Intelligent learning state supervision method and system based on big data

Abstract

The invention relates to the technical field of learning supervision, in particular to a learning state intelligent supervision method and system based on big data, and the method comprises the following steps: capturing the interaction data of a user on a learning platform, analyzing and classifying behavior data portraits, and recognizing a learning mode and configuring labels for behaviors through analyzing the page stay duration and jump behaviors. The course content is dynamically adjusted, the user demand is matched, the course display mode and the interaction frequency are optimized, and the course sequence in the learning path is adjusted. The system can monitor and analyze the behavior pattern of a learner in real time, including key indexes such as click frequency and page dwell time, and can analyze the learning habit of a student through detailed behavior data, thereby adjusting the teaching content and difficulty in real time, better adapting to the actual demand of the student, and improving the teaching efficiency. The instant response capability enables the teaching strategy to be more flexible, and the behaviors of the students can be rapidly optimized, so that the teaching effect is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of learning supervision, and particularly to an intelligent supervision method and system for learning status based on big data. Background Art

[0002] Learning supervision is a field involving educational technology and data analysis, mainly focusing on how to optimize and monitor the learning process through technical means, used to monitor learners' behaviors, evaluate progress, and predict learning outcomes, and can adjust teaching content and strategies according to the specific needs of students, provide feedback and support in real time, so as to improve learning efficiency and effect.

[0003] Among them, the intelligent supervision method for learning status based on big data involves using big data technology to monitor and manage students' learning status, and can understand students' learning habits, progress, and difficulties by analyzing students' behavior data on the online learning platform, so as to provide personalized teaching feedback and resources. Its main purpose is to achieve more accurate teaching intervention through real-time data monitoring and intelligent analysis, enhance the transparency and interactivity of the learning process, and the goal is to improve students' learning achievements and overall educational quality.

[0004] The existing technology mainly relies on preset teaching models and post-feedback in learning supervision, and cannot achieve real-time monitoring and immediate adjustment of the learning process. This static data processing method cannot effectively capture students' immediate reactions and changes during the learning process, resulting in lagging teaching intervention measures, unable to prescribe the right medicine precisely, and the ability to handle individual learning differences is also limited. In addition, due to the lack of in-depth analysis and application of real-time learning data, it is difficult to accurately understand the specific needs of each student, so it is impossible to provide personalized learning support and resource allocation, resulting in waste of learning resources and students' potential not being fully exerted. The learning path planning lacking real-time data support also does not conform to students' actual learning progress and effect, affecting the overall teaching quality and learning achievements. Summary of the Invention

[0005] The purpose of the present invention is to solve the disadvantages existing in the prior art, and propose an intelligent supervision method and system for learning status based on big data.

[0006] In order to achieve the above purpose, the present invention adopts the following technical scheme: An intelligent supervision method for learning status based on big data, including the following steps, S1: Based on the interaction data of users in the learning platform, capture the click frequency and page stay time in real time, monitor the time stamp and behavior occurrence order of each interaction, record the user behaviors under each node, and classify them according to behavior characteristics to obtain a behavior data portrait; S2: Based on the behavioral data portrait, analyze the user's page stay duration and jump behavior, identify the behavior patterns of continuous browsing or frequent page switching, and according to the association with the course structure hierarchy, configure corresponding tags for each behavior to obtain behavior classification metrics; S3: Use the behavior classification metrics to identify the learning nodes marked with task continuous delay and repeated attempts. According to the behavioral characteristics of the learning nodes, dynamically adjust the course content, modify the teaching difficulty or change the content presentation method to obtain teaching adaptability configuration; S4: According to the teaching adaptability configuration, retrieve the content display method and interaction frequency settings in the course resource pool, identify the teaching content that does not match the user behavior trend, analyze the display method and interaction frequency of the teaching content, and combine the user behavior pattern to optimize the course content specifically and generate teaching content adaptation data.

[0007] The improvement of the present invention is that the behavioral data portrait includes user interaction patterns, active time distribution, and course node activity. The behavior classification metrics include pattern recognition tags, behavior duration classification, and interaction frequency. The teaching adaptability configuration includes content difficulty level, display style adjustment, and interaction form adaptation. The teaching content adaptation data includes display method difference, interaction matching index, and teaching feedback adaptability.

[0008] The improvement of the present invention is that the acquisition steps of the behavioral data portrait are specifically as follows: S111: Based on the interaction data of users in the learning platform, capture the click frequency and page stay time of each user interaction in real time, monitor the interaction timestamp and the order of behavior occurrence, and according to the structure nodes of the course content, record the click details, page interaction time, and interaction order and frequency under each node to obtain user interaction records; S112: Use the user interaction records to integrate the behavioral data under each course node, analyze the click frequency, page stay time, interaction timestamp, and behavior order, and identify and classify the behavior patterns to obtain the behavioral data portrait.

[0009] The improvement of the present invention is that the acquisition steps of the behavior classification metrics are specifically as follows: S211: Based on the behavioral data portrait, obtain the access duration, path, and browsing order, remove outliers and duplicate records through data cleaning, calculate the average stay time of users on each page, and count the frequency of page jumps according to user activities to obtain the page jump frequency metric; S212: According to the page jump frequency metric, associate the user browsing behavior with the course hierarchy data of the page structure, calculate the ratio of same-level and cross-level jumps, and obtain the behavior jump structure offset result; S213: Analyze the user behavior according to the offset result of the behavior jump structure, classify each jump behavior into intervals, and configure corresponding tags for each browsing behavior based on the residence time of the user on the different pages, so as to obtain the behavior classification index.

