Personalized educational resource pushing system based on Internet

Abstract

The invention provides a personalized educational resource pushing system based on the Internet, relates to the field of resource pushing, and aims at performing cognitive load monitoring on user learning process data by constructing a personalized educational resource pushing system based on cognitive load, and performing resource pushing on the basis of learning state synchronization, conflict resolution and other technical means. Precise pushing of personalized learning resources is realized, and the learning efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of resource push, and more specifically, to an Internet-based personalized educational resource push system. Background Art

[0002] In recent years, with the rapid development of Internet technology, online education platforms have gradually become popular, and the personalized learning mode has become an important research direction for improving learning efficiency. A personalized education system based on big data and artificial intelligence can analyze the cognitive state of users according to their learning behaviors, and then push adaptive learning resources to improve learning efficiency. However, existing personalized education systems mainly rely on users' basic information (such as age, learning preferences) or historical learning records (such as completed courses, exercise accuracy rates) for resource recommendation, and lack accurate analysis of users' real-time learning states, resulting in a deviation between the pushed content and the current learning needs.

[0003] Specifically, the existing personalized education systems mainly have the following deficiencies: First, in terms of learning resource push, there is a lack of in-depth monitoring of the learning process of users, making it difficult to accurately evaluate the current learning load state of users, resulting in subjectivity in setting the timing and priority of resource push and being unable to fully adapt to the real-time learning needs of users; Second, in terms of multi-terminal learning synchronization, existing technologies only rely on timestamps or basic synchronization rules for progress update, unable to accurately identify the differences in learning states between different devices, and thus resulting in inconsistencies in learning progress; In addition, during the learning process, there may be data conflicts in the operations of different terminals, and existing technologies fail to effectively identify and resolve these conflicts, which may lead to the loss of learning records or incorrect synchronization, etc. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides an Internet-based personalized educational resource push system.

[0005] According to one aspect of the present invention, there is provided an Internet-based personalized educational resource push system, which includes: a cognitive load monitoring module, which forms a three-dimensional load index matrix based on user learning process data according to the attention dispersion degree, operation delay, and error rate; a push control module, which determines the push timing and priority of learning resources according to the three-dimensional load index matrix; a learning state synchronization module, which performs hierarchical state mapping based on user cross-terminal learning data, encodes the learning state into a multi-layer state vector, and calculates the learning progress difference between different devices through a difference matrix; a conflict resolution module, which resolves conflicts based on the relevance of learning content when detecting operation conflicts during the multi-terminal learning state synchronization process; and a data storage module, which is used to store user learning-related data and dynamically updates the stored content based on the instructions of the learning state synchronization module.

[0006] Further, the user learning process data includes eye movement data, mouse operation trajectories, and click time intervals.

[0007] Further, the three-dimensional load index matrix includes: according to the task type, different weights are assigned to the attention offset, operation response time, and error rate to obtain the load contribution values of different features in the current learning task; the load contribution values in different time slices are arranged in chronological order to form a three-dimensional load index matrix.

[0008] Further, the push timing of the learning resources includes: calculating the comprehensive load value in the current time slice according to the three-dimensional load index matrix and comparing it with a preset load threshold: if the comprehensive load value is in the low load range, learning resources are pushed in the corresponding time slice; if the comprehensive load value exceeds the high load threshold, the push is adjusted.

[0009] Further, the calculation of the comprehensive load value is: according to the three-dimensional load index matrix, within a set time window, the load contribution values of consecutive time slices are weighted and summed.

[0010] Further, according to the type of learning resources and the user's learning progress, priorities are set for the resources; a hierarchical buffer queue including high, medium, and low priorities is constructed, and the resources to be pushed are stored in the corresponding queue levels according to their priorities; if in the low load range, the resources to be pushed are selected from the high-priority queue; if the high-priority queue is empty, resources are selected from the medium-priority queue for pushing; in the high load state, the pushing of low-priority resources is suspended, and the push delay time is recorded. When the user's load decreases, the delayed resources are pushed according to the adjusted queue order.

[0011] Further, encoding the learning state into a multi-layer state vector includes: collecting learning progress-related data based on the user's learning behaviors on different devices; encoding the user's learning state into a multi-layer state vector, with each layer corresponding to different types of learning data.

