Learning situation report generation method and system based on AI language model

Abstract

The invention belongs to the technical field of learning condition management, and particularly relates to a learning condition report generation method and system based on an AI language model, and the system specifically comprises a multi-modal learning condition data aggregation module, a learning condition data noise reduction and regularization module, a knowledge point mastery degree decision module, a knowledge point enhancement necessity analysis module, a learning condition report AI generation and output module and a teacher decision end. Individual skilled and weak knowledge points of students are accurately distinguished through a knowledge point mastery degree decision-making module, a knowledge point reinforcement necessity analysis module provides a basis for a teacher to accurately position knowledge points needing key reinforcement and reasonably distribute teaching resources, and a learning enthusiasm quantification output module accurately quantifies learning attitudes of the students. The course teaching management evaluation module comprehensively evaluates the course teaching management condition, and the learning condition report AI generation and output module generates a two-dimensional learning condition report of the whole class and the individual students, thereby providing powerful technical support for personalized teaching and precise teaching management.

Description

Technical Field

[0001] This invention relates to the field of learning management technology, specifically to a method and system for generating learning reports based on AI language models. Background Technology

[0002] With the deepening of educational informatization, the popularization of digital tools such as online learning platforms and smart campus systems has generated massive amounts of learning data. How to deeply mine this multi-source data to generate accurate learning reports has become a core requirement to help personalized teaching and precise teaching management. Learning analysis technology has gradually evolved from traditional grade statistics to multi-dimensional data-driven intelligent assessment, but it still faces many technical bottlenecks that need to be solved in practical applications.

[0003] For example, Chinese invention patent CN117114934A discloses a method, system, and electronic device for generating learning analysis reports. It mainly obtains students' basic information and related test questions, scores, and other data, generates class manuals and student manuals based on preset strategies, and pushes homework accordingly to achieve learning assessment and consolidation exercises. Although it achieves preliminary analysis and application of learning data, the data source is limited to structured information such as scores. It only completes the learning assessment through score statistics and does not combine process indicators such as answering efficiency and error correction to optimize the logic of knowledge point mastery assessment. It is difficult to truly reflect the students' actual mastery level of knowledge points.

[0004] Furthermore, existing technologies often focus on analyzing the learning situation of individual students, lacking accurate identification of common weak knowledge points at the class level, and failing to construct quantitative models for attitude dimensions such as learning enthusiasm. This results in insufficient guidance in the generated learning reports, making it difficult to meet the actual needs of precise teaching and personalized management. Consequently, teachers are unable to fully grasp the teaching situation in their classes and are unable to formulate targeted teaching optimization strategies and resource allocation plans.

[0005] Therefore, a solution is proposed. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for generating learning reports based on AI language models, so as to solve the technical defects mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a learning report generation system based on an AI language model, comprising a multimodal learning data aggregation module, a learning data noise reduction and normalization module, a knowledge point mastery decision module, a knowledge point reinforcement necessity analysis module, an AI-generated learning report output module, and a teacher decision-making terminal;

[0008] The multimodal learning data aggregation module collects multi-source data from students' learning process and forms an initial learning dataset through format unification and deduplication. The learning data denoising and normalization module is used to remove noisy data from the initial learning data, supplement missing values, and standardize data ranges to generate and output a normalized learning dataset.

[0009] The knowledge point mastery decision module assesses students' mastery of each knowledge point based on the normalized learning dataset, generating a visualized knowledge point mastery matrix. The knowledge point reinforcement necessity analysis module analyzes the reinforcement necessity of each knowledge point in the corresponding course of the class, generating high or low reinforcement necessity signals for the corresponding knowledge points, and sending these signals to the learning report AI generation output module. The learning report AI generation output module generates an overall learning report for the corresponding course in the class and individual learning reports for each student for the corresponding course, and sends the generated learning reports to the teacher's decision-making end.

[0010] Furthermore, the multimodal learning data aggregation module pre-sets three types of data collection sources: knowledge point interaction source, ability performance source, and behavior trajectory source. It connects to the three types of data collection sources through API interface, collects static and dynamic data in real time, and calls the built-in data format converter to uniformly convert different format data into standard JSON format.

[0011] A unique index is built based on "student ID + data generation timestamp + data type". Duplicate data is removed by hash value comparison. Finally, an initial learning data set containing fields of "student identifier - data type - data content - timestamp" is generated. The initial learning data set is then transmitted to the learning data noise reduction and normalization module.

