Self-adaptive teaching system and method based on generative large model
Through adaptive teaching methods based on generative large models, the teaching content is dynamically adjusted, and the problem that the existing system fails to match the needs of learners is solved, personalized learning paths are realized, and learning efficiency and teaching quality are improved.
Patent Information
- Application Number
- CN202411043222.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-11-19
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing teaching system fails to effectively consider individual differences, resulting in the teaching content and methods that do not match the learners' needs, knowledge background, cognitive level and interests, limiting learning potential and motivation.
Adaptive teaching methods based on generative large models are adopted, and learning data is collected in real time, the difficulty of exercises and model transformation coefficients are dynamically adjusted, and a personalized learning path is generated to ensure that the teaching content matches the learner's abilities and progress.
It improves learning efficiency and results, maintains learning motivation, reduces resource waste, realizes efficient utilization of teaching resources, provides a personalized learning experience, and optimizes the adaptability of teaching methods.
Smart Images

Figure CN118982445A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of generative large models, and in particular to an adaptive teaching system and method based on a generative large model. Background Art
[0002] With the growing demand for personalized and lifelong learning in the era of globalization and knowledge economy, and the technical support provided by technological advances such as educational technology, artificial intelligence, and big data analysis for personalized teaching, the research results of educational psychology and learning science have provided a theoretical basis for understanding learners' needs and optimizing the learning process. At the same time, the reform of education policies, the rise of online education platforms, and the society's attention to improving national competitiveness are all driving the education field to seek innovative teaching methods. In addition, social and economic changes require the workforce to have higher skills and adaptability, which further emphasizes the necessity of adaptive reforms in the education system. Taking all these factors into consideration, adaptive teaching methods have become an important development direction in the education field, aiming to improve the quality and effectiveness of teaching through data-driven decision-making and personalized learning paths, and to meet rapidly changing social and economic needs.
[0003] The patent document with publication number CN117437099A discloses a smart teaching system and teaching method based on a large language model. The system further realizes intelligent agents through the knowledge graph to assist the large language model, including intelligent tutor module, intelligent teaching assistant module and intelligent learning partner module. Before class, the intelligent teaching assistant module realizes the teacher's intelligent assisted lesson preparation, target dialogue-based autonomous preview and personalized learning path recommendation; during class, the intelligent tutor module realizes guided teaching, and the intelligent learning partner discussion assistant realizes group discussion-assisted teaching; after class, the intelligent teaching assistant module helps teachers generate targeted intelligent assisted evaluations, and intelligently assists subjective questions in judging, helping the target to achieve personalized learning. However, in the teaching process, the individual differences of the targets are not taken into account, and the phenomenon of using unified teaching content, methods or evaluation standards to treat all targets is considered a shortcoming in the field of education because it does not take into account the different learning needs, knowledge background, cognitive level and interests of the targets. Summary of the invention
[0004] To this end, the present invention provides an adaptive teaching system and method based on a generative large model, so as to overcome the limitation of the single teaching method commonly existing in the traditional teaching practice in the prior art, which leads to the problem of suppressing the learning potential and motivation of the target.
[0005] To achieve the above object, the present invention provides an adaptive teaching method based on a generative large model, comprising:
[0006] Step S1, real-time collection of the knowledge level, learning speed, correct rate of exercises and learning time of several targets;
[0007] Step S2, calculating a duration difficulty adjustment factor according to the learning duration and the preset standard duration, and calculating a target difficulty level of exercises suitable for the target to practice according to the knowledge level, learning speed, correct rate of exercises, learning duration, the preset standard duration and the calculated duration difficulty adjustment factor, and multiplying the target difficulty level by the preset model conversion coefficient to calculate an input difficulty level for model generation corresponding to the target difficulty level, and inputting the input difficulty level into the generative large model to generate a number of corresponding large model exercises;
[0008] Step S3, collecting the correctness and completion time of the large model exercises corresponding to the completion of several goals in real time, and calculating the completion time index and the correctness index according to the completion time and the correctness of the exercises, and calculating the comprehensive difficulty representation coefficient according to the correctness index and the completion time index;
[0009] Step S4, based on the comprehensive difficulty characterization coefficient, the preset satisfaction range, the learning time and the standard time, determine whether the difficulty needs to be adjusted, and when it is determined that the difficulty needs to be adjusted, select existing standard exercises of the same difficulty from the preset question bank, and based on the accuracy and completion time of the exercises after completing training for several goals, as well as the calculated standard comprehensive difficulty characterization coefficient and the comprehensive difficulty characterization coefficient, determine whether the model conversion coefficient needs to be adjusted, and adjust the model conversion coefficient based on the adjustment ratio calculated from the satisfaction range and the comprehensive difficulty characterization coefficient.
[0010] Further, the target difficulty level of the exercises suitable for the target to practice is calculated according to the knowledge level, learning speed, correct rate of exercises, learning time, preset standard time and the calculated time difficulty adjustment factor.
[0011]
[0012] Among them, D is the target difficulty level, K is the knowledge level, L is the learning speed, C is the correct rate of exercises, T is the learning time, and T' is the standard time. It is the duration difficulty adjustment factor.