[0010] The improvement of the present invention is that the obtaining step of the teaching adaptability configuration is specifically as follows: S311: Use the behavior classification index to match the task delay feature weight term with the corresponding data of the learning node, judge whether it meets the dual feature determination criteria of task continuous delay and multiple repeated attempts, screen the learning nodes that meet the conditions, and obtain the abnormal feature matching data; S312: Calculate the node content structure adaptability score according to the abnormal feature matching data, compare it with the grade standard, and obtain the course content adjustment index; S313: According to the course content adjustment index, combine and change the presentation order and complexity level of the course content of the corresponding node, rearrange the content presentation structure, and match the learning needs of the user to obtain the teaching adaptability configuration.

[0011] The improvement of the present invention is that the obtaining step of the teaching content adaptation data is specifically as follows: S411: According to the teaching adaptability configuration, retrieve the content display method and interaction frequency setting in the course resource pool, analyze the user browsing time, click frequency and interaction feedback trend, and identify the teaching content with deviation in the user behavior by comparing the display method and interaction frequency, so as to obtain the deviation content index; S412: Based on the deviation content index, analyze the display method and interaction frequency of the teaching content, identify the relative deviation of the display method and interaction frequency, and judge whether it exceeds the benchmark range to obtain the display interaction deviation data; S413: Based on the display interaction deviation data, perform targeted optimization on the course content to obtain the teaching content adaptation data.

[0012] The improvement of the present invention is that the step further includes: S5: According to the teaching content adaptation data, re-evaluate the content sorting information in the learning path, and adjust the order of the course content in the learning path according to the usage frequency and learning effect of the learning content to obtain the learning path optimization result; The learning path optimization result includes the learning path adjustment result, the efficiency improvement index, and the course contact frequency.

[0013] The improvement of the present invention is that the obtaining step of the learning path optimization result is specifically as follows: S511: Adapt data according to the teaching content, obtain the usage frequency data and learning effect data of each learning content in the learning path, summarize and standardize them respectively, and obtain the learning content sorting information by cross-comparing the usage frequency and learning effect; S512: Invoke the learning content sorting information, and adjust the order of the course content in the learning path according to the variation deviation between the frequency distribution, learning effect and current sorting position of the learning content, so as to obtain the optimized result of the learning path.

[0014] Intelligent learning status monitoring system based on big data, the system includes: The user behavior monitoring module captures the click frequency and page stay time in real time based on the interaction data of users in the learning platform, monitors the time stamp and behavior occurrence order of each interaction, records the user behavior under each node, and classifies it according to the behavior characteristics to obtain the behavior data portrait; The behavior pattern analysis module analyzes the page stay duration and jump behavior of users based on the behavior data portrait, identifies the behavior patterns of continuous browsing or frequent page switching, and configures corresponding labels for each behavior according to the association with the course structure level to obtain the behavior classification index; The teaching dynamic adjustment module uses the behavior classification index to identify the learning nodes marked as task continuous delay and repeated attempts, and dynamically adjusts the course content according to the behavior characteristics of the learning nodes, modifies the teaching difficulty or changes the content presentation method to obtain the teaching adaptability configuration; The content adaptation and optimization module retrieves the content display method and interaction frequency setting in the course resource pool according to the teaching adaptability configuration, identifies the teaching content that does not match the user behavior trend, analyzes the display method and interaction frequency of the teaching content, and pertinently optimizes the course content to generate the teaching content adaptation data; The learning path reconstruction module re-evaluates the content sorting information in the learning path according to the teaching content adaptation data, and adjusts the order of the course content in the learning path according to the usage frequency and learning effect of the learning content to obtain the optimized result of the learning path.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, it is possible to monitor and analyze the behavior patterns of learners in real time, including key indicators such as click frequency and page residence time. Through detailed behavioral data, the learning habits of students can be analyzed, so as to adjust the teaching content and difficulty in real time, better adapt to the actual needs of students. This ability of instant response makes the teaching strategy more flexible, can quickly optimize according to the behavior of students, improve the teaching effect. By intelligently analyzing the behavior classification indicators, the teaching content adaptation data can ensure a high degree of matching between the course and the student behavior, enhance the interactivity of the course, and thus improve the participation degree in the learning process to improve the learning efficiency and learning effectiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is the main step flow chart of the present invention; Figure 2 is the flow chart for obtaining the behavioral data portrait in the present invention; Figure 3 is the flow chart for obtaining the behavior classification indicators in the present invention; Figure 4 is the flow chart for obtaining the teaching adaptability configuration in the present invention; Figure 5 is the flow chart for obtaining the teaching content adaptation data in the present invention; Figure 6 is the flow chart for obtaining the learning path optimization result in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0018] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as limiting the present invention. In addition, in the description of the present invention, the meaning of "a plurality" is two or more, unless otherwise specifically defined.