[0012] Furthermore, the learning data noise reduction and normalization module receives the initial learning dataset, filters invalid data through a preset noise identification rule base, including data with abnormal time, invalid content, and logically contradictory data; it uses hierarchical interpolation to supplement missing data, and performs standardization processing on all data, finally generating and outputting the normalized learning dataset.

[0013] Furthermore, the specific operation process of the knowledge point mastery decision module is as follows:

[0014] Let i be the identifier of the corresponding student in the class, and i = {1, 2, ..., n}, where n is the total number of students in the corresponding class. Let k be the identifier of the knowledge point, and k = {1, 2, ..., m}, where m is the total number of knowledge points in the corresponding course. Obtain the average score of student i on the questions related to knowledge point k within the period and mark it as the knowledge point score coefficient Sik. Obtain the average answering time of student i on the questions related to knowledge point k within the period and mark it as the knowledge point answer efficiency coefficient Tik. Obtain the proportion of the incorrect questions of student i on knowledge point k that were answered correctly again within the period and mark it as the error correction rate Cik.

[0015] The mastery coefficient Mik of student i for knowledge point k is calculated by mastering the quantification formula. The mastery coefficient Mik is compared with the preset mastery coefficient threshold. If the mastery coefficient Mik exceeds the preset mastery coefficient threshold, knowledge point k is marked as a proficient knowledge point of student i. If the mastery coefficient Mik does not exceed the preset mastery coefficient threshold, knowledge point k is marked as a weak knowledge point of student i. An n×m "knowledge point mastery matrix" is generated and sent to the knowledge point reinforcement necessity analysis module and the learning report AI generation output module.

[0016] Furthermore, the specific analysis process of the knowledge point reinforcement necessity analysis module includes:

[0017] Obtain the labeling information of all students in the class regarding knowledge point k within the period, calculate the ratio of the number of students corresponding to the mastery of knowledge points to the value n to obtain the pass rate test value, and calculate the average of the mastery coefficients of all students regarding knowledge point k to obtain the mastery performance value.

[0018] The reinforcement judgment value is obtained by weighted summation of the pass rate detection value and the mastery performance value. The reinforcement judgment value is compared with the preset reinforcement judgment threshold. If the reinforcement judgment value exceeds the preset reinforcement judgment threshold, a low reinforcement necessity signal for knowledge point k is generated; if the reinforcement judgment value does not exceed the preset reinforcement judgment threshold, a high reinforcement necessity signal for knowledge point k is generated.

[0019] Furthermore, the knowledge point reinforcement necessity analysis module communicates with the course teaching management evaluation module. The course teaching management evaluation module will conduct a comprehensive evaluation of the teaching management status of the corresponding courses in the class, and through analysis, determine whether to generate the corresponding course optimization management signal, and send the comprehensive course teaching management evaluation information to the learning report AI generation output module.

[0020] Furthermore, the specific analysis process for the course teaching management evaluation module is as follows:

[0021] The number of knowledge points corresponding to the high necessity signal in the corresponding course is obtained and the ratio of the number of knowledge points is calculated with the value m to obtain the reinforcement urgency value. The ratio of the preset reinforcement judgment threshold to the reinforcement judgment value of the corresponding knowledge point is calculated to obtain the back test value. The average of the back test values ​​of all knowledge points in the corresponding course is calculated to obtain the reinforcement summary value. The reinforcement urgency value and the reinforcement summary value are compared with the preset reinforcement urgency threshold and the preset reinforcement summary threshold respectively. If the reinforcement urgency value or the reinforcement summary value exceeds the corresponding preset threshold, the optimization management signal of the corresponding course in the class is generated.

[0022] If neither the reinforcement emergency value nor the reinforcement summary value exceeds the corresponding preset threshold, the enthusiasm coefficient of the corresponding course is obtained. The course teaching and management evaluation value is calculated by weighted summation of the reinforcement emergency value, reinforcement summary value and enthusiasm anomaly coefficient. The course teaching and management evaluation value is compared with the preset course teaching and management evaluation threshold. If the course teaching and management evaluation value exceeds the preset course teaching and management evaluation threshold, the optimization management signal of the corresponding course in the class is generated.