[0012] Further, calculating the learning progress difference between different devices through a difference matrix includes: Based on the learning state vector of the current terminal and the learning state vectors of other terminals, the state difference of each dimension is calculated to generate a difference matrix; according to the timestamp information of the learning state, it is judged whether the data in the difference matrix comes from the same time slice. If the time slices are the same, the difference value is directly calculated; if the time slices are different, the calculation is completed by supplementing according to the nearest matching principle; based on the data in the difference matrix, the learning progress lag degree between the current terminal and other terminals is calculated. If the lag value exceeds the set threshold, it is determined that the learning progress of the current terminal lags.

[0013] Further, the operation conflict is as follows: compare the learning status update times of the current terminal and other terminals, and determine whether there are learning status change records with overlapping times; collect the learning status data of all terminals, establish a learning status comparison matrix, analyze the conflict points between the statuses, and if the statuses of the same task are inconsistent on different terminals, it is determined as an operation conflict.

[0014] Further, the conflict resolution based on the relevance of learning content includes: According to the hierarchical relationship of learning resources, establish a task relevance network, and label the sequence and dependency relationships between tasks; if an operation conflict occurs, enter the conflict resolution process: query the task relevance network to determine the priorities of the conflicting tasks; if the conflicting tasks have a dependency relationship, give priority to retaining the completed status of the previous task; if the conflicting tasks are at the same level, select the status according to the learning progress; according to the priorities of the conflicting tasks, determine the reasonable status of the conflicting tasks and synchronize it to all terminals.

[0015] Compared with the prior art, the personalized educational resource push system based on the Internet provided by the present invention monitors the attention dispersion degree, operation delay and error rate during the user's learning process through constructing a personalized learning resource push system based on cognitive load, forms a three-dimensional load index matrix, so as to realize accurate learning status evaluation and intelligent push control. Analyze the learner's cognitive load level through this matrix, dynamically adjust the push timing and priority of learning resources, ensure that the content is pushed in the best cognitive state, and improve learning efficiency. At the same time, through the learning status synchronization module, the cross-terminal learning data is hierarchically mapped, and the learning progress difference between devices is calculated using the difference matrix to ensure the consistency of learning content on different devices and improve the coherence of cross-device learning. In addition, aiming at the problem of multi-terminal operation conflicts, conflict resolution is carried out based on the relevance of learning content to ensure the integrity and sequential rationality of learning tasks. Finally, the present invention effectively reduces the cognitive load during the learning process, improves the matching degree and push accuracy of personalized learning resources, enhances the cross-device learning experience, and optimizes the adaptability and user-friendliness of the intelligent education system. The present invention focuses on monitoring the cognitive load of the user's learning process data, and realizes the accurate push of personalized learning resources based on technical means such as learning status synchronization, conflict resolution and data optimized storage, so as to improve learning efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to these drawings without creative efforts. In the drawings: Figure 1 It is a system block diagram of an Internet-based personalized education resource push system according to an embodiment of the present invention.

[0017] Figure 2 It is a block diagram of a learning status synchronization module in an Internet-based personalized education resource push system according to an embodiment of the present invention. Detailed implementation manners

[0018] Next, exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments of the present invention. It should be understood that the present invention is not limited by the exemplary embodiments described herein.

[0019] As mentioned in the above background art, the existing personalized education systems mainly have the following deficiencies: First, in terms of learning resource push, there is a lack of in-depth monitoring of the user's learning process, making it difficult to accurately evaluate the user's current learning load status, resulting in subjectivity in setting the timing and priority of resource push and being unable to fully adapt to the user's real-time learning needs; Second, in terms of multi-terminal learning synchronization, the existing technologies only rely on timestamps or basic synchronization rules for progress update, unable to accurately identify the differences in learning status between different devices, and thus leading to inconsistencies in learning progress; In addition, during the learning process, there may be data conflicts in the operations of different terminals, and the existing technologies fail to effectively identify and resolve these conflicts, which may result in the loss or incorrect synchronization of learning records, etc. The present invention proposes an Internet-based personalized education resource push system to solve the above problems.

[0020] Figure 1 It is a system block diagram of an Internet-based personalized education resource push system according to an embodiment of the present invention. As Figure 1 shown, in the Internet-based personalized education resource push system, it includes: A cognitive load monitoring module 100, which forms a three-dimensional load index matrix based on the user's learning process data according to the attention dispersion degree, operation delay, and error rate. A push control module 200, which determines the push timing and priority of learning resources according to the three-dimensional load index matrix. A learning status synchronization module 300, which performs hierarchical status mapping based on the user's cross-terminal learning data, encodes the learning status into a multi-layer status vector, and calculates the learning progress difference between different devices through a difference matrix. A conflict resolution module 400, which, during the multi-terminal learning status synchronization process, when detecting an operation conflict, resolves the conflict based on the learning interaction data in combination with a timestamp matrix and an operation weight rule; A data storage module 500, which is used to store the user's learning-related data and dynamically updates the stored content based on the instructions of the learning status synchronization module.

