Intelligent agent-based learning behavior data analysis system and method

Abstract

The invention relates to the technical field of data processing, in particular to a learning behavior data analysis system and method based on an agent, and the method comprises the steps: determining knowledge points in learning content, generating a visual knowledge point graph, testing the mastering degree of a learner for the knowledge points, generating a visual knowledge point mastering graph, and analyzing the knowledge points. The method comprises the steps of generating an optimal learning task which is most adaptive to the state of a learner based on the mastery degree of the learner and a learning behavior library, collecting real-time learning feature data of the learner during task execution, sensing the emotion and attention state of the learner in real time, and generating a personalized learning path in combination with a knowledge point graph, learning behavior analysis and the state of the learner. Learning tasks are dynamically adjusted according to real-time data, and learning strategies are optimized according to feedback of learners. Through dynamic perception and adjustment capability, the problem that a learning behavior analysis system in the prior art lacks real-time perception and dynamic adjustment of the state of a learner can be effectively solved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an agent-based learning behavior data analysis system and method. Background Art

[0002] In today's information age, fields such as education, training, and online learning are booming, and the analysis of learners' personalized needs and learning behaviors has become a focus. However, existing technologies lack effective analysis and planning for learners' behaviors. This makes it difficult to conduct targeted learning behavior planning analysis based on learners' understanding of each knowledge point in the current learning content, and it is difficult to provide learners with learning behavior plans that maintain optimal learning efficiency.

[0003] Existing announcement number CN117094863B discloses a learning behavior intelligent analysis system based on smart learning, including: a knowledge point analysis module, which is used to determine the knowledge points in the current learning content and generate a visual knowledge point map; a learning behavior library, which stores a number of learning behaviors related to the knowledge points; a testing module, which is electrically connected to the knowledge point analysis module; a learning task formulation module, which is electrically connected to the learning behavior library and the testing module; and a learning behavior analysis module, which is electrically connected to the learning task formulation module. Learners can achieve maximum mastery of the learning content within a limited learning time, assist learners in maximizing their learning efficiency, and improve the accuracy and efficiency of learners' learning behavior management.

[0004] However, during the learning process, the learner's state may change, such as fluctuating emotions or distracted attention. Such systems may not be able to perceive these changes in real time and dynamically adjust learning tasks. An intelligent agent is an entity that can perceive its environment and make decisions based on this information. A key characteristic of an intelligent agent is its ability to perceive its environment and dynamically adjust its behavior. If the system could incorporate intelligent agent technology, enabling the agent to perceive the learner's state, such as emotions and attention, in real time and dynamically adjust learning tasks based on these states, the system's dynamic adaptability would be greatly improved. However, existing intelligent learning-based learning behavior analysis systems may lack this dynamic perception and adjustment capability. Summary of the Invention

[0005] The purpose of the present invention is to provide a learning behavior data analysis system and method based on intelligent agents, which solves the problem that the existing learning behavior intelligent analysis system based on smart learning lacks dynamic perception and adjustment capabilities.

[0006] To achieve the above object, the present invention provides an agent-based learning behavior data analysis method, comprising the following steps:

[0007] Determine the knowledge points in the current learning content and generate a visual knowledge point map;

[0008] Test and collect learners' mastery of each knowledge point in the current learning content, and generate a visual knowledge point mastery map based on the visual knowledge point map;

[0009] Based on the learner's mastery of each knowledge point in the current learning content and the learning behaviors stored in the learning behavior library, several learning behaviors that best suit the learner's current state are generated to form the optimal learning task;

[0010] The learning data generated by learners when performing the optimal learning task is collected and recorded as real-time learning feature data. Based on the real-time learning feature data and the optimal learning task, intelligent analysis is performed to determine whether the learner's learning behavior is in the optimal learning task. At the same time, the learner's emotions and attention state are perceived in real time.

[0011] Combining knowledge point maps, learning behavior analysis, and learner status, it generates personalized learning paths for learners and dynamically adjusts learning tasks based on learners' real-time learning data.

[0012] Automatically adjust learning strategies based on learners' feedback and learning behavior data to optimize subsequent learning tasks.

[0013] Among them, determine the knowledge points in the current learning content and generate a visual knowledge point map. The specific steps are:

[0014] Conduct text analysis on the current learning content and extract the knowledge points;

[0015] Build a knowledge graph based on the association between knowledge points;

[0016] Visualize the knowledge graph to form a visual knowledge point graph.

[0017] Among them, the learners' mastery of each knowledge point in the current learning content is tested and collected, and a visual knowledge point mastery map is generated based on the visual knowledge point map. The specific steps are:

[0018] Design test questions for each knowledge point;

[0019] Collect learners' responses to test questions;

[0020] Evaluate learners' mastery of each knowledge point based on their answers.