[0023] Furthermore, the course teaching management and evaluation module communicates with the learning motivation quantification output module. The learning motivation quantification output module is used to analyze the learning motivation of all students in the class for the corresponding course one by one, determine the motivation level of each student for the corresponding course, and generate the motivation anomaly coefficient for the corresponding course through analysis, and send the motivation anomaly coefficient for the corresponding course to the course teaching management and evaluation module.

[0024] Furthermore, the specific analysis process of the learning motivation quantification output module is as follows:

[0025] The learning time of student i for each knowledge point in the corresponding course within the period is collected and summed to obtain the knowledge point learning time value Ti. The number of active interactions of student i in the corresponding course within the period is collected and marked as the course main interaction value Ii. The number of tasks completed by student i within the period is collected and the ratio of the number of tasks to the total number of tasks is calculated to obtain the task completion rate Qi.

[0026] The student i's motivation score Si for the corresponding course is calculated. The motivation score Si is compared with the set threshold Smin. If Si ≥ Smin, the student i is marked as a motivated student for the corresponding course. If Si < Smin, the student i is marked as a non-motivated student for the corresponding course. The motivation marking information of all students is sent to the learning report AI generation output module.

[0027] The number of non-active students in the class corresponding to the corresponding course is obtained and the ratio of this number to the value n is calculated to obtain the non-active number detection value. The ratio of Smin to Si is marked as the negative motivation value. The negative motivation values ​​of all students for the corresponding course are averaged to obtain the negative motivation table value. The negative motivation coefficient is obtained by weighted summation of the non-active number detection value and the negative motivation table value.

[0028] This invention also proposes a learning report generation method based on an AI language model, comprising the following steps:

[0029] Step 1: Collect multi-source data from students' learning process, and form an initial learning data set through format standardization and deduplication.

[0030] Step 2: Remove noisy data from the initial learning data, add missing values, and standardize the data range to generate and output the normalized learning data set;

[0031] Step 3: Based on the normalized learning dataset, assess students' mastery of each knowledge point and generate a visualized knowledge point mastery matrix;

[0032] Step 4: Analyze the necessity of strengthening each knowledge point in the corresponding course for the class, and generate signals indicating high or low necessity for strengthening the corresponding knowledge points through analysis.

[0033] Step 5: Generate overall learning reports and individual learning reports for the corresponding courses in the class and send them to the teacher's decision-making end.

[0034] Compared with the prior art, the beneficial effects of the present invention are:

[0035] 1. In this invention, a multi-modal learning data aggregation module collects data from multiple sources and unifies the format; a learning data noise reduction and normalization module ensures data quality through multiple processing steps; a knowledge point mastery decision module accurately distinguishes between students' proficient and weak knowledge points, overcoming the one-sidedness of traditionally judging mastery solely by scores; and a knowledge point reinforcement necessity analysis module provides teachers with a basis for accurately identifying knowledge points that need to be strengthened and rationally allocating teaching resources. This facilitates the implementation of data-driven teaching optimization and significantly improves the efficiency of learning management and the quality of teaching.

[0036] 2. In this invention, the learning motivation quantification output module innovatively quantifies students' learning attitudes, while the course teaching management evaluation module integrates knowledge point reinforcement data and motivation anomaly coefficients to comprehensively evaluate the course teaching management status. This helps teachers adjust their teaching management plans in a timely manner, which is conducive to developing more targeted teaching strategies and management plans. It further improves the efficiency of learning management and teaching quality, provides strong technical support for personalized teaching and precise teaching management, and significantly enhances the comprehensiveness, accuracy, and practicality of learning analysis and teaching guidance. Attached Figure Description

[0037] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings;

[0038] Figure 1 This is a system block diagram of Embodiment 1 of the present invention;

[0039] Figure 2 This is a system block diagram of Embodiments 2 and 3 of the present invention;

[0040] Figure 3 This is a flowchart of the method in Embodiment 4 of the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] Example 1: As Figure 1 As shown, the AI ​​language model-based learning report generation system proposed in this invention includes a multimodal learning data aggregation module, a learning data noise reduction and normalization module, a knowledge point mastery decision module, a knowledge point reinforcement necessity analysis module, an AI-generated learning report output module, and a teacher decision-making terminal. The multimodal learning data aggregation module collects multi-source data from students during their learning process, and through format unification and deduplication processing, forms an initial learning dataset covering "knowledge point interaction - ability performance - behavior trajectory", solving the problems of "single data source and chaotic format" in existing technologies.