[0019] Embodiment Please refer to Figure 1 , the present invention provides a technical solution: a method for intelligent supervision of learning status based on big data, including the following steps: S1: Based on the interactive data of users in the learning platform, the click frequency and page dwell time are captured in real time, the timestamp and behavior order of each interaction are monitored, and the user behavior under each node is recorded according to the structural nodes of the course content, and classified according to the behavioral characteristics to obtain the behavioral data portrait; S2: Based on the behavior data portrait, analyze the user's page dwell time and jump behavior, identify the behavior pattern of continuous browsing or frequent page switching, match the behavior with the course hierarchy according to the association with the course structure level, and configure the corresponding label for each behavior to obtain the behavior classification index; S3: Use behavioral classification indicators to identify learning nodes marked as continuous task delays and multiple repeated attempts. According to the behavioral characteristics of the learning nodes, dynamically adjust the course content, modify the teaching difficulty or change the content presentation method to match the user's learning needs and obtain teaching adaptability configuration; S4: According to the teaching adaptability configuration, retrieve the content display mode and interaction frequency settings in the course resource pool, identify the teaching content that does not match the user behavior trend, analyze the display mode and interaction frequency of the teaching content, and optimize the course content in a targeted manner based on the user behavior pattern to generate teaching content adaptation data; S5: Based on the teaching content adaptation data, re-evaluate the content sorting information in the learning path, and adjust the order of course content in the learning path according to the usage frequency and learning effect of the learning content to obtain the learning path optimization result.

[0020] Behavioral data portraits include user interaction patterns, active time distribution, and course node activity. Behavioral classification indicators include pattern recognition labels, behavior duration classification, and interaction frequency. Teaching adaptability configuration includes content difficulty level, presentation style adjustment, and interaction form adaptation. Teaching content adaptation data includes presentation method differences, interaction matching index, and teaching feedback adaptability. Learning path optimization results include learning path adjustment results, efficiency improvement indicators, and course contact frequency.

[0021] The structural nodes of the course content refer to the specific parts within the course, which can be chapters, class periods, exercises, quizzes, or any other form of course content. Each node represents an independent part of the course, and the nodes form a complete learning content in sequence. Nodes usually have specific learning objectives and content arrangements, aiming to help learners gradually master knowledge points; User behavior refers to a series of activities and interactions of learners on the online learning platform, which can include clicking on a certain link, the time spent staying on a specific page, participating in interactions (such as answering questions, commenting, watching videos, etc.), the order of page jumps, etc.; The course hierarchical relationship refers to the structured organization method of the course content, usually arranged according to the difficulty or logical order of knowledge. For example, a complete course can be divided into multiple levels, progressing from basic knowledge to advanced content; Learning nodes refer to specific links or learning tasks in the learning process, usually associated with specific learning activities or learning objectives. Learning nodes can be a certain task that learners need to complete, a small quiz, a learning video, a discussion, or a group activity, etc.

[0022] Please refer to Figure 2 , and the specific steps for obtaining the behavioral data portrait are as follows: S111: Based on the interaction data of users in the learning platform, capture the click frequency and page stay time of each user interaction in real time, monitor the interaction timestamps and the order of behavior occurrence, and record the click details and page interaction time under each node, as well as the interaction order and frequency according to the structural nodes of the course content, to obtain the user interaction record; Precisely record the timestamp for each user interaction event with the platform, obtain the time of each click, the page stay duration, and the order of occurrence of behaviors. This process involves two key parameters: click frequency and page stay time. Click frequency refers to the number of times each user interaction behavior occurs, while page stay time represents the total duration that the user stays on a certain page. Through the data, the details of each user interaction can be accurately recorded, constructing a sequential data stream of interactions. For example, suppose a user clicks three times during a certain learning process, and the click intervals are 5 seconds, 3 seconds, and 7 seconds respectively. During the click process, the platform will record the timestamp, page content, and behavior type (such as: simple click, stay, jump, etc.) at each moment. Next, relying on the structural nodes of the course content, the platform will classify the behavior data according to the content involved in the nodes. For example, the behaviors involved in a certain node include "watching a video", "viewing a document", "filling out a test question", and each behavior has a different stay time and interaction frequency. For example, under a certain video node, the user stays for 180 seconds, and the interaction frequency is two times (one time is playing the video, and the other time is adjusting the video volume). The behavior data will be separately recorded in the interaction record of this node, thus obtaining a detailed record of the user's behavior under each learning node. The data not only involves the time characteristics of user behaviors but also includes the order of occurrence of behaviors and the frequency of interactions.