[0021] In an embodiment of the present invention, the cognitive load monitoring module 100 specifically includes collecting eye movement trajectory data, mouse movement data, and click time interval data during the user's learning process, storing them in the load data buffer, and marking corresponding data category labels in the load data buffer according to the current learning task type of the user for load calculation in different scenarios.

[0022] It is known that in traditional learning systems, static task records or user self-evaluation methods are usually used to measure learning load. For example, a self-evaluation questionnaire is popped up regularly, and users are required to subjectively evaluate the learning difficulty and fatigue level. However, this method has great limitations. For example, user feedback may be affected by factors such as personal emotions and subjective biases, and it is difficult to reflect the actual learning state in real time. Therefore, the present invention adopts a real-time acquisition method based on user behavior data to improve the accuracy and objectivity of load monitoring.

[0023] It should be noted that the data acquisition in the present invention covers three main categories: eye movement trajectory data, mouse movement data, and click time interval data. Among them, the eye movement trajectory data is used to analyze the stay situation of the user's line of sight in the learning content area, record the moving path and stay time of the line of sight focus; the mouse movement data reflects the operation method of the user on the interface during the interaction process, including the mouse trajectory, moving speed, and stay duration; the click time interval data is used to calculate the operation delay of the user when completing learning tasks (such as answering questions, switching pages). All these data need to be stored in the load data buffer to ensure quick access during subsequent calculation processes.

[0024] Preferably, during the data storage process, in order to facilitate subsequent analysis, the present invention marks data category labels in the load data buffer. Specifically, each piece of data is associated with the current learning task type and marked with corresponding labels according to the task nature. For example, in a reading task, the eye movement trajectory data may be more critical, while in an interactive experiment task, the mouse movement data and click time interval data may be more representative. In this way, load calculation can be carried out specifically in different task scenarios, avoiding interference of irrelevant data on the calculation results, and improving the pertinence and accuracy of load analysis.

[0025] During the load calculation process, it is necessary to select data within a time window in the load data buffer to ensure the time continuity of the calculation. The selection of the time window is one of the key steps in calculating cognitive load. Usually, traditional methods may use a fixed-length window (such as 30 seconds, 1 minute) for data sampling, but this method may lead to mutations between data segments and affect the smoothness of the calculation. Therefore, the present invention adopts a dynamic time window strategy, which can adaptively adjust the window length according to the user's learning rhythm to more accurately reflect the user's current learning state.

[0026] After the selected time window, according to the data category labels, the attention deviation, operation response time, and error rate are calculated respectively. Among them, the attention deviation represents the frequency and amplitude of the user's line of sight deviating from the learning content area. This indicator can be used to evaluate the user's concentration, and the calculation considers the proportion of time when the user's line of sight deviates from the area. The operation response time is used to measure the interaction delay of the user for learning tasks, such as the reaction time required for the user to click a button or switch pages with the mouse. The present invention not only focuses on the response time itself but also combines the user's past operation patterns to determine whether the current response time is abnormal. For example, if the user usually has a short response time, but the response time for a certain task increases abnormally, it may indicate that the task is too difficult or the cognitive load is overloaded. The error rate represents the proportion of errors that occur during the user's learning process, such as the proportion of the number of incorrect answers to the total number of answers.

[0027] In the present invention, in order to accurately evaluate the cognitive load under different tasks, an adaptive weight allocation strategy is adopted to assign different weights to the attention deviation, operation response time, and error rate. Traditional methods may directly use fixed weights for weighted calculation. For example, the same influence ratio is given to all features. However, this method ignores the characteristics of different tasks and may lead to distorted calculation results of the cognitive load. For example, in logical reasoning tasks, the attention deviation may be the main influencing factor, while in real-time interaction tasks, the operation response time is more critical. Therefore, the present invention proposes a strategy of dynamically adjusting the feature weights according to the task type. For example, in high-concentration tasks (such as reading or mathematical derivation), the weight of the attention deviation is increased to highlight the impact of the change in the user's attention on the learning effect. In operation-sensitive tasks (such as experiment simulation or online quizzes), the weight of the operation response time is increased to more accurately measure the user's task completion ability. In addition, in tasks with a high error rate (such as programming training), the weight of the error rate can be appropriately increased to reflect the degree of the user's learning obstacles.