[0013] Furthermore, the duration difficulty adjustment factor is calculated based on the learning duration and the preset standard duration.
[0014]
[0015] Among them, α is a constant that controls the degree of influence, and β is the upper limit of the duration difficulty adjustment factor.
[0016] Furthermore, the target difficulty level is multiplied by a preset model conversion coefficient to calculate an input difficulty level for model generation corresponding to the target difficulty level.
[0017] Furthermore, the completion time index is calculated according to the ratio of the calculated completion time and the preset minimum completion time.
[0018] Furthermore, the accuracy index is calculated based on the ratio of the accuracy of the exercise questions to a preset maximum accuracy.
[0019] Furthermore, the comprehensive difficulty representation coefficient is calculated based on the correct rate and completion time indicators of the exercises.
[0020] Dz=w1×Cz+w2×Tz,
[0021] Among them, Dz is the comprehensive difficulty representation coefficient, w1 is the accuracy weight coefficient, and w2 is the completion time weight coefficient.
[0022] Furthermore, by comparing the comprehensive difficulty representation coefficient with the preset satisfactory range,
[0023] If the comprehensive difficulty representation coefficient is not within a satisfactory range, it is determined that it is necessary to compare the learning time with the standard time.
[0024] If the learning time is less than the standard time, it is determined that there is no need to adjust the difficulty;
[0025] If the learning time is greater than or equal to the standard time, it is determined that the difficulty needs to be adjusted.
[0026] Furthermore, when it is determined that the difficulty needs to be adjusted, the difference between the standard comprehensive difficulty characterization coefficient and the comprehensive difficulty characterization coefficient is calculated, and based on the difference and a preset difference threshold, it is determined whether the model conversion coefficient needs to be adjusted.
[0027] If the difference is greater than the preset difference threshold, it is determined that the model conversion coefficient needs to be adjusted, the adjustment ratio is calculated according to the minimum value of the satisfactory range and the comprehensive difficulty characterization coefficient, and the model conversion coefficient is adjusted according to the adjustment ratio.
[0028]
[0029] Among them, P is the adjustment ratio, Mmin is the minimum value of the satisfactory range, and R' is the adjusted model conversion coefficient.
[0030] An adaptive teaching system based on a generative large model, based on the above-mentioned adaptive teaching method based on a generative large model, comprises:
[0031] The data collection module is used to collect the knowledge level, learning speed, correctness of exercises and learning time of several targets in real time;
[0032] A large model generation module, connected to the data acquisition module, is used to calculate a time difficulty adjustment factor according to the learning time and the preset standard time, and calculate a target difficulty level of exercises suitable for the target to practice according to the knowledge level, learning speed, correct rate of exercises, learning time, the preset standard time and the calculated time difficulty adjustment factor, and calculate an input difficulty level for model generation corresponding to the target difficulty level by multiplying the target difficulty level and the preset model conversion coefficient, and generate a number of corresponding large model exercises by inputting the input difficulty level into the generative large model;
[0033] The data collection module is also used to collect in real time the correctness and completion time of the large model exercises corresponding to the several targets;
[0034] The determination module is connected to the data acquisition module and the large model generation module respectively, and is used to calculate the completion time index and the accuracy index according to the completion time and the accuracy of the exercises respectively, and calculate the comprehensive difficulty characterization coefficient according to the accuracy index and the completion time index, and determine whether the difficulty needs to be adjusted according to the comprehensive difficulty characterization coefficient, the preset satisfaction range, the learning time and the standard time.
[0035] The adjustment module is connected to the determination module and is used to select existing standard exercises of the same difficulty from a preset question bank when it is determined that the difficulty needs to be adjusted, and determine whether the model conversion coefficient needs to be adjusted based on the accuracy and completion time of the exercises after completing training for several goals, as well as the calculated standard comprehensive difficulty characterization coefficient and the comprehensive difficulty characterization coefficient, and adjust the model conversion coefficient based on the satisfaction range and the adjustment ratio calculated by the comprehensive difficulty characterization coefficient when it is determined that the model conversion coefficient needs to be adjusted.
[0036] Compared with the prior art, the present invention has the following beneficial effects:
[0037] By collecting the target's learning data in real time, this method can provide each target with tailored learning materials to ensure that the teaching content matches the target's personal ability. By using the time-difficulty adjustment factor and the model conversion coefficient, the difficulty of the exercises can be dynamically adjusted to adapt to the target's learning progress and comprehension ability. By accurately matching the target's learning level, the target can reduce the time wasted on questions of inappropriate difficulty, thereby improving learning efficiency. According to the target's practice feedback, not only can the difficulty of the exercises be adjusted, but the model conversion coefficient can also be optimized, so that the teaching method can be continuously improved to better adapt to the target's learning needs. Through the comprehensive difficulty representation coefficient, the target's learning performance can be monitored and feedback and support can be provided in a timely manner. The use of the preset question bank and model conversion coefficient makes teaching resources more efficiently utilized and reduces resource waste.