[0023] S112: Utilize the user interaction records to integrate the behavior data under each course node, analyze the click frequency, page stay time, interaction timestamp, and behavior order, and identify and classify the behavior patterns to obtain a behavior data portrait; By integrating the collected user interaction records, the behavior data under each course node can be obtained. During the analysis process, first, the click frequency and page dwell time under each node are calculated. The click frequency refers to the number of times a user interacts with a node. For example, if the click frequency of a user on a certain node is 5 times, this indicates that the content of this node has been repeatedly operated or there are frequent user interaction behaviors. The page dwell time refers to the total dwell duration of this node. For example, if a user stays on this node for 300 seconds, the page dwell time of this node is 300 seconds. Next, based on the interaction timestamps and the order of behavior occurrence, the user's behavior patterns will be classified according to the temporal characteristics of the behavior. For example, assuming that on a certain learning node, the user shows a short browsing behavior (such as jumping away after only staying for 20 seconds), it will be classified as "low dwell time behavior". If the user stays for a long time (such as not jumping for 5 minutes continuously), it can be classified as "high dwell time behavior". Through classification analysis, the user's behavior patterns can be discovered. For example, some users show a frequent page jump pattern (such as switching pages every 30 seconds), while other users show a behavior pattern of continuous staying and few jumps. Through the analysis of the behavior order, it is also possible to identify whether the user progresses step by step or frequently jumps over some nodes during the learning process. Finally, based on the analysis results, a user's behavior data portrait is generated. For example, for a certain learning, if a user shows a behavior pattern of "high-frequency clicks" and "low dwell duration", the behavior portrait of this user will be defined as a "quick browsing user". The generation of the data portrait not only helps to accurately understand the user's learning habits but also provides data support for subsequent teaching content adjustment and path optimization.

[0024] Please refer to Figure 3 , the specific steps for obtaining the behavior classification indicators are as follows: S211: Based on the behavior data portrait, obtain the access duration, path, and browsing order. Remove outliers and duplicate records through data cleaning, calculate the average dwell time of users on each page, and conduct frequency statistics on the page jumps according to user activities to obtain the page jump frequency indicator; The access duration refers to the total time a user stays at each learning node. The path refers to the transfer process of a user from one node to another. The browsing order refers to the sequence in which a user accesses each learning item. By cleaning and processing the user behavior data, outliers and duplicate records will be removed. The criteria for determining outliers can be based on abnormal fluctuations in parameters such as stay time and page jump frequency. For example, a stay time greater than a certain threshold (such as 300 seconds) indicates system problems or invalid data, or an abnormally high jump frequency, indicating that the user's behavior does not conform to the normal learning path and needs to be excluded. Then, calculate the average stay time of users on each page. This calculation is based on the stay duration of each user on the page and the number of times the page is accessed. For example, if user A stays on page X for 240 seconds and user B stays on the same page for 300 seconds, and these two users access the page 3 times and 2 times respectively, then the average stay time of page X is: (240×3 + 300×2) / (3 + 2) = 264 seconds. After that, according to the user's activities, count the jump frequency of each page, that is, the number of times each page is jumped to other pages. Suppose the number of times page A is jumped to page B is 5 times and the number of times page B is jumped to page C is 3 times. Then the jump frequency can be used as an indicator to measure the jump correlation between page A and page B. Through these steps, the page jump frequency index is finally obtained. The index helps to evaluate the user's behavior pattern and understand which pages are more strongly correlated and which pages have a lower jump frequency.

[0025] S212: According to the page jump frequency index, correlate the user browsing behavior with the course hierarchy data of the page structure, and calculate the ratio of same-level and cross-level jumps. Use the formula: ; Obtain the behavior jump structure offset result , which represents the correlation and offset degree between the page jump behavior of a user within the website and the corresponding course structure hierarchy. Among them, is the total number of user jumps, and are adjustment coefficients, which respectively affect the proportion of same-level and cross-level jumps, is the same-level jump ratio in the th jump, reflecting the frequency of jumps between pages within the same course hierarchy, is the cross-level jump ratio in the th jump, referring to the frequency of jumps between pages in different course hierarchies, is the adjusted hierarchical position influence factor, used to adjust the jump influence between different page hierarchies, is the th jump of the user and the The course hierarchical position of a page is a hierarchical identifier that shows the positioning of the page in the course structure. is the number of pages included in each jump; According to the page jump frequency index, the user behavior data is sorted into a jump path sequence in chronological order, the course structure hierarchical data is called, and the page is mapped to the corresponding course level through index matching. Further, the jump type is judged by whether the hierarchical numbers of the pages before and after the jump are the same. If the numbers are the same, it is counted as a same-level jump; if different, it is counted as a cross-level jump. Suppose there are 3 segments of paths in the first jump, 2 of which are same-level and 1 is cross-level, then we get and , and the corresponding page levels are , and the sum is , suppose the structure adjustment factor of this jump is , and the adjustment coefficient is , substitute into the formula: ; For the second jump, suppose and and and , substitute into: ; For the third jump, suppose and and and , substitute into: ; Calculate the average structure deviation index of the three jumps: ; This structure jump deviation value represents the average degree of change in the course hierarchical structure during the user's page jumps and is used as the judgment basis for subsequent behavior classification to obtain the behavior jump structure deviation index.

[0026] S213: According to the behavior jump structure deviation result, analyze the user behavior, classify each jump behavior into intervals, and configure corresponding labels for each browsing behavior based on the user's stay time on different pages to obtain the behavior classification index; According to the offset index of the behavior jump structure, sort and divide the set of offset values generated by the user in multiple jump behaviors. Construct the interval boundaries based on the median value and upper and lower quantiles of the preset distribution. For example, consider the offset index value less than 0.10 as a stable type of jump, 0.10 - 0.20 as a medium jump, and more than 0.20 as a frequent jump type. Further, in combination with the page stay time of the user in various jump paths, extract the browsing duration sequences corresponding to different types of jump behaviors. If the average browsing duration is less than 30 seconds, it is marked as a quick exit type. If it is between 60 - 120 seconds, it is classified as a stable browsing type. In this way, a combined label set of jump structure types and stay behavior types is formed, such as frequent jump + quick exit type, stable jump + stable browsing type, etc. Generate label fields through automatic tagging processing to construct a complete behavior classification dimension system.