[0043] Specifically, the multimodal learning data aggregation module first presets three types of data collection sources: knowledge point interaction sources (such as text answers in the homework submission system, objective question options / subjective question answers in the test platform, and wrong question correction records), ability performance sources (such as classroom interaction voice-to-text conversion, group discussion collaboration logs, and essay / experiment report scoring data), and behavioral trajectory sources (such as learning platform login / logout timestamps, video learning progress bar dragging records, question bank practice times / duration statistics, classroom hand-raising / questioning records, and the difference between homework submission time and deadline).

[0044] Secondly, by connecting to the above three types of data collection sources through API interfaces, static data (such as student ID, class affiliation, and course outline knowledge point annotation table) and dynamic data (such as learning behavior logs refreshed every 5 minutes, real-time submitted homework / test data, and real-time records of classroom interactions) are collected in real time. The built-in data format converter is called to uniformly convert different format data (JSON, Excel, TXT, speech-to-text XML) into standard JSON format.

[0045] Subsequently, a unique index is constructed based on "student ID + data generation timestamp + data type". Duplicate data (such as multiple submissions of the same assignment or repeated uploads of the same classroom interaction) is removed by comparing hash values. Finally, an initial learning data set containing fields of "student identifier - data type - data content - timestamp" is generated and transmitted to the learning data noise reduction and normalization module.

[0046] The learning data denoising and normalization module is used to remove noisy data from the initial learning data, supplement missing values, and standardize the data range, generating and outputting a normalized learning dataset. This improves data quality, provides a highly reliable data foundation for subsequent analysis processes such as knowledge mastery, avoids analytical biases caused by noisy data, and ensures the preparedness of subsequent analysis results.

[0047] Specifically, firstly, the learning data noise reduction and normalization module receives the initial learning dataset and filters invalid data through a preset "noise recognition rule base". This includes data with abnormal time (such as objective question records with answer time < 3 seconds, single-day records with learning duration > 24 hours, and homework submission time earlier than assignment time, which are judged as misoperation or data abnormality), invalid content data (such as garbled text, subjective question answers containing only punctuation marks, and classroom interaction records without actual content, which are judged by UTF-8 encoding verification and text length threshold < 5 characters), and logically contradictory data (such as test scores of 100 points but with wrong question records, and learning video progress of 100% but viewing time < 5 minutes, which are judged by "score - number of wrong questions" and "progress - duration" logical verification).

[0048] Next, a hierarchical interpolation method is used to supplement the missing data. For example, if the missing data is "knowledge point test score", the supplementary value is calculated based on "student's score on other knowledge points in the same chapter (weight 0.6) + class average score on the same knowledge point (weight 0.4)"; if the missing data is "learning time", it is supplemented based on "student's average time in the same period of the previous 3 days"; if the missing data is "number of classroom interactions", it is supplemented based on "student's average number of interactions in the previous 2 classes (weight 0.7) + class average number of interactions (weight 0.3)".

[0049] Subsequently, all data is standardized. For example, scores of different tests (such as full marks of 50 and 100) are uniformly converted into a range of 0-100 using a linear mapping formula. The submission time of assignments (in minutes, positive numbers indicate overtime and negative numbers indicate early submission) is converted into a standardized time value of 0-1 (overtime ≤ 0 → 0, early submission ≥ 30 minutes → 1, with intermediate values ​​calculated using linear mapping). Finally, the standardized learning data set is generated and output.

[0050] The knowledge point mastery decision module, based on the normalized learning dataset, assesses students' mastery of each knowledge point and generates a visualized knowledge point mastery matrix. This addresses the shortcomings of existing technologies that rely solely on scores to determine mastery and neglect learning process indicators, making the knowledge point mastery analysis process more reasonable and the results more accurate. The specific operation process of the knowledge point mastery decision module is as follows:

[0051] Let i be the identifier of the corresponding student in the class, and i = {1, 2, ..., n}, where n is the total number of students in the corresponding class. Let k be the identifier of the knowledge point, and k = {1, 2, ..., m}, where m is the total number of knowledge points in the corresponding course. Obtain the average score of student i on the questions related to knowledge point k within the period and mark it as the knowledge point score coefficient Sik. Obtain the average answering time of student i on the questions related to knowledge point k within the period and mark it as the knowledge point answer efficiency coefficient Tik. Obtain the proportion of the incorrect questions of student i on knowledge point k that were answered correctly again within the period and mark it as the error correction rate Cik.