[0021] Among them, based on the learner's mastery of each knowledge point in the current learning content and the learning behaviors stored in the learning behavior library, several learning behaviors that best suit the learner's current state are generated to form the optimal learning task. The specific steps include:

[0022] Determine the learner's current learning status based on the learner's mastery of each knowledge point;

[0023] Combine the learning behavior data in the learning behavior library to select the learning behavior that best suits the learner's current learning status;

[0024] Combine the screened learning behaviors into the optimal learning task.

[0025] The learning data generated by learners when performing the optimal learning task is collected and recorded as real-time learning feature data. Based on the real-time learning feature data and the optimal learning task, an intelligent analysis is performed to determine whether the learner's learning behavior is in the optimal learning task. At the same time, the learner's emotions and attention state are perceived in real time. The specific steps include:

[0026] Collect learners' behavioral data when performing learning tasks through learning management systems, online learning platforms, or mobile learning apps, including task completion time, accuracy rate, page dwell time, and operation path;

[0027] Use cameras to perform facial expression recognition and collect learners’ emotional state data;

[0028] The microphone collects the learner's voice information, analyzes their tone and speaking speed, and assists in judging their emotional state;

[0029] Use eye tracking equipment to monitor learners’ gaze focus and eye movement trajectory to assess their attention state;

[0030] Collect learners' physiological and behavioral data through wearable devices;

[0031] Use photo search technology to collect learners' performance data in homework exercises through image recognition;

[0032] Preprocess the collected multi-source data, including data cleaning, normalization and dimensionality reduction;

[0033] Extract key features from the preprocessed data and store the extracted features.

[0034] Among them, after extracting key features from the preprocessed data and storing the extracted features, the specific steps include:

[0035] Compare the time learners spend completing tasks with the preset optimal time range to determine if there is a time deviation;

[0036] Evaluate whether the learner's task accuracy meets the expected standard and determine whether the learner has mastered the task content;

[0037] Analyze interaction frequency and dwell time to determine whether learners have difficulty or hesitation with certain knowledge points or tasks;

[0038] Evaluate whether learners' task progress meets expectations and determine whether they are lagging behind or ahead of schedule;

[0039] Based on the analysis results, comprehensively evaluate whether the learner's learning behavior is in the best learning task;

[0040] At the same time, the emotion and attention monitoring results are integrated to evaluate the learner's emotion and attention state.

[0041] The system combines knowledge point maps, learning behavior analysis, and learner status to generate personalized learning paths for learners and dynamically adjust learning tasks based on their real-time learning data. The specific steps include:

[0042] Determine the learner's current knowledge mastery level and knowledge weaknesses based on a visual knowledge point map;

[0043] Assess learners' learning styles and preferences based on the results of their learning behavior analysis;

[0044] Combine learners' emotions and attention states to predict learners' potential needs at different learning stages;

[0045] Plan a personalized learning path for learners from their current knowledge level to their target knowledge level, including the sequence of learning content, the difficulty of learning tasks, and recommendations for learning resources.

[0046] The system combines knowledge point maps, learning behavior analysis, and learner status to generate personalized learning paths for learners and dynamically adjust learning tasks based on their real-time learning data. Specific steps include:

[0047] Real-time monitoring of learners' learning data, including changes in task completion time, accuracy, emotional state, and attention state;

[0048] Compare real-time learning data with pre-set learning paths and learning task objectives to determine whether learners have deviated from the optimal learning path;

[0049] If learners deviate from the optimal learning path, the content, difficulty, or sequence of learning tasks will be dynamically adjusted based on the specific circumstances of the deviation to guide the learners back to the optimal learning path;

[0050] When it is detected that the learner is depressed or distracted, the difficulty of the learning task is automatically adjusted, motivating learning content is introduced, or the presentation of the learning task is adjusted to help the learner regain a positive learning state.

[0051] A learning behavior data analysis system based on an intelligent agent includes a graph generation module, a mastery evaluation module, an optimal learning task generation module, a real-time learning acquisition module, a personalized learning generation module, and an optimization module. The graph generation module is connected to the mastery evaluation module, the optimal learning task generation module is connected to the mastery evaluation module, the real-time learning acquisition module is connected to the optimal learning task generation module, the personalized learning generation module is connected to the real-time learning acquisition module, and the optimization module is connected to the optimal learning task generation module and the personalized learning generation module respectively.

[0052] The graph generation module is used to determine each knowledge point in the current learning content and generate a visual knowledge point graph;

[0053] The mastery evaluation module is used to test and collect the learner's mastery of each knowledge point in the current learning content, and generate a visual knowledge point mastery map based on the visual knowledge point map;

[0054] The optimal learning task generation module is used to generate a number of learning behaviors that are most suitable for the learner's current state based on the learner's mastery of each knowledge point in the current learning content and the learning behaviors stored in the learning behavior library to form the optimal learning task;

[0055] The real-time learning acquisition module is used to collect learning data generated by learners when performing optimal learning tasks, record them as real-time learning feature data, and intelligently analyze whether the learners' learning behavior is within the optimal learning task based on the real-time learning feature data and the optimal learning task. At the same time, it perceives the learners' emotions and attention status in real time;

[0056] The personalized learning generation module is used to combine the knowledge point map, learning behavior analysis and learner status to generate a personalized learning path for the learner and dynamically adjust the learning tasks according to the learner's real-time learning data;

[0057] The optimization module is used to automatically adjust the learning strategy based on the learner's feedback and learning behavior data to optimize subsequent learning tasks.