[0027] Please refer to Figure 4 , and the specific steps for obtaining the teaching adaptability configuration are as follows: S311: Use the behavior classification index to match the task delay feature weight item with the corresponding data of the learning node, and determine whether it meets the dual feature determination criteria of task continuous delay and multiple repeated attempts. Screen the learning nodes that meet the conditions to obtain abnormal feature matching data; The task delay feature weight item includes the task completion duration of the learning node, the user's jump frequency, and the behavior feature of multiple repeated attempts. For example, assume that the completion duration of a certain learning node is 10 minutes, and the user related to this node has made 3 repeated attempts on this node, with the completion time of each attempt being 5 minutes, 7 minutes, and 12 minutes. The "task delay feature weight" of this node will be obtained by calculating the average task completion time of this node. For example, the average task completion time is (5 + 7 + 12) / 3 = 8 minutes. This feature will be used as a weight item for task delay to participate in subsequent matching calculations. Then, it will be further determined whether the task of this node meets the dual feature determination criteria of task continuous delay and multiple repeated attempts. The dual feature standard means that the task is repeatedly attempted within a certain delay time, and this behavior feature is reflected in the behaviors of multiple users. By comparing the behavior data of each user, calculate the number of repeated attempts and the corresponding delay duration of each learning node to determine whether it meets this standard. For example, if the number of repeated attempts of a node is greater than 2 times and the duration of each attempt exceeds 7 minutes, then this node meets the dual feature standard of task delay and multiple repeated attempts. Through this analysis process, the learning nodes that meet the conditions can be screened out, further forming a set of learning nodes containing task delay and repeated attempt features, and finally obtaining abnormal feature matching data. For example, assume that in a certain course, there are multiple learning nodes with task delay and multiple repeated attempts, and the nodes will be automatically marked as abnormal feature nodes.

[0028] S312: Match the data according to the abnormal features and use the formula: ; Calculate the adaptability score of the node content structure , compare it with the grade standard to obtain the curriculum content adjustment index, where is the task difficulty level of the th task, reflecting the quantitative rating of the task in terms of difficulty, is the content organization dimension of the th task, quantitatively representing the structure and organization complexity of the task content, is the user feedback response volume of the th task, measuring the user's response to the task, such as completion rate, feedback enthusiasm, etc., is the number of tasks included under the target learning node; For three tasks T1, T2, and T3 under a certain learning node, their data are as follows: T1: Task difficulty level , content organization dimension , user feedback response volume ; T2: Task difficulty level , content organization dimension , user feedback response volume ; T3: Task difficulty level , content organization dimension , user feedback response volume ; Next, calculate according to the formula to calculate the numerator: ; ; ; Calculate the denominator: ; Then, substitute the numerator and denominator into the formula: ; The calculation result is the adaptability score of the node content structure , this result indicates that the adaptability of the node is within the medium range, and an adaptability adjustment plan needs to be further formulated according to this score, such as adjusting the task order or the content presentation order to make it more in line with the learning needs.

[0029] S313: Adjust the indicators according to the course content, combine and change the presentation order and complexity level of the course content of the corresponding nodes, rearrange the content presentation structure, and match the learning needs of users to obtain the teaching adaptability configuration; Based on the adaptability score of the content, rearrange the presentation order and complexity level of the nodes in the "medium adaptability level" interval of the adaptability score. First, according to the content organization dimension ( ) and the user feedback response volume ( ) of each task, sort the tasks. The content organization dimension of task T1 is 2.5, T2 is 3.0, and T3 is 2.0. Since the content organization dimension of task T2 is the largest, it will be displayed first, while the content organization dimension of T3 is the smallest, so it will be placed at the end. In addition, the user feedback response volume will also be considered, and the task with the most positive user response will be advanced. For example, although the content dimension of task T3 is small, due to its high user feedback response volume ( ), it is also displayed first. When specifically executed, the structure dimension of task T1 is 2.5, the feedback volume is 80, the structure dimension of task T2 is 3.0, the feedback volume is 65, and the structure dimension of task T3 is 2.0, the feedback volume is 90. After the priority sorting, the task display order becomes T2, T1, T3. Adjust the presentation order of the learning content according to the sorting of the tasks, complete the optimization of the task order and presentation method, and rearrange the content complexity so that the learning process of the tasks better meets the user needs. Finally, through the recombination and sorting of the tasks, a new learning task configuration plan is generated to ensure that users can learn the content according to the new adaptability configuration.