[0052] The mastery coefficient Mik of student i in knowledge point k is calculated by mastering the quantification formula, as follows: Where α, β, and γ are preset weighting coefficients; and the larger the value of the mastery coefficient Mik, the more proficient student i is in terms of overall mastery of knowledge point k.

[0053] The mastery coefficient Mik is compared with the preset mastery coefficient threshold. If the mastery coefficient Mik exceeds the preset mastery coefficient threshold, it indicates that student i has a relatively high overall mastery of knowledge point k, and knowledge point k is marked as a proficient knowledge point of student i. If the mastery coefficient Mik does not exceed the preset mastery coefficient threshold, it indicates that student i has a relatively weak overall mastery of knowledge point k, and knowledge point k is marked as a weak knowledge point of student i. An n×m "knowledge point mastery matrix" is generated and sent to the knowledge point reinforcement necessity analysis module and the learning report AI generation output module.

[0054] The knowledge point reinforcement necessity analysis module analyzes the reinforcement necessity of each knowledge point in a class's corresponding course. Through analysis, it generates high-needness or low-needness reinforcement signals for each knowledge point and sends these signals to the learning report AI generation output module. This accurately determines whether each knowledge point in a class's corresponding course needs targeted reinforcement, avoiding blind reinforcement in teaching. It provides teachers with a clear basis for developing targeted teaching plans and rationally allocating teaching resources, improving the relevance and efficiency of knowledge point teaching. The specific analysis process of the knowledge point reinforcement necessity analysis module is as follows:

[0055] Obtain the labeling information of all students in the class regarding knowledge point k within the period, calculate the ratio of the number of students corresponding to the mastery of knowledge points to the value n to obtain the pass rate test value, and calculate the average of the mastery coefficients of all students regarding knowledge point k to obtain the mastery performance value.

[0056] The reinforcement judgment value is obtained by weighted summation of the pass rate test value and the mastery performance value. Specifically, the pass rate test value and the mastery performance value are each assigned a corresponding preset weight value, and the pass rate test value and the mastery performance value are multiplied by the corresponding preset weight value. The sum of the two sets of products is marked as the reinforcement judgment value. It should be noted that the larger the reinforcement judgment value, the worse the overall mastery of knowledge point k in the corresponding course is.

[0057] The reinforcement judgment value is compared with the preset reinforcement judgment threshold. If the reinforcement judgment value exceeds the preset reinforcement judgment threshold, it indicates that the overall mastery of knowledge point k in the corresponding course is good, and there is no need to focus on strengthening the teaching of knowledge point k. In this case, a low necessity signal for strengthening knowledge point k is generated. If the reinforcement judgment value does not exceed the preset reinforcement judgment threshold, it indicates that the overall mastery of knowledge point k in the corresponding course is poor, and there is a need to focus on strengthening the teaching of knowledge point k. In this case, a high necessity signal for strengthening knowledge point k is generated.

[0058] The AI-generated learning report module generates overall learning reports for the corresponding courses in the class and individual learning reports for each student for the corresponding courses. The generated learning reports are then sent to the teacher's decision-making end, enabling the teacher to intuitively and comprehensively grasp the overall learning situation of the class and the individual learning situation of each student. This provides the teacher with intuitive and comprehensive reference for formulating teaching strategies, adjusting teaching plans, and carrying out personalized teaching, effectively improving the efficiency and pertinence of the teacher's teaching decisions.

[0059] Example 2: Figure 2 As shown, the difference between this embodiment and Embodiment 1 is that the knowledge point reinforcement necessity analysis module is connected to the course teaching management and evaluation module. The course teaching management and evaluation module will comprehensively evaluate the teaching management status of the corresponding course in the class, and generate an optimization management signal for the corresponding course through analysis. The comprehensive evaluation information of course teaching management is then sent to the learning report AI generation output module, effectively solving the problem of the inability to quantify learning attitudes in traditional technologies. This allows for accurate identification of individual student learning motivation differences, providing a clear basis for teachers to conduct targeted incentive guidance, and also providing key motivational data support for the course teaching management and evaluation module. The specific analysis process of the course teaching management and evaluation module is as follows:

[0060] The number of knowledge points corresponding to the high necessity signal in the corresponding course is obtained and the ratio of this number to the value m is calculated to obtain the reinforcement urgency value. The ratio of the preset reinforcement judgment threshold to the reinforcement judgment value of the corresponding knowledge point is calculated to obtain the back test value. The average of the back test values ​​of all knowledge points in the corresponding course is calculated to obtain the reinforcement summary value. The reinforcement urgency value and the reinforcement summary value are compared with the preset reinforcement urgency threshold and the preset reinforcement summary threshold respectively. If the reinforcement urgency value or the reinforcement summary value exceeds the corresponding preset threshold, it indicates that the teaching management performance of the corresponding course in the class is worse and the teaching management plan of the corresponding course needs to be optimized in time. Then, the optimization management signal of the corresponding course in the class is generated.

[0061] If neither the reinforcement urgency value nor the reinforcement summary value exceeds the corresponding preset threshold, the corresponding course's motivation coefficient is obtained. The course management evaluation value is calculated by weighted summation of the reinforcement urgency value, reinforcement summary value, and motivation anomaly coefficient. That is, the reinforcement urgency value, reinforcement summary value, and motivation anomaly coefficient are each assigned a corresponding preset weight value, and the reinforcement urgency value, reinforcement summary value, and motivation anomaly coefficient are each multiplied by the corresponding preset weight value. The sum of the three product results is marked as the course management evaluation value.

[0062] It should be noted that the higher the course management evaluation value, the worse the overall teaching management performance of the corresponding course in the class. By comparing the course management evaluation value with the preset course management evaluation threshold, if the course management evaluation value exceeds the preset course management evaluation threshold, it indicates that the overall teaching management performance of the corresponding course in the class is worse, and the teaching management plan for the corresponding course needs to be optimized in a timely manner, thus generating an optimization management signal for the corresponding course in the class.

[0063] Example 3: Figure 2 As shown, the difference between this embodiment and Embodiment 1 and Embodiment 2 is that the course teaching management and evaluation module is connected to the learning motivation quantification output module. The learning motivation quantification output module is used to analyze the learning motivation of all students in the class for the corresponding course one by one and determine the motivation level of each student for the corresponding course.

[0064] Furthermore, by analyzing and generating the corresponding course's motivational anomaly coefficient, and sending this coefficient to the course teaching management evaluation module, the system can comprehensively and accurately assess the teaching management status of the corresponding courses in a class, promptly identify problems in teaching management, provide scientific guidance for teachers to optimize course teaching management plans, and help improve the overall quality of course teaching management. The specific analysis process of the learning motivation quantification output module is as follows:

[0065] The learning time of student i for each knowledge point in the corresponding course within the period is collected and summed to obtain the knowledge point learning time value Ti. The number of times student i actively interacts in the corresponding course within the period (active interaction includes raising hands in class, asking questions on the platform, speaking in group discussions, etc.) is collected and marked as the course main interaction value Ii. The number of tasks completed by student i within the period is collected and the ratio with the total number of tasks (tasks include homework, quizzes, video learning, etc.) is calculated to obtain the task completion rate Qi.

[0066] The student i's motivation score Si for the corresponding course is calculated using the formula Si=e1×Ti+e2×Ii+e3×Qi, where e1, e2, and e3 are preset weighting coefficients. The larger the motivation score Si is, the more motivated the student i is in learning the corresponding course overall.

[0067] The motivation score Si is compared with the set threshold Smin. If Si ≥ Smin, it indicates that student i is more motivated to learn the corresponding course, and student i is marked as a motivated student for the corresponding course. If Si < Smin, it indicates that student i is not motivated to learn the corresponding course, and student i is marked as a non-motivated student for the corresponding course. The motivation marking information of all students is sent to the learning report AI generation output module.

[0068] The number of non-active students in the class corresponding to the corresponding course is obtained and the ratio of the non-active student to the value n is calculated to obtain the non-active student count. The ratio of Smin to Si is marked as the negative student count. The average of the negative student counts for the corresponding course of all students is calculated to obtain the negative student count.

[0069] The positive anomaly coefficient is calculated by weighted summation of the non-positive test values ​​and the positive negative table values. Specifically, each non-positive test value and the positive negative table value is assigned a corresponding preset weight value, and each non-positive test value and the positive negative table value is multiplied by its corresponding preset weight value. The sum of the two products is then marked as the positive anomaly coefficient. It should be noted that the larger the positive anomaly coefficient, the more active all students in the class are in learning the corresponding course.