[0058] The present invention provides an agent-based learning behavior data analysis system and method. First, it identifies knowledge points in the learning content and generates a visual knowledge point map. Second, it tests the learner's mastery of the knowledge points and generates a visual knowledge point mastery map. Next, based on the learner's mastery and learning behavior database, it generates the optimal learning task that best matches the learner's state. Then, it collects real-time learning feature data while the learner is performing the task and perceives their mood and attention state in real time. Finally, it combines the knowledge point map, learning behavior analysis, and learner state to generate a personalized learning path. It dynamically adjusts learning tasks based on real-time data and optimizes learning strategies based on learner feedback. This dynamic perception and adjustment capability effectively addresses the problem of existing learning behavior analysis systems lacking real-time perception and dynamic adjustment of learner states. Through agent technology, it can monitor changes in the learner's mood, attention, and other states in real time and dynamically adjust the difficulty, content, and presentation of learning tasks accordingly. For example, if a learner is detected to be depressed or distracted, the system can automatically reduce the task difficulty or introduce motivational content to help the learner regain a positive learning state. Furthermore, the system dynamically adjusts learning paths based on real-time learning data, ensuring that learners are always on the optimal learning path, thereby improving learning efficiency and effectiveness. This dynamic adaptability, which is lacking in existing technologies, enables the present invention to better meet learners' personalized needs and enhance both learning experience and learning outcomes. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art.

[0060] Figure 1 This is a flowchart of the steps of the agent-based learning behavior data analysis method of the first embodiment of the present invention.

[0061] Figure 2 It is a principle block diagram of the agent-based learning behavior data analysis system of the second embodiment of the present invention.

[0062] In the figure: 201-graph generation module, 202-mastery assessment module, 203-optimal learning task generation module, 204-real-time learning acquisition module, 205-personalized learning generation module, 206-optimization module. DETAILED DESCRIPTION

[0063] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, but should not be understood as limiting the present invention.

[0064] The first embodiment of this application is:

[0065] See also Figure 1 ,in, Figure 1 This is a flowchart of the steps of the agent-based learning behavior data analysis method of the first embodiment of the present invention.

[0066] The present invention provides an agent-based learning behavior data analysis method, comprising the following steps:

[0067] S101: Determine each knowledge point in the current learning content and generate a visual knowledge point map;

[0068] Specifically, learning content, such as textbooks and course outlines, is converted to text, removing irrelevant symbols and stop words. Natural language processing techniques, such as TF-IDF and TextRank, are used to extract keywords from the text. These keywords are then used as potential knowledge points. Part-of-speech tagging and dependency parsing are used to further screen and confirm the knowledge points, ensuring they possess clear semantics and educational significance. Based on the extracted knowledge points, machine learning algorithms (such as association rule mining and graph neural networks) are used to analyze the logical relationships between them, such as "inclusion," "parallelism," and "cause and effect." A knowledge graph is constructed, with knowledge points as nodes and relationships as edges. Each node contains detailed information about the knowledge point, and each edge represents the type and weight of the relationship between knowledge points. The knowledge graph is optimized through expert review and user feedback to ensure its accuracy and practicality. Appropriate visualization tools, such as D3.js and Gephi, are used to design a visual interface for the knowledge graph. This interface should intuitively display the hierarchical structure and relationships between knowledge points. Interactive features, such as node clicks and path queries, are added to the knowledge graph to facilitate learners' exploration of the connections between knowledge points. Based on learners' learning progress and feedback, the display content of the knowledge graph is dynamically adjusted to highlight key knowledge points and learning paths.

[0069] S102: Testing and collecting learners' mastery of each knowledge point in the current learning content, and generating a visual knowledge point mastery map based on the visual knowledge point map;