[0028] Since the units of the attention deviation, operation response time, and error rate are different, directly performing weighted calculation may lead to an imbalance in the calculation results. For example, the attention deviation may be in degrees or time, the operation response time is in milliseconds, and the error rate is in percentage. To ensure calculation consistency, the present invention needs to perform normalization processing on the three types of data to ensure that all data are within the same numerical range.

[0029] The present invention adopts the Min-Max normalization method, that is, each feature value is converted to the range of 0 to 1, so that the contribution degrees of different features are comparable. After normalization, the attention deviation, operation response time, and error rate are multiplied by their corresponding feature weights respectively to obtain the weighted load contribution values of each feature.

[0030] Finally, arrange the weighted load contribution values under different time slices in chronological order to form a three-dimensional load index matrix. The time slice is the segment after dynamic window adjustment. Different from the traditional single-index analysis method, this matrix can provide more comprehensive cognitive load monitoring data to support the dynamic adjustment of personalized learning paths. For example, if the matrix data shows that the load value of the user has increased abnormally during a certain period, the system can actively reduce the task difficulty to prevent cognitive overload and improve learning efficiency.

[0031] It should be noted that compared with the traditional load evaluation method, this solution realizes more accurate cognitive load evaluation through a dynamic time window, adaptive weight adjustment, and the construction of a three-dimensional load matrix, providing efficient data support for the intelligent learning system.

[0032] In the embodiment of the present invention, the push control module 200 specifically includes: determining the push timing and priority of learning resources according to the three-dimensional load index matrix provided by the cognitive load monitoring module. Using a hierarchical buffer queue, resources with higher priority (such as wrong-question analysis) are placed at the front of the queue, while low-priority resources (such as extended reading) dynamically adjust the push order according to the cognitive load index to avoid learning interference.

[0033] The push timing of the learning resources includes: calculating the comprehensive load value in the current time slice according to the three-dimensional load index matrix and comparing it with a preset load threshold: if the comprehensive load value is in the low-load interval, push learning resources in the corresponding time slice; if the comprehensive load value exceeds the high-load threshold, adjust the push.

[0034] Specifically, in the cognitive load monitoring module, a three-dimensional load index matrix has been constructed through the attention offset, operation response time, and error rate. This matrix can intuitively reflect the cognitive load of the user under different time slices; since the load value of a single time slice may fluctuate, directly relying on a single time slice may lead to unstable push strategies. Therefore, it is necessary to perform weighted summation on the load contribution values of consecutive time slices within a set time window to calculate a more stable comprehensive load value.

[0035] The present invention presets three intervals: low load, medium load, and high load. Among them: Low-load interval: The comprehensive load value is relatively low, indicating that the user is in a relatively relaxed state. At this time, new learning resources can be pushed; Medium-load interval: The load value is moderate, indicating that the user is in a normal learning state. At this time, the current rhythm can be maintained without additional push; High-load interval: The comprehensive load value is relatively high, indicating that the user's cognitive load is close to the upper limit. At this time, the push should be paused to prevent the fatigue effect caused by overlearning.

[0036] When the comprehensive load value is in the low load range, the system will push learning resources during the corresponding time slice to maintain the learning rhythm; when the comprehensive load value exceeds the high load threshold, the system will suspend the push of low-priority resources and record the push delay time to avoid overloading the learning pressure.

[0037] Furthermore, in traditional learning systems, the push of learning resources usually adopts a single queue and is pushed in a fixed order. However, this method has the following disadvantages: Key learning resources (such as wrong-question analysis) and secondary resources (such as extended reading) are mixed together, without specifically optimizing the push order, and the fixed-order push strategy cannot dynamically adjust the resource push according to the user's cognitive load, which may lead to pushing the wrong resources at the wrong time. To solve the above problems, the present invention adopts a hierarchical buffer queue mechanism to set priorities for resources based on the type of learning resources and the user's learning progress.

[0038] Exemplarily, first, the system will classify all resources to be pushed into three categories: high-priority, medium-priority, and low-priority according to the type of learning resources. For example: High-priority: Resources such as the user's wrong-question analysis and intensive practice of important knowledge points, which are crucial to the learning effect; Medium-priority: Resources such as after-class consolidation exercises and periodic tests, which help the user check the learning results; Low-priority: Resources such as extended reading and additional reference materials, which have less impact on the current learning task. Build a hierarchical buffer queue: High-priority resources enter the high-priority queue to ensure the push of key knowledge points first; Medium-priority resources enter the medium-priority queue to ensure the timely presentation of core learning content; Low-priority resources enter the low-priority queue, and the push timing is dynamically adjusted according to the user's load status.