[0070] Example 4: Figure 3 As shown, the difference between this embodiment and Embodiments 1, 2, and 3 is that the learning report generation method based on AI language model proposed in this invention includes the following steps:

[0071] Step 1: Collect multi-source data from students' learning process, and form an initial learning data set through format standardization and deduplication.

[0072] Step 2: Remove noisy data from the initial learning data, add missing values, and standardize the data range to generate and output the normalized learning data set;

[0073] Step 3: Based on the normalized learning dataset, assess students' mastery of each knowledge point and generate a visualized knowledge point mastery matrix;

[0074] Step 4: Analyze the necessity of strengthening each knowledge point in the corresponding course for the class, and generate signals indicating high or low necessity for strengthening the corresponding knowledge points through analysis.

[0075] Step 5: Generate overall learning reports and individual learning reports for the corresponding courses in the class and send them to the teacher's decision-making end.

[0076] The working principle of this invention is as follows: In use, the multi-modal learning data aggregation module collects multi-source data and unifies the format; the learning data noise reduction and normalization module further ensures data quality through noise removal, missing data filling, and standardization; the knowledge point mastery decision module calculates the mastery coefficient based on multi-dimensional data, accurately distinguishing the proficiency and weakness of individual students' knowledge points, thus overcoming the one-sidedness of traditionally judging mastery solely by scores; and the knowledge point reinforcement necessity analysis module accurately determines whether to generate reinforcement signals based on the overall knowledge point mastery data of the class, providing a basis for teachers to accurately locate the knowledge points that need to be strengthened and rationally allocate teaching resources.

[0077] Furthermore, the learning motivation quantification module innovatively quantifies students' learning attitudes, providing teachers with direction for targeted student motivation. The course teaching management evaluation module integrates knowledge point reinforcement data and motivation anomaly coefficients to comprehensively evaluate the course teaching management status, helping teachers adjust teaching management plans in a timely manner. The learning report AI generation output module integrates the analysis results of all modules to generate a dual-dimensional learning report for the class as a whole and for individual students, which is then pushed to the teacher's decision-making end. This allows teachers to intuitively and comprehensively grasp the learning situation, thereby formulating more targeted teaching strategies and management plans, improving the overall efficiency of learning management and teaching quality. It provides strong technical support for personalized teaching and precise teaching management, significantly improving the comprehensiveness, accuracy, and practicality of learning analysis and teaching guidance.

[0078] In this invention, the threshold, preset value, or preset range settings are for result comparison and analysis to determine whether the result is good or bad. The magnitude of these values ​​is determined by a combination of large-scale model analysis of sample data and human experience, and can also be appropriately adjusted based on seasonal or common-sense influence conditions. Similarly, the preset weight coefficients and influence factors are assigned specific values ​​based on the magnitude of each parameter's influence on the result, ultimately reflecting the impact on the result. These settings are also determined by a combination of large-scale model analysis of sample data and human experience, and can also be appropriately adjusted based on seasonal or common-sense influence conditions.