[0070] Specifically, based on the visual knowledge point map, knowledge points are divided into different categories, such as basic knowledge points, advanced knowledge points, and comprehensive knowledge points. Test questions are designed for each knowledge point, including multiple-choice questions, fill-in-the-blank questions, and short-answer questions. Questions should cover different levels of difficulty for each knowledge point to ensure a comprehensive assessment of learners' mastery. The designed test questions are stored in a question bank, which can be dynamically updated and expanded to accommodate changes in learning content. A testing environment is provided for learners through a learning management system (LMS) or online learning platform. This testing environment should include automatic timing and scoring to ensure fairness and accuracy. After learners complete the test, their responses are collected, including time taken for each question, correctness of answers, and order of answers. This data will serve as an important basis for assessing learners' mastery of the knowledge points. The collected response data is stored in a database with efficient data reading and writing capabilities and data security mechanisms to ensure data integrity and availability. Based on the learners' responses, statistical analysis methods are used to assess their mastery of each knowledge point. Evaluation indicators include accuracy, average answering time, knowledge point coverage, etc. For example, accuracy can be calculated using the following formula: Based on the assessment results, learners' mastery of each knowledge point is classified into different levels, such as "Mastered," "Basically Mastered," and "Not Mastered." Based on the assessment results and the knowledge point map, a visual knowledge point mastery map is generated. Within the map, different colors or icons are used to represent the learner's mastery level for each knowledge point, visually demonstrating the learner's knowledge mastery. For example, green represents "Mastered," yellow represents "Basically Mastered," and red represents "Not Mastered."

[0071] S103: Based on the learner's mastery of each knowledge point in the current learning content and the learning behaviors stored in the learning behavior library, a number of learning behaviors that best suit the learner's current state are generated to form the optimal learning task;

[0072] Specifically, a learner's mastery of all knowledge points is comprehensively considered to determine their current learning status. For example, if a learner has "mastered" most basic knowledge points but "not mastered" advanced knowledge points, their learning status is "advanced stage." Learning behavior data, such as study time, study frequency, and learning preferences, is used to assess the learner's learning style, such as visual, auditory, or hands-on. A learning behavior library is established, containing multiple learning behaviors, each corresponding to a different knowledge point and learning stage. For example, for knowledge points that are "not mastered," learning behaviors such as "video explanation" and "interactive practice" can be provided. Based on the learner's current learning status and learning style, the most suitable learning behavior is selected from the learning behavior library. For example, for visual learners, "video explanation" is recommended, while for hands-on learners, "interactive practice" is recommended. The selected learning behaviors are prioritized to ensure that learners can learn in the optimal order. The selected learning behaviors are combined into learning tasks, each containing multiple learning behaviors covering the knowledge points that the learner currently needs to improve. The content and sequence of the learning tasks are dynamically adjusted based on the learner's learning progress and feedback. For example, if a learner performs well in a certain learning behavior, they can skip that behavior and go directly to the next one. Through the learning management system or online learning platform, the best learning tasks can be pushed to the learner, and real-time guidance and feedback can be provided.

[0073] S104: Collecting learning data generated by the learner when performing the optimal learning task, recording it as real-time learning feature data, and intelligently analyzing whether the learner's learning behavior is in the optimal learning task based on the real-time learning feature data and the optimal learning task. At the same time, the learner's emotions and attention state are perceived in real time;

[0074] Specifically, using a learning management system, online learning platform, or mobile learning app, record the time it takes learners to complete each task. Record the learner's accuracy rate in the task. Record the time learners spend on each page. Record the learner's operation path within the learning platform. Use a camera to capture the learner's facial expressions and analyze their emotional state using a deep learning algorithm. Use a microphone to collect learner voice information and analyze their intonation and speaking speed to assist in determining their emotional state. Use an eye-tracking device to monitor the learner's gaze focus and eye movement trajectory to assess their attention state. Use wearable devices to collect learner physiological data, such as heart rate and skin conductance. Use image recognition technology to collect learner performance data on homework exercises. Based on the collected multi-source data, remove invalid data and outliers. Normalize data from different sources to the same dimension to facilitate subsequent analysis. Use methods such as principal component analysis (PCA) to reduce data dimensionality and computational complexity. Extract key features, including time deviation, accuracy rate, interaction frequency, and average dwell time. If the time deviation exceeds a preset threshold (e.g., 20%), it is determined that the learner has a deviation in task completion time. If the accuracy rate is lower than a preset standard (e.g., 70%), the learner is judged to have not mastered the task content. If the interaction frequency is lower than a preset threshold (e.g., 5 times per minute), the learner may be judged to have difficulty or hesitation in grasping certain knowledge points or tasks. If the average dwell time is too long, the learner may be judged to have difficulty in grasping certain knowledge points or tasks. If the learner lags behind or advances in task progress, the difficulty or sequence of learning tasks is adjusted. Emotional states (e.g., happiness, sadness, anxiety, etc.) are output through deep learning models. Emotional states are analyzed through intonation and speech rate. Attention states are assessed through gaze focus and eye movement trajectories. Physiological data such as heart rate and skin conductance are used to assist in assessing attention states. Based on the above analysis results, a comprehensive assessment is made to determine whether the learner's learning behavior is in the optimal learning task. Based on adjustment suggestions, learning tasks are dynamically updated to ensure that learners are always in the optimal learning state. When a learner is detected to be depressed or distracted, the difficulty of the learning task is automatically adjusted, motivational content is introduced, or the task presentation method is adjusted. If learning behavior deviates from the optimal state, adjustment suggestions are generated.