[0030] Please refer to Figure 5 , and the specific steps for obtaining the teaching content adaptability data are as follows: S411: According to the teaching adaptability configuration, retrieve the content presentation method and interaction frequency setting in the course resource pool, analyze the user browsing time, click frequency, and interaction feedback trend, and identify the teaching content with deviations in the user behavior by comparing the presentation method and interaction frequency to obtain the deviation content index; Retrieve the content display method and interaction frequency settings in the course resource pool. The content display method refers to the way of presenting content for each learning node in the course, such as text, video, image, etc., while the interaction frequency refers to the frequency of interacting with the content, such as clicking, commenting, rating, etc. Analyze the user's browsing time, click frequency, and interaction feedback trend. Capture the user's behavior characteristics by calculating the time the user stays at each node, the number of click behaviors, and the user's interaction feedback (such as submitting homework, answering questions). If at a certain video node, the user stays for 5 minutes within 10 minutes and pauses or adjusts the volume twice during the viewing process, this indicates that the user has a relatively high interaction frequency at this node. Also, compare the display method and interaction frequency based on the data to determine whether the user's behavior matches the current course content's display method and interaction frequency. For example, if a user's stay time at a video node is much lower than the average value, while the stay time at a text node is longer, it will be considered that the user is not interested in the video content or the video's display method is not attractive enough. Through this analysis, it is possible to identify teaching content with deviations. The deviation of the content is due to the mismatch between the display method and the user's preferences, or the interaction frequency does not meet the expectations. The retrieval index of the deviated content will be obtained through the identification of the deviated content.

[0031] S412: Based on the retrieval index of the deviated content, analyze the content display method and interaction frequency of the teaching content, identify the relative deviation between the display method and the interaction frequency, determine whether it exceeds the benchmark range, and obtain the display interaction deviation data; By comparing the display methods and interaction frequencies of different teaching nodes, determine whether there is an excessive deviation in the display method or interaction frequency of a certain node. For example, if in a certain learning, the display frequency of the video content is very low, while its corresponding interaction frequency is extremely high, it indicates that the video content cannot attract users or there are other problems. Calculate the deviation degree between each node by calculating the display frequency and interaction frequency of each node. Assume the display frequency is A and the interaction frequency is B, and calculate the difference between A and B. For example, represent the deviation by calculating the absolute difference |A - B|. Next, determine whether this deviation exceeds the set benchmark range. The setting of the benchmark range can be based on historical data or preset business rules. For example, if the preset benchmark range is 10%, when the deviation between the display frequency and interaction frequency of a certain node exceeds 10%, that node will be marked as deviated content. Finally, obtain the display interaction deviation data according to the analysis results.

[0032] S413: Based on the display interaction deviation data, conduct targeted optimization on the course content, using the formula: ; Obtain the teaching content adaptation data , used to evaluate the matching degree between the teaching content configuration and user behavior, where represents the original display mode of the th teaching content item, that is, before optimization, the display mode parameters of a certain teaching content (for example: display duration, page design, etc.). represents the adjusted display mode of the th teaching content item, that is, the display mode parameters after optimization based on user behavior analysis. represents the original interaction frequency of the th teaching content item, that is, before optimization, the interaction frequency parameters of a certain teaching content (for example: number of clicks, feedback frequency, etc.). represents the adjusted interaction frequency of the th teaching content item, that is, the interaction frequency parameters after optimization based on user behavior analysis. and are weight coefficients. used to balance the influence degree between the display mode and the interaction frequency. used to balance the relative influence of the display mode and the interaction frequency during the optimization process. is the total number of teaching content items; Adjust the deviated content. The adjusted content includes modifying the display mode and interaction frequency settings of the teaching content. The core goal of optimization is to make it more in line with the user behavior pattern. Use a formula to calculate the optimized teaching content adaptation data. Suppose there are 3 teaching content items with different deviations in display mode and interaction frequency respectively. Set the weight coefficients as: , ; The 1st teaching content item: The original display time is 10 minutes, and after optimization, it is 12 minutes; the original interaction frequency is 5 times, and after optimization, it is 7 times; The 2nd teaching content item: The original display time is 8 minutes, and after optimization, it is 10 minutes; the original interaction frequency is 4 times, and after optimization, it is 6 times; The 3rd teaching content item: The original display time is 15 minutes, and after optimization, it is 16 minutes; the original interaction frequency is 6 times, and after optimization, it is 8 times; According to the formula, calculate the deviation of each teaching content item: The 1st item: Display deviation = ; Interaction frequency deviation = ; The 2nd item: Display deviation = ; Interaction frequency deviation = ; The 3rd item: Display deviation = ; Interaction frequency deviation = ; Substitute the deviation value into the formula for calculation: ; ; The result shows that the teaching content adaptation data ( ) is 1.8, which means that after optimizing the display method and interaction frequency, the adaptation degree of the teaching content in terms of user behavior response has been quantified. The value reflects the matching degree of the adjusted content display method and interaction frequency with the user behavior trend. A lower CZ value indicates that the optimized content is more in line with the user behavior trend, while a higher CZ value means that the optimization effect is not obvious or there are still deviations.