[0039] When the system detects that the user is in a low-load state, it preferentially selects learning resources from the high-priority queue for pushing to ensure the timely consolidation of key knowledge; if the high-priority queue is empty, it selects appropriate resources from the medium-priority queue for pushing to ensure the coherence of the learning process; when the user is in a high-load state, the push of low-priority resources is suspended, and the push delay time is recorded to ensure that the user's learning rhythm will not be disturbed by excessive information.

[0040] It can be seen that the present invention adopts a hierarchical buffer queue mechanism to classify learning resources according to priorities and dynamically adjusts the push order in combination with the user's real-time load status. This mechanism effectively avoids the problem of mixed pushing of key learning resources and secondary resources in traditional push systems, ensures that important resources are pushed first when the user's cognitive load is light, and improves the learning efficiency. When the user is in a high-load state, the system will suspend the push of low-priority resources and record the push delay time to reduce learning fatigue caused by excessive information interference.

[0041] To further improve the adaptability of the push strategy, the present invention also adds the following optimization strategies: The system dynamically adjusts the push order of the hierarchical buffer queue according to the user's learning progress and the changing trend of cognitive load. For example, if the user has a high error rate within a certain period of time, the wrong-question analysis can be pushed first, while reducing the recommendation of extended reading; if the user's operation response time significantly increases, indicating a greater learning pressure, the system will reduce the task difficulty and the resource push frequency. When the system delays the push of some resources due to a high-load state, it will adjust the push order according to the user's learning progress after the user's cognitive load recovers. For example, if a low-priority resource is delayed in pushing, but its relevance is strong (such as supplementary materials for a certain chapter), it can be pushed timely before the user enters the next learning stage to avoid knowledge gaps.

[0042] The present invention also introduces a dynamic adjustment strategy. According to indicators such as the user's error rate and response time, it further optimizes the push content and frequency, thereby reducing the task difficulty and the resource push frequency when the user's load is high, and improving the self-adaptability of the system. Generally speaking, this push control module has significant beneficial effects in improving the accuracy of learning resource push, optimizing the push timing, reducing learning interference, and dynamic regulation strategies, which helps to enhance the personalized learning experience and improve the learning effect.

[0043] In the embodiment of the present invention, as Figure 2 shown, the learning status synchronization module 300 includes a hierarchical status mapping unit 310, which is used to encode the learning status of the user on different terminals into a multi-layer status vector to represent the learning status characteristics of each terminal; a difference calculation unit 320, which is used to calculate the learning progress difference between different devices based on the difference matrix and determine the incremental data that needs to be synchronized; and a synchronization instruction generation unit 330, which is used to send a synchronization instruction to the data storage module according to the learning progress difference to realize the incremental update of cross-terminal learning data.

[0044] To accurately describe the user's learning status, the system needs to collect the user's learning behavior data on different terminals and encode it into a multi-layer status vector. This status vector contains multiple levels, and each level corresponds to different types of learning data. The specific operation of the hierarchical status mapping unit 310 is as follows: The system monitors and records the user's learning behavior on different terminals, including: progress information, such as the chapters, units, and task completion degrees learned by the user; interaction data, such as the completion status, correct rate, and submission time of practice questions; time information, such as learning duration, recent access time, and pause time points; and behavior characteristics, such as page stay duration, scrolling speed, and jump behavior.

[0045] Exemplarily, the collected data is encoded into a multi-layer state vector, with each layer representing different types of data. For example: Layer 1: Basic progress information (chapter and unit completion status); Layer 2: Practice and quiz data (number of completed practice questions, accuracy rate); Layer 3: Interaction and behavior data (page dwell time, jump frequency); Layer 4: Timestamp information (recent learning time, cumulative learning duration). Each terminal will regularly generate a learning state vector and store it in the local cache or cloud database. When the user switches terminals, the system will compare the learning state vectors on different devices to ensure data synchronization.

[0046] When the user switches between different terminals for learning, the system needs to identify the progress differences of each terminal and perform intelligent synchronization to ensure the integrity of the learning state. The difference calculation unit 320 specifically includes: Calculating the learning progress differences between different devices through the difference matrix includes: calculating the state differences of each dimension based on the learning state vector of the current terminal and the learning state vectors of other terminals, and generating a difference matrix. Each column of the difference matrix represents the learning state of a certain layer (such as progress, practice completion rate, behavior characteristics, etc.), and each row represents the state of different terminals.