[0075] S105: Combine knowledge point maps, learning behavior analysis, and learner status to generate personalized learning paths for learners and dynamically adjust learning tasks based on learners' real-time learning data;

[0076] Specifically, based on a visual knowledge point map, learners' mastery of each knowledge point is analyzed. The results of the knowledge point mastery test are used to determine the learner's current knowledge mastery level and weaknesses. Knowledge points with a mastery level below a preset threshold (e.g., 70%) are marked as weak points. Learners' learning styles and preferences are assessed based on their behavioral data when performing optimal learning tasks (e.g., task completion time, accuracy rate, page dwell time, etc.). Learning styles are categorized into visual, auditory, and hands-on, and classified based on learners' behavioral data. Real-time emotional and attention state data is combined to predict learners' potential needs at different learning stages. Machine learning algorithms (e.g., decision trees and neural networks) are used to build a demand prediction model, predicting learners' needs at different stages based on their emotional, attention, and learning behavior data. Based on the knowledge point map and the learner's current knowledge level, a personalized learning path is planned from the current level to the target level. The learning path is optimized based on the learner's learning style and preferences, including the sequence of learning content, the difficulty of learning tasks, and recommended learning resources. Starting with weak points in basic knowledge, the learning path gradually transitions to advanced knowledge points. Dynamically adjust task difficulty based on learner progress. Recommend learning resources that match learning styles, such as video explanations (visual), audio explanations (auditory), etc. Monitor learners' real-time learning data, including changes in task completion time, accuracy, emotional state, and attention state. Compare real-time learning data with the preset learning path and learning task goals to determine whether the learner has deviated from the optimal learning path. Deviation judgment formula: If the deviation exceeds a preset threshold (such as 10%), the learner is judged to have deviated from the optimal learning path. If the learner deviates from the optimal learning path, the content, difficulty or sequence of the learning task is dynamically adjusted according to the specific circumstances of the deviation. When the learner is detected to be depressed, the difficulty of the learning task is automatically adjusted, and motivational learning content is introduced or the task presentation method is adjusted. When the learner is detected to be distracted, the presentation method of the task is adjusted, such as adding interactive elements, adjusting the task rhythm, etc. The formula for adjusting task difficulty is: New difficulty = current difficulty × (1-α × depression coefficient), where α is the adjustment coefficient, and the depression coefficient is dynamically adjusted according to the emotional state. Reorder the learning tasks according to the learner's mood and attention state, and give priority to simple tasks to restore the learner's confidence.

[0077] S106: Automatically adjust learning strategies based on learner feedback and learning behavior data to optimize subsequent learning tasks.

[0078] Specifically, through questionnaires and learner evaluations, learners' subjective feedback on current learning tasks is collected, including information on satisfaction, difficulty, and learning experience. Learners' performance data on the learning tasks, such as task completion time, accuracy rate, and page dwell time, is analyzed as objective feedback. The collected learning behavior data is preprocessed, including data cleaning, normalization, and dimensionality reduction. Key features, such as task completion time, accuracy rate, interaction frequency, and dwell time, are extracted. Machine learning algorithms (such as cluster analysis and decision trees) are used to analyze learners' behavioral patterns and identify their behavioral characteristics during the learning process. Based on learner feedback and behavioral data, the effectiveness of the current learning strategy is evaluated. If learners' satisfaction with the learning task is low, or if task completion time is too long or accuracy rate is low, the current learning strategy needs to be adjusted. If learners' task completion time is too long or accuracy rate is low, the task difficulty should be reduced; if learners' task completion time is short and accuracy rate is high, the task difficulty should be appropriately increased. Based on learners' behavioral patterns and knowledge mastery, the order of learning tasks should be adjusted, prioritizing knowledge points with lower mastery. Adjust learning resource recommendations based on learners' learning styles and preferences, such as adding video tutorials and interactive exercises. If learners are feeling down or distracted, introduce motivational content or adjust task presentation to help them regain a positive learning state. Optimize subsequent learning tasks based on the recommended adjustments to ensure they align with the learner's current state. Monitor learners' real-time learning data and dynamically update learning tasks based on their progress and status to ensure they remain on the optimal learning path. Establish a feedback loop mechanism to regularly collect learners' feedback and behavioral data to continuously optimize learning strategies. The formula for adjusting task difficulty is: New difficulty = Current difficulty × (1-α × Feedback coefficient), where α is the adjustment coefficient, which is dynamically adjusted based on learners' feedback and behavioral data. The formula for adjusting task order is: New task order = Current task order + β × Priority adjustment, where β is the adjustment coefficient, which is dynamically adjusted based on learners' behavioral patterns and knowledge acquisition.

