization process lies in its organic integration of intelligent optimization, risk control, and efficient operation and maintenance. By monitoring storage anomaly indices in real time and dynamically adjusting data page sizes and TCP window parameters, the system significantly improves storage throughput and network transmission performance, adaptively responding to fluctuations in resource access. A three-level progressive verification mechanism is employed, ensuring that data is only stored in the core storage area when the anomaly index reaches a certain threshold, thus strictly guaranteeing the integrity and reliability of stored data. If the threshold is still exceeded after automatic optimization, an early warning is immediately triggered, and manual intervention is initiated. This not only prevents the system from running out of control under abnormal conditions but also provides a fallback solution for complex problems.

[0081] The educational resource recommendation module is used to extract feature information of user behavior, compare the feature information with the data of various categories in educational resources, and thus make accurate recommendations for educational resources.

[0082] Specifically, the process of extracting user behavior feature information involves: collecting user behavior data and processing the data based on a pre-set artificial intelligence algorithm model.

[0083] Extract feature parameters that reflect user behavior, including feature vectors of learning preferences, ability levels, and resource usage patterns.

[0084] A structured user feature dataset is generated based on feature vectors of learning preferences, ability levels, and resource usage patterns.

[0085] Figure 5 The flowchart for recommending educational resources in this invention analyzes and extracts key feature information from user behavior data; similarity calculations or matching comparisons are performed between this user feature information and various types of data pre-classified in the education area of ​​the database; finally, based on the matching results, educational resources that are highly compatible with the user's interests and needs are accurately recommended.

[0086] To extract feature vectors reflecting users' learning preferences, ability levels, and resource usage patterns, the system first collects and cleans raw user behavior data. Then, it automatically analyzes the processed data based on pre-defined artificial intelligence algorithm models. These models transform raw behavior records into quantitative indicators by performing complex feature engineering and pattern recognition tasks (such as statistical computation, sequence analysis, latent trait inference, or deep learning), ultimately constructing structured feature vectors that accurately characterize the three core dimensions of user behavior.

[0087] Raw data includes resource usage patterns such as resource click records, search keywords, play / pause operations, access duration, and device information; ability levels include exercise answer results, answer duration, test scores, error patterns, and learning progress; and learning preferences include content type selection, note / save behavior, learning time distribution, and behavioral sequence paths.

[0088] The core advantage of this user behavior feature extraction and analysis process lies in its efficient and automated processing of raw behavioral data through pre-defined artificial intelligence algorithms. This process accurately extracts key feature vectors representing users' learning preferences, ability levels, and resource usage patterns, generating a structured dataset. This significantly improves the efficiency and objectivity of obtaining in-depth user insights from massive and complex behavioral data, avoiding the limitations of manual analysis. The generated structured feature set not only comprehensively and quantitatively depicts user profiles but, more importantly, provides a highly operable foundation for subsequent applications. It directly supports intelligent services such as personalized learning resource recommendations, adaptive learning path planning, and precise adjustments to teaching strategies. Simultaneously, it provides a unified and reliable quantitative basis for product function optimization, rational resource allocation, and data-driven scientific decision-making, ultimately powerfully driving user experience improvement and service efficiency optimization.

[0089] Specifically, this allows for precise recommendations of educational resources. The specific analysis process involves calculating the similarity between the structured user feature dataset and various educational resources in the database storage area to obtain the matching degree of each type of educational resource.

[0090] If the matching degree of a certain type of educational resource is higher than the preset recommendation threshold, the educational resource of that type will be automatically placed in the recommendation queue and pushed to the user.

[0091] If the matching degree of a certain type of educational resource is lower than the preset recommendation threshold, the exploratory recommendation process will be triggered. The system will automatically filter and generate a supplementary recommendation queue based on the potential interest mining, resource popularity and diversity strategy of the structured user feature dataset.

[0092] Similarity calculation uses vectorized user features and educational resource metadata, employs a multi-algorithm fusion strategy (cosine similarity to handle sparse features, Euclidean distance to measure numerical differences) to perform multi-dimensional comparisons, and generates a weighted comprehensive matching degree.