[0047] Based on the timestamp information of the learning state, determine whether the data in the difference matrix comes from the same time slice. If the time slices are the same (i.e., the user used two terminals within a close time), directly calculate the difference value; if the time slices are different, perform a complement calculation based on the nearest matching principle, that is, find the state vector corresponding to the closest timestamp for comparison to avoid misjudgment caused by time differences.

[0048] Exemplarily, assume that the timestamp of the state vector of a certain user on the PC side is "T1", and the timestamp of the state vector on the mobile side is "T2". If "T2" is close to "T1", the system will regard the two time slices as the same time slice; if "T2" is far away, the system will look for other state vectors near "T1" and perform a complement calculation based on the nearest matching principle.

[0049] Based on the data in the difference matrix, calculate the lag degree of the learning progress of the current terminal and other terminals. If the lag value exceeds the set threshold, it is determined that the learning progress of the current terminal lags. In this way, the system can avoid data loss or misjudgment, thereby ensuring the accuracy of data difference judgment.

[0050] When the system detects that the learning progress lags, it needs to send a synchronization instruction to the data storage module for incremental update. The synchronization instruction generation unit 330 specifically includes: Based on the lag data in the difference matrix, filter the learning states that need to be updated and send a synchronization instruction to the data storage module to achieve data incremental update.

[0051] Exemplarily, if the user has completed the wrong-question analysis of Chapter 3 on the PC side but the mobile side has not been updated, only the wrong-question analysis data of this chapter will be synchronized, rather than the overall progress. If the user has taken a quiz on the mobile side but the PC side has not recorded it, only the quiz data will be synchronized, rather than the data of all chapters.

[0052] Furthermore, to avoid redundant data transmission, the system only synchronizes the different parts, reducing the amount of data transmission and improving the synchronization efficiency. In addition, to ensure that the learning experience of the current end-user is not affected after the data is updated, the system avoids automatic page refreshing or content mutation during the synchronization process to ensure the learning coherence of the user. Through the above incremental update method, the present invention avoids the resource waste caused by traditional full-volume data coverage and effectively prevents learning interference caused by frequent data refreshing, thereby further improving the accuracy and stability of data synchronization.

[0053] It can be seen that the present invention adopts a hierarchical state mapping method to encode the learning state into a multi-layer state vector to more precisely describe the learning state of the user, and calculates the progress difference between different devices through a difference matrix to achieve efficient cross-terminal synchronization.

[0054] During the multi-terminal learning state synchronization process, the user may operate on the same learning task on different terminals, resulting in conflicts in the learning state. For example, the user may have completed a certain exercise on the mobile side, while the PC side still shows that the question is not completed; or the user may have completed the reading of a certain chapter on the tablet side, while the PC side still shows that it has not started. Such conflicts may affect the learning experience of the user and lead to inconsistent learning records. Therefore, an efficient and accurate conflict resolution mechanism is needed to ensure that the learning state is consistent on all terminals.

[0055] The present invention adopts a conflict resolution method based on the relevance of learning content. By analyzing the hierarchical relationship of learning tasks, constructing a task relevance network, determining the priority of conflicting tasks, and selecting a reasonable learning state for synchronization accordingly. This method not only considers time conflicts but also the dependencies between tasks to ensure the consistency and integrity of learning data.

[0056] In the embodiment of the present invention, the conflict resolution module 400 specifically includes: Before conflict resolution, the system first needs to detect whether there are conflicts in the learning state. The present invention adopts the method of a learning state comparison matrix to compare the data of multiple terminals. The specific operation is as follows: Collect the learning status data of all terminals, divide the learning status data of different terminals according to learning tasks, and construct a learning status comparison matrix. Each row of this matrix represents a certain learning task (such as the reading progress of a certain chapter), and each column represents the status records of different terminals. The example is as follows: Task Number Status of Terminal A Status of Terminal B Status of Terminal C Update Time Task 1 Completed Not Completed Completed 3-18 10:00 Task 2 In Progress In Progress Completed 3-18 09:50 Task 3 Not Started Not Started In Progress 3-18 09:45 The purpose of the comparison matrix is to find tasks with inconsistent statuses and further analyze whether there are operation conflicts.

[0057] If the statuses of a certain task are inconsistent among multiple terminals and the status update times overlap (that is, multiple terminals modify the status of the same task within a close time), it is determined as an operation conflict. For example, in the above table, the status of Task 1 shows "completed" on both Terminal A and Terminal C, but it is still "uncompleted" on Terminal B and the update time is relatively new, which may indicate that the user has made modifications on the B side but failed to synchronize successfully. Once an operation conflict is detected, the system will enter the conflict resolution process.