[0079] Through dynamic perception and adjustment capabilities, it is possible to effectively solve the problem in the existing technology that the learning behavior analysis system lacks real-time perception and dynamic adjustment of the learner's state. Through intelligent agent technology, it is possible to monitor the learner's mood, attention and other state changes in real time, and dynamically adjust the difficulty, content and presentation of the learning task accordingly. For example, when it is detected that the learner is depressed or distracted, the task difficulty can be automatically reduced or motivational content can be introduced to help the learner restore a positive learning state. In addition, the learning path can be dynamically adjusted according to the learner's real-time learning data to ensure that the learner is always on the best learning path, thereby improving learning efficiency and effectiveness. This dynamic adaptability is lacking in the existing technology, which enables the present invention to better meet the learner's personalized needs and improve the learning experience and learning outcomes.

[0080] The second embodiment of this application is:

[0081] Based on the first embodiment, please refer to Figure 2 ,in, Figure 2 It is a principle block diagram of the agent-based learning behavior data analysis system of the second embodiment of the present invention.

[0082] The present embodiment provides an agent-based learning behavior data analysis system, comprising a graph generation module 201, a mastery evaluation module 202, an optimal learning task generation module 203, a real-time learning acquisition module 204, a personalized learning generation module 205, and an optimization module 206.

[0083] For this specific implementation, the graph generation module 201 is connected to the mastery evaluation module 202, the optimal learning task generation module 203 is connected to the mastery evaluation module 202, the real-time learning acquisition module 204 is connected to the optimal learning task generation module 203, the personalized learning generation module 205 is connected to the real-time learning acquisition module 204, and the optimization module 206 is connected to the optimal learning task generation module 203 and the personalized learning generation module 205 respectively;

[0084] The graph generation module 201 is used to determine each knowledge point in the current learning content and generate a visual knowledge point graph;

[0085] The mastery evaluation module 202 is used to test and collect the learner's mastery of each knowledge point in the current learning content, and generate a visual knowledge point mastery map based on the visual knowledge point map;

[0086] The optimal learning task generation module 203 is used to generate a number of learning behaviors that are most suitable for the learner's current state based on the learner's mastery of each knowledge point in the current learning content and the learning behaviors stored in the learning behavior library to form the optimal learning task;

[0087] The real-time learning acquisition module 204 is used to collect learning data generated by learners when performing the optimal learning task, record it as real-time learning feature data, and intelligently analyze whether the learner's learning behavior is in the optimal learning task based on the real-time learning feature data and the optimal learning task. At the same time, it perceives the learner's emotions and attention status in real time;

[0088] The personalized learning generation module 205 is used to combine the knowledge point map, learning behavior analysis and learner status to generate a personalized learning path for the learner and dynamically adjust the learning tasks according to the learner's real-time learning data;

[0089] The optimization module 206 is used to automatically adjust the learning strategy according to the learner's feedback and learning behavior data, and optimize subsequent learning tasks.

[0090] Using the agent-based learning behavior data analysis system of this embodiment, the graph generation module 201 is first activated to perform text analysis on the current learning content and extract knowledge points. Based on the associations between knowledge points, a knowledge graph is constructed and visualized to form a visual knowledge point graph. The generated visual knowledge point graph serves as the basis for subsequent modules and is passed to the mastery assessment module 202. The mastery assessment module 202 receives the visual knowledge point graph transmitted by the graph generation module 201. Test questions are designed for each knowledge point and pushed to learners via a learning management system, online learning platform, or mobile learning app. The learner's response to the test questions is collected, including information such as the time taken to answer the questions and whether the answer was correct. Based on the response, the learner's mastery of each knowledge point is evaluated and a visual knowledge point mastery graph is generated. The learner's mastery data on each knowledge point is passed to the optimal learning task generation module 203. The optimal learning task generation module 203 receives the learner's mastery data on the knowledge point transmitted by the mastery assessment module 202. Combined with the learning behavior data stored in the learning behavior library, the learning behavior that best matches the learner's current learning state is screened. The screened learning behaviors are combined into an optimal learning task, and the optimal learning task is passed to the real-time learning acquisition module 204. The real-time learning acquisition module 204 receives the optimal learning task passed by the optimal learning task generation module 203. Learning data generated by the learner while performing the optimal learning task is collected through a learning management system, online learning platform, or mobile learning app, including task completion time, accuracy, page dwell time, and operation path. Simultaneously, facial expression recognition is performed using a camera to collect learner emotional state data; a microphone is used to collect learner voice information and analyze its intonation and speed to assist in determining emotional state; an eye tracking device is used to monitor the learner's gaze focus and eye movement trajectory to assess their attention state; wearable devices are used to collect learner physiological and behavioral data; and photo-search technology is used to collect learner performance data on homework exercises through image recognition. The collected multi-source data is preprocessed, including data cleaning, normalization, and dimensionality reduction. Key features are extracted from the preprocessed data and stored. Based on real-time learning feature data and optimal learning tasks, an intelligent analysis is performed to determine whether the learner's learning behavior is within the optimal learning task, and the analysis results are transmitted to the personalized learning generation module 205. The personalized learning generation module 205 receives the analysis results transmitted by the real-time learning acquisition module 204. Combining the knowledge point map, learning behavior analysis, and learner status, a personalized learning path is generated for the learner, including the sequence of learning content, the difficulty of learning tasks, and recommended learning resources. The learner's real-time learning data is monitored in real time, including changes in task completion time, accuracy, emotional state, and attention status.Real-time learning data is compared with the preset learning path and learning task objectives to determine whether the learner has deviated from the optimal learning path. If the learner deviates from the optimal learning path, the content, difficulty, or sequence of the learning tasks are dynamically adjusted based on the specific circumstances of the deviation to guide the learner back to the optimal learning path. When the learner is detected to be depressed or distracted, the difficulty of the learning task is automatically adjusted, motivating learning content is introduced, or the presentation of the learning task is adjusted to help the learner regain a positive learning state. The personalized learning path and dynamically adjusted learning tasks are transmitted to the optimization module 206. The optimization module 206 receives the personalized learning path and dynamically adjusted learning tasks transmitted by the personalized learning generation module 205. The learning strategy is automatically adjusted based on the learner's feedback and learning behavior data. The learner's subjective feedback and objective behavior data are collected to evaluate the effectiveness of the current learning strategy. Based on the evaluation results, learning strategy adjustment suggestions are generated, including adjustments to task difficulty, task sequence, resource recommendations, and incentive measures. The adjusted learning strategy is applied to subsequent learning tasks to optimize the subsequent learning tasks, ensure that the learning tasks match the learner's current state, and further improve learning outcomes and learning experience.