[0093] The exploratory recommendation process first mines potential interests by analyzing similar user behavior patterns through association rules, decomposing users using latent semantic models, utilizing resource interaction matrices, and traversing the knowledge association graph using graph neural networks to predict unexpressed potential needs. Second, it introduces a resource popularity supplementation mechanism, selecting top resources based on real-time dynamically calculated comprehensive popularity indicators to ensure the effectiveness of basic recommendations. Finally, it implements a diversity guarantee strategy, forcibly dispersing resource types, inserting cross-domain exploratory content, and overlaying new resource exposure weighting and a historical recommendation cooling mechanism to generate the final supplementary queue.

[0094] The preset recommendation threshold is used to characterize the lower limit of the matching degree of a certain type of educational resources.

[0095] This educational resource recommendation mechanism boasts several significant advantages: It achieves personalized recommendations through precise similarity calculations, ensuring users receive resources highly aligned with their characteristics, directly improving learning efficiency and user satisfaction. By introducing intelligent judgment with preset recommendation thresholds, it proactively breaks through inherent matching patterns through an exploratory recommendation process, cleverly balancing accuracy and exploration. This system not only matches resources based on explicit user characteristics but also proactively discovers valuable learning content that users may not yet be aware of by mining potential interests and combining resource popularity and diversity strategies. This mechanism, through automated processes, achieves efficient, dynamic, and comprehensive resource supply, creating continuously optimized learning paths for users and building a more intelligent and attractive resource distribution ecosystem for the platform.

[0096] Reference Figure 2As shown, the second aspect of this invention provides a method for intelligent classification and precise recommendation of educational resources based on artificial intelligence, comprising: Step 1, extracting features from educational resources based on artificial intelligence, monitoring the feature extraction process of educational resources, collecting and analyzing abnormal parameters of the feature extraction process, determining whether to adjust the feature extraction process of educational resources, thereby intelligently classifying educational resources and storing them in the education area of ​​the database; Step 2, monitoring the storage process of educational resources classified and stored in the education area of ​​the database, acquiring and parsing abnormal parameters of the classification and storage process of educational resources, thereby determining whether to optimize the classification and storage process of educational resources, until all educational resources are classified and stored in the storage area of ​​the database; Step 3, extracting feature information of user behavior, comparing the feature information with the classification data of each category in the educational resources, thereby making precise recommendations for educational resources.

[0097] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined by the present invention, they should all fall within the protection scope of the present invention.

Claims

1. An artificial intelligence-based intelligent classification and accurate recommendation system for educational resources, characterized in that, The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. 2.The artificial intelligence-based education resource intelligent classification and accurate recommendation system according to claim 1, characterized in that: The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. 3.The AI-based education resource intelligent classification and accurate recommendation system according to claim 1, characterized in that: The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education resource classification and storage system. The application relates to an intelligent education Re-extracting features of the education resources in the temporary area of the database using the optimized AI model, obtaining a secondary feature extraction anomaly index, comparing the secondary feature extraction anomaly index with the feature extraction anomaly threshold value, if the secondary feature extraction anomaly index is less than or equal to the feature extraction anomaly threshold value, intelligently classifying and storing the education resources in the education area of the database; If the secondary feature extraction anomaly index is greater than the feature extraction anomaly threshold value, the feature extraction process of the education resources is judged to be adjusted twice.