[0033] Please refer to Figure 6 for the specific steps to obtain the learning path optimization result: S511: According to the teaching content adaptation data, obtain the usage frequency data and learning effect data of each learning content in the learning path, summarize and standardize them respectively, and obtain the learning content sorting information by cross - comparing the usage frequency and learning effect; The frequency data refers to the frequency of users accessing or operating each learning during the learning process. For example, if a certain learning node (such as a video) is accessed by 100 users within a week, and each user accesses the node 3 times, then the usage frequency data of this node is 300 times. The learning effect data is related to the learning outcome and is obtained by evaluating the performance of the trainees after this learning, such as exam scores, task completion, user feedback, etc. For example, if after learning the video, 80 out of 100 trainees completed the corresponding test and the average score was 85 points, it indicates that the learning effect of this video node is good. Summarize the data and use standardization processing to ensure that the data can be compared under the same dimension. Standardization usually converts the data into z - score or normalized form. The standardized usage frequency and learning effect data can facilitate subsequent comparison and analysis. Then, by cross - comparing the usage frequency and learning effect, identify which learnings have higher learning effects at high usage frequencies and which ones have low learning effects despite excessive access. For example, if the usage frequency of a certain video is extremely high but the learning effect evaluation is low, it is considered that there is a place that needs to be optimized. After cross - comparison, generate the learning content sorting information.

[0034] S512: Invoke the learning content sorting information, and adjust the order of the course content in the learning path according to the variation deviation between the frequency distribution, learning effect and the current sorting position of the learning content to obtain the learning path optimization result; The frequency distribution of learning content refers to the frequency distribution of each learning content during the user's access process. If a certain node has a high frequency in the learning path, it will be evaluated whether it is in a suitable position, and the variation deviation between the learning effect and the current sorting position will also be analyzed to determine whether a certain node is in a high-frequency position due to its low learning effect. At this time, adjustment is required. If a learning content is located at the front end of the learning path, but its usage frequency and learning effect are much lower than those of other nodes, it will be considered to be adjusted to the back end of the path or the content and presentation method will be optimized. It is also necessary to calculate the variation deviation between learning contents. The deviation refers to the difference between the actual content sorting and the expected sorting. For example, if a certain learning content is expected to be ranked third, but the actual position is sixth, and there is a large difference in the learning effect and usage frequency of this node, it will be judged whether its position needs to be adjusted. Through analysis, the learning path optimization result is finally obtained, and the result will reflect the optimal arrangement order of each node in the learning path to ensure that students can obtain higher effects and more suitable learning experiences during the learning process.

[0035] Intelligent learning status supervision system based on big data, the system includes: The user behavior monitoring module captures the click frequency and page stay time in real time based on the interaction data of users on the learning platform, monitors the time stamp and behavior occurrence order of each interaction, records the user behavior under each node, and classifies it according to behavior characteristics to obtain a behavior data portrait. The behavior pattern analysis module analyzes the page stay duration and jump behavior of users based on the behavior data portrait, identifies the behavior patterns of continuous browsing or frequent page switching, and configures corresponding labels for each behavior according to the association with the course structure level to obtain behavior classification indicators. The teaching dynamic adjustment module uses the behavior classification indicators to identify the learning nodes marked as task continuous delay and repeated attempts, and dynamically adjusts the course content, modifies the teaching difficulty or changes the content presentation method according to the behavior characteristics of the learning nodes to obtain a teaching adaptability configuration. The content adaptation optimization module retrieves the content display method and interaction frequency settings in the course resource pool according to the teaching adaptability configuration, identifies the teaching content that does not match the user behavior trend, analyzes the display method and interaction frequency of the teaching content, and optimizes the course content specifically to generate teaching content adaptation data. The learning path reconstruction module re-evaluates the content sorting information in the learning path according to the teaching content adaptation data, and adjusts the order of the course content in the learning path according to the usage frequency and learning effect of the learning content to obtain the learning path optimization result.

[0036] The above are only the preferred embodiments of the present invention, and do not limit the present invention in other forms. Any person skilled in the relevant art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent supervision method for learning status based on big data, characterized in that, It includes the following steps: S1: Based on the interaction data of users in the learning platform, capture the click frequency and page stay time in real time, monitor the time stamp and behavior occurrence order of each interaction, record the user behavior under each node, and classify it according to the behavior characteristics to obtain a behavior data portrait; S2: Based on the behavior data portrait, analyze the page stay duration and jump behavior of users, identify the behavior patterns of continuous browsing or frequent page switching, and configure corresponding tags for each behavior according to the association with the course structure level to obtain behavior classification indicators; S3: Use the behavior classification indicators to identify the learning nodes marked with task continuous delay and multiple repeated attempts, and dynamically adjust the course content, modify the teaching difficulty or change the content presentation method according to the behavior characteristics of the learning nodes to obtain a teaching adaptability configuration; S4: According to the teaching adaptability configuration, retrieve the content display method and interaction frequency setting in the course resource pool, identify the teaching content that does not match the user behavior trend, analyze the display method and interaction frequency of the teaching content, and combine the user behavior pattern to optimize the course content specifically to generate teaching content adaptation data.

2. The intelligent supervision method for learning status based on big data according to claim 1, characterized in that The behavior data portrait includes user interaction patterns, active time distribution, and course node activity. The behavior classification indicators include pattern recognition tags, behavior duration classification, and interaction frequency. The teaching adaptability configuration includes content difficulty level, display style adjustment, and interaction form adaptation. The teaching content adaptation data includes display method differences, interaction matching indexes, and teaching feedback adaptability.