[0091] Through dynamic perception and adjustment capabilities, it is possible to effectively solve the problem in the existing technology that the learning behavior analysis system lacks real-time perception and dynamic adjustment of the learner's status. Through intelligent agent technology, it is possible to monitor the learner's mood, attention and other state changes in real time, and dynamically adjust the difficulty, content and presentation of the learning task accordingly. For example, when it is detected that the learner is depressed or distracted, the system can automatically reduce the difficulty of the task or introduce motivational content to help the learner restore a positive learning state. In addition, the system can also dynamically adjust the learning path based on the learner's real-time learning data to ensure that the learner is always on the best learning path, thereby improving learning efficiency and effectiveness. This dynamic adaptability is lacking in the existing technology, which enables the present invention to better meet the learner's personalized needs and improve the learning experience and learning outcomes.

[0092] The above disclosure is merely one or more preferred embodiments of the present application and is not intended to limit the scope of the present application. A person skilled in the art will understand that all or part of the processes of the above embodiments and equivalent changes made in accordance with the claims of the present application are still within the scope of the present application.

Claims

1. A method for analyzing learning behavior data based on an intelligent agent, characterized in that: The following steps are involved: Determine the knowledge points in the current learning content and generate a visual knowledge point map; Test and collect learners' mastery of each knowledge point in the current learning content, and generate a visual knowledge point mastery map based on the visual knowledge point map; Based on the learner's mastery of each knowledge point in the current learning content and the learning behaviors stored in the learning behavior library, several learning behaviors that best suit the learner's current state are generated to form the optimal learning task; The learning data generated by learners when performing the optimal learning task is collected and recorded as real-time learning feature data. Based on the real-time learning feature data and the optimal learning task, intelligent analysis is performed to determine whether the learner's learning behavior is in the optimal learning task. At the same time, the learner's emotions and attention state are perceived in real time. Combining knowledge point maps, learning behavior analysis, and learner status, it generates personalized learning paths for learners and dynamically adjusts learning tasks based on learners' real-time learning data. Automatically adjust learning strategies based on learners' feedback and learning behavior data to optimize subsequent learning tasks.

2. The method for analyzing learning behavior data based on an agent according to claim 1, wherein: Determine the knowledge points in the current learning content and generate a visual knowledge point map. The specific steps are: Conduct text analysis on the current learning content and extract the knowledge points; Build a knowledge graph based on the association between knowledge points; Visualize the knowledge graph to form a visual knowledge point graph.

3. The method for analyzing learning behavior data based on an agent according to claim 2, wherein: Test and collect learners' mastery of each knowledge point in the current learning content, and generate a visual knowledge point mastery map based on the visual knowledge point map. The specific steps are as follows: Design test questions for each knowledge point; Collect learners' responses to test questions; Evaluate learners' mastery of each knowledge point based on their answers.

4. The method for analyzing learning behavior data based on an agent according to claim 3, wherein: Based on the learner's mastery of each knowledge point in the current learning content and the learning behaviors stored in the learning behavior library, several learning behaviors that best suit the learner's current state are generated to form the optimal learning task. The specific steps include: Determine the learner's current learning status based on the learner's mastery of each knowledge point; Combine the learning behavior data in the learning behavior library to select the learning behavior that best suits the learner's current learning status; Combine the screened learning behaviors into the optimal learning task.