4. The artificial intelligence-based education resource intelligent classification and accurate recommendation system according to claim 3, characterized in that: The specific adjustment process of the judgment of the secondary adjustment of the feature extraction process of the education resources is: The education resources are classified and stored in the temporary area of the database, and then the initial learning rate of the AI model is reduced and the noise intensity of the input data is increased based on the secondary feature extraction anomaly index; After the secondary adjustment is completed, the education resources in the temporary area of the database are re-extracted using the optimized AI model, the parameters of the feature extraction process are re-monitored, the tertiary feature extraction anomaly index is compared with the feature extraction anomaly threshold value, if the tertiary feature extraction anomaly index is less than or equal to the feature extraction anomaly threshold value, the education resources are intelligently classified and stored in the education area of the database; If the tertiary feature extraction anomaly index is greater than the feature extraction anomaly threshold value, a warning is given and manual intervention is performed. 5.The AI-based education resource intelligent classification and accurate recommendation system according to claim 1, characterized in that: The specific analysis process of the classified storage process anomaly parameter of the education resources is: The classified storage process anomaly parameter of the education resources includes the RAID utilization rate of the storage device to which the classified storage of the education resources belongs, the SSD reserved space of the storage device to which the classified storage of the education resources belongs, and the queue depth of the storage device to which the classified storage of the education resources belongs; The feature extraction anomaly index final value is obtained, and the resource classification storage anomaly exclusion value is matched from the database; The education resource classification storage anomaly index is used for quantitative evaluation of the abnormality degree in the education resource classification storage process; The preset reference RAID utilization rate, reference SSD reserved space and reference queue depth are obtained from the database, the RAID utilization rate of the storage device to which the classified storage of the education resources belongs, the SSD reserved space of the storage device to which the classified storage of the education resources belongs, and the queue depth of the storage device to which the classified storage of the education resources belongs are monitored in real time, the deviation rates of the three from the corresponding reference values are calculated, each deviation rate is assigned to the preset measurement proportion in the database, and finally the resource classification storage anomaly index is obtained by excluding the resource classification storage anomaly exclusion value. 6.The AI-based education resource intelligent classification and accurate recommendation system according to claim 1, characterized in that: The specific judgment process of the judgment of whether to optimize the classified storage process of the education resources is: Comparing the education resource classification storage anomaly index with the storage anomaly threshold value, if the education resource classification storage anomaly index is less than or equal to the storage anomaly threshold value, the education resources in the education area of the database are classified and stored in the storage area in the database; If the education resource classification storage anomaly index is greater than the storage anomaly threshold value, the education resources are classified and stored in the secondary storage area in the database, and the data page size in the storage process is increased based on the education resource classification storage anomaly index; After the adjustment, if the education resource classification storage abnormality index of the secondary storage area in the database is less than or equal to the storage abnormality threshold value, the education resource classification storage area of the secondary storage area in the database is stored in the storage area of the database; If the education resource classification storage abnormality index of the secondary storage area in the database is greater than the storage abnormality threshold value, the classified education resources are stored in the tertiary storage area of the database, and it is judged whether to perform secondary optimization on the education resource classification storage process.

7. The artificial intelligence-based education resource intelligent classification and accurate recommendation system according to claim 6, characterized in that: The specific judgment process is: Based on the secondary increase of the education resource classification storage abnormality index in the secondary storage area in the database, the data page size and the TCP window size of the network transmission layer are increased; If the education resource classification storage abnormality index in the tertiary storage area of the database is less than or equal to the storage abnormality threshold value after the secondary adjustment, the classified education resources are stored in the storage area of the database; If the education resource classification storage abnormality index in the tertiary storage area of the database is greater than the storage abnormality threshold value, a warning is given, and manual intervention is performed. 8.The AI-based education resource intelligent classification and accurate recommendation system according to claim 1, characterized in that: The specific analysis process of extracting the feature information of user behavior is: Collect user behavior data, process the user behavior data based on a preset artificial intelligence algorithm model; Extract feature parameters reflecting user behavior, including learning preference, ability level and resource use mode feature vectors; Based on the learning preference, ability level and resource use mode feature vectors, a structured user feature data set is generated. 9.The AI-based education resource intelligent classification and accurate recommendation system according to claim 1, characterized in that: The specific analysis process of accurately recommending education resources is: Calculate the similarity between the structured user feature data set and each type of education resource in the database storage area to obtain the matching degree of each type of education resource; If the matching degree of a certain type of education resource is higher than the preset recommendation threshold, the system automatically places the education resource in the recommendation queue and pushes it to the user; If the matching degree of a certain type of education resource is lower than the preset recommendation threshold, the exploratory recommendation process is triggered, and the system will automatically filter and generate a supplementary recommendation queue based on the potential interest mining of the structured user feature data set, resource popularity and diversity strategy.

10. An AI-based intelligent classification and precise recommendation method for educational resources, applied to implement the AI-based intelligent classification and precise recommendation system for educational resources as described in any one of claims 1-9, characterized in that: It includes: Step 1: Based on artificial intelligence, the features of education resources are extracted, the feature extraction process of education resources is monitored, abnormal parameters of the feature extraction process are collected and analyzed, and it is judged whether to adjust the feature extraction process of education resources, so as to intelligently classify and store education resources in the education area of the database; Step 2: Monitor the storage process of the classified education resources to the education area of the database, obtain and analyze the classification storage process abnormality parameters of the education resources, and judge whether to optimize the classification storage process of the education resources until all the education resources are classified and stored in the storage area of the database; Step 3: Extract the feature information of user behavior, compare the feature information with each classification data in the education resources, and accurately recommend the education resources.

Images