[0079] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, enabling those skilled in the art to better understand and utilize it. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. An AI language model-based learning situation report generation system, characterized by, The system comprises a multi-modal learning data aggregation module, a learning data noise reduction and normalization module, a knowledge point mastery decision module, a knowledge point reinforcement necessity analysis module, a learning report AI generation and output module, and a teacher decision end. The multi-modal learning data aggregation module collects multi-source data in the learning process of students, and forms an initial learning data set through format unification and de-duplication processing. The learning data noise reduction and normalization module is used for removing noise data in the initial learning data, supplementing missing values, and standardizing the data range, to generate a normalized learning data set and output. The knowledge point mastery decision module evaluates the mastery degree of students on each knowledge point based on the normalized learning data set, and generates a visual knowledge point mastery matrix. The knowledge point reinforcement necessity analysis module analyzes the reinforcement necessity of each knowledge point in the corresponding course of the class, and generates a reinforcement high necessity signal or a reinforcement low necessity signal for the corresponding knowledge point. The learning report AI generation and output module is used for generating an overall learning report and an individual learning report, and sending the generated learning report to the teacher decision end. 2.The AI language model-based learning situation report generation system of claim 1, wherein, The multi-modal learning data aggregation module presets three types of data collection sources, and connects the three types of data collection sources through API interface to collect static data and dynamic data in real time, and calls a built-in data format converter to convert different format data into standard JSON format. Based on "student ID + data generation timestamp + data type", a unique index is constructed, and repeated data is removed by comparing hash values, to finally generate an initial learning data set containing "student identification-data type-data content-timestamp" fields, and transmit the initial learning data set to the learning data noise reduction and normalization module. 3.The AI language model-based learning situation report generation system of claim 2, characterized in that, The learning data noise reduction and normalization module receives the initial learning data set, filters invalid data through a preset noise recognition rule library, supplements missing data by using a hierarchical interpolation method, and standardizes all data, to finally generate a normalized learning data set and output. 4.The AI language model-based learning situation report generation system of claim 1, wherein, The specific operation process of the knowledge point mastery decision module is as follows: Set the identification of the corresponding student in the class as i, and i={=1,2,...,n}, n is the total number of students in the corresponding class, and set the knowledge point identification as k, and k={1,2,...,m}, m is the total number of knowledge points of the corresponding course; the mastery coefficient of student i on knowledge point k is calculated, if the mastery coefficient exceeds the preset mastery coefficient threshold, knowledge point k is marked as a proficient knowledge point of student i; otherwise, knowledge point k is marked as a weak knowledge point of student i, and a "knowledge point mastery matrix" with a dimension of n x m is generated and sent to the knowledge point reinforcement necessity analysis module and the learning report AI generation and output module. 5.The AI language model-based learning situation report generation system of claim 4, characterized in that, The specific analysis process of the knowledge point reinforcement necessity analysis module is as follows: The reinforcement decision value is calculated by weighted sum of the pass rate detection value and the mastery performance value, if the reinforcement decision value exceeds the preset reinforcement decision threshold, the reinforcement low necessity signal of knowledge point k is generated; otherwise, the reinforcement high necessity signal of knowledge point k is generated. 6.The AI language model-based learning situation report generation system of claim 5, wherein, The knowledge point reinforcement necessity analysis module is connected in communication with the course teaching management evaluation module. The course teaching management evaluation module evaluates the teaching management condition of the corresponding course in the class, generates an optimization management signal for the corresponding course by analysis, and sends the course teaching management comprehensive evaluation information to the learning situation report AI generation output module. 7.The AI language model-based learning situation report generation system of claim 6, wherein, The specific analysis process of the course teaching management evaluation module is as follows: if the reinforcement emergency value or the reinforcement summary value exceeds the corresponding preset threshold, an optimization management signal for the corresponding course in the class is generated; if the reinforcement emergency value and the reinforcement summary value do not exceed the corresponding preset threshold, a course teaching management evaluation value is calculated by weighted summation of the reinforcement emergency value, the reinforcement summary value and the positivity abnormality coefficient; if the course teaching management evaluation value exceeds the preset course teaching management evaluation threshold, an optimization management signal for the corresponding course in the class is generated. 8.The AI language model-based learning situation report generation system of claim 7, wherein, The course teaching management evaluation module is connected in communication with the learning positivity quantification output module. The learning positivity quantification output module is used to analyze the learning positivity of all students in the class for the corresponding course one by one, judge the positivity grade of each student for the corresponding course, generate a positivity abnormality coefficient for the corresponding course by analysis, and send the positivity abnormality coefficient of the corresponding course to the course teaching management evaluation module. 9.The AI language model-based learning situation report generation system of claim 8, wherein, The specific analysis of the learning positivity quantification output module is as follows: the positivity score Si of student i for the corresponding course is calculated; if Si≥Smin, student i is marked as an active object of the corresponding course; otherwise, student i is marked as a non-active object of the corresponding course; the positivity abnormality coefficient is calculated by weighted summation of the non-active number detection value and the positivity negative table value.

10. The method for generating an academic information report based on an AI language model, using the system for generating an academic information report based on an AI language model according to any one of claims 1-9, characterized in that, The method comprises the following steps: Step one, collect and output the initial learning situation data set; step two, generate and output the regularized learning situation data set; step three, evaluate the mastery degree of students for each knowledge point and output the knowledge point mastery matrix; step four, analyze the reinforcement necessity of each knowledge point in the corresponding course of the class; step five, generate the overall learning situation report and the individual learning situation report and send them to the teacher decision end.

Images