5. The method for analyzing learning behavior data based on an agent according to claim 4, wherein: The learning data generated by learners when performing the optimal learning task is collected and recorded as real-time learning feature data. Based on the real-time learning feature data and the optimal learning task, an intelligent analysis is performed to determine whether the learner's learning behavior is in the optimal learning task. At the same time, the learner's emotions and attention state are perceived in real time. The specific steps include: Collect learners' behavioral data when performing learning tasks through learning management systems, online learning platforms, or mobile learning apps, including task completion time, accuracy rate, page dwell time, and operation path; Use cameras to perform facial expression recognition and collect learners’ emotional state data; The microphone collects the learner's voice information, analyzes their tone and speaking speed, and assists in judging their emotional state; Use eye tracking equipment to monitor learners’ gaze focus and eye movement trajectory to assess their attention state; Collect learners' physiological and behavioral data through wearable devices; Use photo search technology to collect learners' performance data in homework exercises through image recognition; Preprocess the collected multi-source data, including data cleaning, normalization and dimensionality reduction; Extract key features from the preprocessed data and store the extracted features.

6. The method for analyzing learning behavior data based on an agent according to claim 5, wherein: After extracting key features from the preprocessed data and storing the extracted features, the specific steps include: Compare the time learners spend completing tasks with the preset optimal time range to determine if there is a time deviation; Evaluate whether the learner's task accuracy meets the expected standard and determine whether the learner has mastered the task content; Analyze interaction frequency and dwell time to determine whether learners have difficulty or hesitation with certain knowledge points or tasks; Evaluate whether learners' task progress meets expectations and determine whether they are lagging behind or ahead of schedule; Based on the analysis results, comprehensively evaluate whether the learner's learning behavior is in the best learning task; At the same time, the emotion and attention monitoring results are integrated to evaluate the learner's emotion and attention state.

7. The method for analyzing learning behavior data based on an agent according to claim 6, wherein: Combining knowledge point maps, learning behavior analysis, and learner status, we generate personalized learning paths for learners and dynamically adjust learning tasks based on their real-time learning data. The specific steps include: Determine the learner's current knowledge mastery level and knowledge weaknesses based on a visual knowledge point map; Assess learners' learning styles and preferences based on the results of their learning behavior analysis; Combine learners' emotions and attention states to predict learners' potential needs at different learning stages; Plan a personalized learning path for learners from their current knowledge level to their target knowledge level, including the sequence of learning content, the difficulty of learning tasks, and recommendations for learning resources.

8. The agent-based learning behavior data analysis method according to claim 6, characterized in that: Combining knowledge point maps, learning behavior analysis, and learner status, we generate personalized learning paths for learners and dynamically adjust learning tasks based on their real-time learning data. Specific steps include: Real-time monitoring of learners' learning data, including changes in task completion time, accuracy, emotional state, and attention state; Compare real-time learning data with pre-set learning paths and learning task objectives to determine whether learners have deviated from the optimal learning path; If learners deviate from the optimal learning path, the content, difficulty, or sequence of learning tasks will be dynamically adjusted based on the specific circumstances of the deviation to guide the learners back to the optimal learning path; When it is detected that the learner is depressed or distracted, the difficulty of the learning task is automatically adjusted, motivating learning content is introduced, or the presentation of the learning task is adjusted to help the learner regain a positive learning state.

9. An agent-based learning behavior data analysis system, applicable to the agent-based learning behavior data analysis method according to any one of claims 1 to 8, characterized in that: It includes a graph generation module, a mastery evaluation module, an optimal learning task generation module, a real-time learning acquisition module, a personalized learning generation module and an optimization module, wherein the graph generation module is connected to the mastery evaluation module, the optimal learning task generation module is connected to the mastery evaluation module, the real-time learning acquisition module is connected to the optimal learning task generation module, the personalized learning generation module is connected to the real-time learning acquisition module, and the optimization module is connected to the optimal learning task generation module and the personalized learning generation module respectively; The graph generation module is used to determine each knowledge point in the current learning content and generate a visual knowledge point graph; The mastery evaluation module is used to test and collect the learner's mastery of each knowledge point in the current learning content, and generate a visual knowledge point mastery map based on the visual knowledge point map; The optimal learning task generation module is used to generate a number of learning behaviors that are most suitable for the learner's current state based on the learner's mastery of each knowledge point in the current learning content and the learning behaviors stored in the learning behavior library to form the optimal learning task; The real-time learning acquisition module is used to collect learning data generated by learners when performing optimal learning tasks, record them as real-time learning feature data, and intelligently analyze whether the learners' learning behavior is within the optimal learning task based on the real-time learning feature data and the optimal learning task. At the same time, it perceives the learners' emotions and attention status in real time; The personalized learning generation module is used to combine the knowledge point map, learning behavior analysis and learner status to generate a personalized learning path for the learner and dynamically adjust the learning tasks according to the learner's real-time learning data; The optimization module is used to automatically adjust the learning strategy based on the learner's feedback and learning behavior data to optimize subsequent learning tasks.

Images