[0058] The conflict resolution strategy of the present invention not only considers time conflicts but also introduces the hierarchical relationship of learning tasks to ensure that task dependencies are not damaged. That is, the conflict resolution based on the relevance of learning content includes: According to the hierarchical relationship of learning resources, establish a task relevance network, and mark the sequence and dependency relationships between tasks; if an operation conflict occurs, enter the conflict resolution process: query the task relevance network to determine the priority of conflict tasks; if the conflict tasks have a dependency relationship, give priority to retaining the completed status of the pre-task; if the conflict tasks are at the same level, select the status according to the learning progress; according to the priority of conflict tasks, determine the reasonable status of conflict tasks and synchronize it to all terminals.

[0059] Specifically as follows: In order to judge the priority of conflict tasks, the system will construct a task relevance network according to the hierarchical relationship of learning content. The main steps are as follows: mark the sequence between tasks. For example, the reading of Chapter 1 should precede that of Chapter 2, and the exercises should precede the chapter test; if Task B depends on the completion of Task A, the learning status of Task B cannot precede that of Task A. For example, a more difficult question requires the user to complete the basic questions first; for different types of tasks, set different influence factors. For example, the weight of the learning progress of important chapters is greater than that of additional reading materials. When conflict tasks are detected, the system will execute the following resolution strategy according to the task relevance network: If conflicting tasks have a pre - dependency relationship, the completed state of the pre - task is preferentially retained; if the conflicting tasks are at the same level (for example, the reading progress of chapters on different terminals is out of sync), the latest state is selected according to the learning progress; if there are differences in the learning states of different terminals but the dependency relationship is not violated, the latest modified state is selected. When the priorities are the same, the state with the most recent update time is selected for synchronization; if the update time differences are small, the device weights are referred to (for example, learning on the PC side may be more accurate, and the PC - side data is preferentially used). If the conflicting states cannot be automatically resolved (such as Terminal A shows completed while Terminal B does not), a user confirmation window is popped up to allow the user to select the final state; for critical tasks (such as exam scores), the system can record all state changes and allow the user to trace back the historical states.

[0060] After the conflict resolution is completed, the system synchronizes the finally determined learning state to all terminals to ensure that users can see the latest learning records regardless of which device they log in to.

[0061] It can be seen that the present invention proposes a conflict resolution method based on the relevance of learning content, determines the priorities of conflicting tasks through a task relevance network, and adopts a multi - strategy conflict resolution mechanism to ensure the consistency of cross - terminal learning states. Compared with traditional methods, this solution can not only efficiently resolve operation conflicts, but also ensure the hierarchical logic of learning content and improve the user experience.

[0062] In the embodiment of the present invention, the data storage module 500 specifically includes storing user learning - related data and dynamically updating the stored content based on the instructions of the learning state synchronization module.

[0063] This module includes a data backup mechanism to prevent data loss caused by abnormal situations (such as power outages, device failures). The backup data includes: writing the user's learning state data, load data, push records, etc. into the backup area according to a preset period (such as every 24 hours). When the user switches terminals, pushes key resources, or completes a stage test, the system automatically triggers an immediate backup to ensure that the latest learning achievements are saved. When the user data is abnormal or lost, the system will preferentially restore according to the latest backup data to minimize the loss of learning data.

[0064] In summary, the personalized educational resource push system based on the Internet according to the embodiments of the present invention is elucidated. The personalized educational resource push system based on the Internet provided by the present invention constructs a personalized learning resource push system based on cognitive load, comprehensively monitors the attention dispersion degree, operation delay, and error rate during the user's learning process, forms a three-dimensional load index matrix, thereby realizing accurate learning state evaluation and intelligent push control. By analyzing the cognitive load level of learners through this matrix, the push timing and priority of learning resources are dynamically adjusted to ensure content push in the optimal cognitive state and improve learning efficiency. At the same time, through the learning state synchronization module, the cross-terminal learning data is hierarchically mapped, and the learning progress difference between devices is calculated using the difference matrix to ensure the consistency of learning content on different devices and enhance the coherence of cross-device learning. In addition, aiming at the problem of multi-terminal operation conflicts, conflict resolution is carried out based on the relevance of learning content to ensure the integrity and sequential rationality of learning tasks. Finally, the present invention effectively reduces the cognitive load during the learning process, improves the matching degree and push accuracy of personalized learning resources, enhances the cross-device learning experience, and optimizes the adaptability and user-friendliness of the intelligent education system. The present invention focuses on monitoring the cognitive load of user learning process data, and based on technical means such as learning state synchronization, conflict resolution, and data optimized storage, realizes the accurate push of personalized learning resources and improves learning efficiency.