3. The intelligent supervision method for learning status based on big data according to claim 1, characterized in that, The specific steps for obtaining the behavior data portrait are as follows: S111: Based on the interaction data of users in the learning platform, capture the click frequency and page stay time of each user interaction in real time, monitor the interaction time stamp and behavior occurrence order, and record the click details and page interaction time under each node according to the structure nodes of the course content, as well as the interaction sequence and frequency to obtain user interaction records; S112: Use the user interaction records to integrate the behavior data under each course node, analyze the click frequency, page stay time, interaction time stamp, and behavior order, and identify and classify the behavior patterns to obtain a behavior data portrait.

4. The intelligent supervision method for learning status based on big data according to claim 1, wherein The specific steps for obtaining the behavior classification indicators are as follows: S211: Based on the behavior data portrait, obtain the access duration, path, and browsing order, remove outliers and duplicate records through data cleaning, calculate the average stay time of users on each page, and count the frequency of page jumps according to user activities to obtain a page jump frequency index; S212: According to the page jump frequency index, associate the user browsing behavior with the course hierarchy data of the page structure, calculate the ratio of same-level and cross-level jumps, and obtain the behavior jump structure deviation result; S213: According to the behavior jump structure deviation result, analyze the user behavior, classify each jump behavior into intervals, and configure corresponding tags for each browsing behavior according to the stay time of the user on different pages to obtain behavior classification indicators.

5. The intelligent supervision method for learning status based on big data according to claim 1, characterized in that The specific steps for obtaining the teaching adaptability configuration are as follows: S311: Using the said behavior classification metrics, match the task delay feature weight terms with the corresponding data of the learning nodes, determine whether the dual feature determination criteria of task continuous delay and multiple repeated attempts are met, screen the learning nodes that meet the conditions, and obtain abnormal feature matching data; S312: According to the abnormal feature matching data, calculate the node content structure adaptability score, compare it with the grade standard, and obtain the course content adjustment metrics; S313: According to the course content adjustment metrics, combine and change the presentation order and complexity level of the course content of the corresponding nodes, rearrange the content presentation structure, and match the learning needs of users to obtain the teaching adaptability configuration.

6. The intelligent supervision method for learning status based on big data according to claim 1, wherein The specific steps for obtaining the teaching content adaptation data are as follows: S411: According to the teaching adaptability configuration, retrieve the content display method and interaction frequency settings in the course resource pool, analyze the user browsing time, click frequency, and interaction feedback trend, and identify the teaching content with deviations in the user behavior by comparing the display method and interaction frequency, to obtain the deviation content index; S412: Based on the deviation content index, analyze the display method and interaction frequency of the teaching content, identify the relative deviation between the display method and interaction frequency, and determine whether it exceeds the benchmark range to obtain the display interaction deviation data; S413: Based on the display interaction deviation data, perform targeted optimization on the course content to obtain the teaching content adaptation data.

7. The intelligent supervision method for learning status based on big data according to claim 1, characterized in that The said steps further include: S5: According to the teaching content adaptation data, re-evaluate the content sorting information in the learning path, and adjust the order of the course content in the learning path based on the usage frequency and learning effect of the learning content to obtain the learning path optimization result; The learning path optimization result includes the learning path adjustment result, the efficiency improvement index, and the course contact frequency.

8. The intelligent supervision method for learning status based on big data according to claim 7, characterized in that The specific steps for obtaining the learning path optimization result are as follows: S511: According to the teaching content adaptation data, obtain the usage frequency data and learning effect data of each learning content in the learning path, perform summary and standardization processing on them respectively, and obtain the learning content sorting information by cross-comparing the usage frequency and learning effect; S512: Invoke the learning content sorting information, and adjust the order of the course content in the learning path based on the variation deviation between the frequency distribution, learning effect, and the current sorting position of the learning content to obtain the learning path optimization result.

9. The intelligent supervision system for learning status based on big data, characterized in that, The system is used to implement the intelligent supervision method for learning status based on big data according to any one of claims 1-8. The system includes: The user behavior monitoring module, based on the interaction data of users in the learning platform, real-time captures the click frequency and page stay time, monitors the time stamp and behavior occurrence order of each interaction, records the user behavior under each node, and classifies it according to the behavior characteristics to obtain the behavior data portrait; The behavior pattern analysis module, based on the behavior data portrait, analyzes the user's page stay duration and jump behavior, identifies the behavior patterns of continuous browsing or frequent page switching, and configures corresponding tags for each behavior according to the association with the course structure hierarchy to obtain the behavior classification metrics; The teaching dynamic adjustment module uses the behavior classification indicators to identify learning nodes marked as task continuous delay and multiple repeated attempts. According to the behavior characteristics of the learning nodes, it dynamically adjusts the course content, modifies the teaching difficulty or changes the content presentation method to obtain a teaching adaptability configuration; The content adaptation optimization module retrieves the content display method and interaction frequency settings in the course resource pool according to the teaching adaptability configuration, identifies teaching content that does not match the user behavior trend, analyzes the display method and interaction frequency of the teaching content, and conducts targeted optimization on the course content to generate teaching content adaptation data; The learning path reconstruction module re-evaluates the content sorting information in the learning path according to the teaching content adaptation data, and adjusts the order of the course content in the learning path based on the usage frequency and learning effect of the learning content to obtain an optimized learning path result.

Images