[0065] It should be understood that the present invention is not limited to the exact structure described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present invention is only limited by the appended claims.

Claims

1. The personalized education resource push system based on the Internet is characterized by: include: The cognitive load monitoring module forms a three-dimensional load indicator matrix based on the user learning process data, according to the degree of distraction, operation delay and error rate; A push control module determines the push timing and priority of learning resources according to the three-dimensional load indicator matrix; The learning state synchronization module performs hierarchical state mapping based on the user's cross-terminal learning data, encodes the learning state into a multi-layer state vector, and calculates the learning progress differences between different devices through the difference matrix; The conflict resolution module resolves conflicts based on the relevance of learning content when an operation conflict is detected during the multi-terminal learning status synchronization process; The data storage module is used to store user learning related data and dynamically update the storage content based on the instructions of the learning status synchronization module.

2. The Internet-based personalized education resource push system according to claim 1, characterized in that: The user learning process data includes eye movement data, mouse operation trajectory and click time interval.

3. The Internet-based personalized education resource push system according to claim 2, characterized in that: The three-dimensional load index matrix includes: According to the task type, different weights are assigned to attention offset, operation response time and error rate to obtain the load contribution values ​​of different features in the current learning task; the load contribution values ​​under different time slices are arranged in chronological order to form a three-dimensional load indicator matrix.

4. The Internet-based personalized education resource push system according to claim 1, characterized in that: The timing of pushing the learning resources includes: calculating the comprehensive load value in the current time slice based on the three-dimensional load indicator matrix, and comparing it with the preset load threshold: if the comprehensive load value is in the low load range, the learning resources are pushed in the corresponding time slice; if the comprehensive load value exceeds the high load threshold, the push is adjusted.

5. The Internet-based personalized education resource push system according to claim 4, characterized in that: The calculation of the comprehensive load value is as follows: according to the three-dimensional load index matrix, within a set time window, the load contribution values ​​of consecutive time slices are weighted and summed to obtain the value.

6. The Internet-based personalized education resource push system according to claim 5, characterized in that: Set priorities for resources based on the type of learning resources and the user's learning progress; build a hierarchical buffer queue with high, medium and low priorities, and store the resources to be pushed in the corresponding queue levels according to their priorities; if it is in a low load range, select the resources to be pushed from the high priority queue; if the high priority queue is empty, select resources from the medium priority queue for push; under high load conditions, suspend the push of low priority resources and record the push delay time. When the user load is reduced, adjust the order of the queue to push the delayed resources.

7. The Internet-based personalized education resource push system according to claim 1, characterized in that: The encoding of the learning state into a multi-layer state vector includes: collecting learning progress related data according to the user's learning behavior on different devices; encoding the user's learning state into a multi-layer state vector, each layer corresponding to a different type of learning data.

8. The Internet-based personalized education resource push system according to claim 7, characterized in that: Calculating the learning progress differences between different devices through the difference matrix includes: Based on the learning state vector of the current terminal and the learning state vectors of other terminals, the state difference of each dimension is calculated to generate a difference matrix; based on the timestamp information of the learning state, it is determined whether the data in the difference matrix comes from the same time slice. If the time slices are consistent, the difference value is directly calculated; if the time slices are different, the completion calculation is performed based on the nearest match principle; based on the data in the difference matrix, the degree of lag in learning progress between the current terminal and other terminals is calculated. If the lag value exceeds the set threshold, it is determined that the learning progress of the current terminal is lagging.

9. The Internet-based personalized education resource push system according to claim 1, characterized in that: The operation conflict is: compare the learning status update time of the current terminal with that of other terminals to determine whether there are learning status change records with overlapping time; collect learning status data of all terminals, establish a learning status comparison matrix, and analyze the conflict points between states. If the status of the same task in different terminals is inconsistent, it is determined to be an operation conflict.

10. The Internet-based personalized education resource push system according to claim 9, characterized in that: The conflict resolution based on the relevance of learning content includes: Based on the hierarchical relationship of learning resources, a task relevance network is established to mark the order and dependency between tasks; if an operation conflict occurs, the conflict resolution process is entered: query the task relevance network to determine the priority of the conflicting task; if the conflicting tasks have dependencies, the completed status of the predecessor task is retained first; if the conflicting tasks are at the same level, the status is selected according to the learning progress; based on the priority of the conflicting tasks, the reasonable status of the conflicting tasks is determined and synchronized to all terminals.

Images