[0038] Furthermore, by utilizing the learning data collected in real time, such as knowledge level, learning speed, correctness of exercises and learning time, combined with the preset standard time and time difficulty adjustment factor, the learning difficulty value suitable for the target is dynamically calculated. It can provide a personalized learning path for each target, dynamically adjust the learning difficulty to match the target's learning progress and ability, thereby improving learning efficiency and results.
[0039] Furthermore, by calculating the duration difficulty adjustment factor, the adaptive teaching system can adjust the teaching difficulty more finely to adapt to the target's current learning status and progress. This method helps maintain the target's learning motivation and ensures that they will not be frustrated by questions that are too difficult, nor lose interest because the questions are too simple.
[0040] Furthermore, this conversion mechanism provides a flexible way to adjust the difficulty of the exercises to match the target's ability and learning needs. By precisely controlling the conversion of difficulty values, a series of exercises can be generated that are neither too difficult nor too easy for the target, thereby promoting the target's learning progress and cognitive development.
[0041] Furthermore, by calculating the completion time indicator, the teaching system can evaluate the target's learning speed and understanding. If the completion time indicator shows that the target takes too long, it means that the question is too difficult or the target has difficulty understanding certain concepts; if the time is too short, it means that the question is too difficult or the target has mastered the knowledge points well. Such feedback can help the adaptive teaching system adjust the difficulty of subsequent exercises to ensure that the target learns at an appropriate challenge level.
[0042] Furthermore, the accuracy index Cz provides a key parameter for the adaptive teaching system to judge the target's mastery of the learning content. If the accuracy index is high, it means that the target has a good understanding and mastery of the relevant concepts, and the system can appropriately increase the difficulty of the questions to provide further challenges. Conversely, if the accuracy index is lower than expected, the system can adjust the teaching strategy, such as reviewing relevant knowledge points or providing additional exercises, to help the target strengthen understanding and learning.
[0043] Furthermore, the comprehensive difficulty representation coefficient provides a multi-dimensional evaluation tool, enabling the teaching system to more accurately grasp the target's learning status. By considering the two key factors of accuracy and completion time, the system can adjust the teaching difficulty more carefully to ensure that the exercises can effectively test the target's learning outcomes while maintaining the challenge of learning.
[0044] Furthermore, if the target finds the exercises too difficult or too easy, the system can optimize the learning experience by adjusting the difficulty to ensure that the target remains in the optimal learning zone. This helps maintain the target's learning motivation, improve learning efficiency, and promote cognitive development. By considering the length of study, the system can evaluate the target's adaptability to the difficulty of exercises. A shorter study time may indicate that the target has a good grasp of the material, while a longer study time means that the target is having difficulty. This approach helps the system make more accurate difficulty adjustment decisions, thereby improving the target's learning efficiency and quality.
[0045] Furthermore, this adjustment mechanism enables the adaptive teaching system to respond more accurately to the target's learning performance and adjust the teaching difficulty in a timely manner to meet the target's personalized learning needs. By optimizing the model conversion coefficient, the system can generate exercises that are more in line with the target's current learning level, avoiding problems of too high or too low difficulty, thereby improving the target's learning efficiency and learning outcomes.
[0046] Furthermore, the teaching content is adjusted in real time to meet the needs of different learners. The system dynamically generates exercises to ensure that learners always learn at an appropriate level of challenge, thereby improving learning efficiency and effectiveness. In addition, the system can reduce the workload of teachers, allowing them to focus more on teaching design and other creative tasks. As learners progress, the system will continue to optimize the learning path to help learners master knowledge at a pace that suits them. Ultimately, this system helps to improve the overall quality of education, achieve optimal allocation of educational resources, and bring learners a richer and deeper learning experience. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] Figure 1 is a flow chart of the adaptive teaching method based on the generative large model of this embodiment;
[0048] Figure 2 This is a decision logic diagram for determining whether to compare learning time in this embodiment;
[0049] Figure 3 This is a decision logic diagram for determining whether to adjust the difficulty in this embodiment;
[0050] Figure 4 This is a decision logic diagram for determining whether to adjust the model conversion coefficient in this embodiment;
[0051] Figure 5 It is a structural diagram of the adaptive teaching system of the generative large model of this embodiment. DETAILED DESCRIPTION
[0052] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0053] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.
[0054] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.
[0055] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0056] See also Figure 1 As shown, it is a flow chart of the adaptive teaching method based on the generative large model of this embodiment;
[0057] This embodiment provides an adaptive teaching method based on a generative large model, including:
[0058] Step S1, real-time collection of the knowledge level, learning speed, correct rate of exercises and learning time of several targets;
[0059] Step S2, calculating a duration difficulty adjustment factor according to the learning duration and the preset standard duration, and calculating a target difficulty level of exercises suitable for the target to practice according to the knowledge level, learning speed, correct rate of exercises, learning duration, the preset standard duration and the calculated duration difficulty adjustment factor, and multiplying the target difficulty level by the preset model conversion coefficient to calculate an input difficulty level for model generation corresponding to the target difficulty level, and inputting the input difficulty level into the generative large model to generate a number of corresponding large model exercises;
[0060] Step S3, collecting the correctness and completion time of the large model exercises corresponding to the completion of several goals in real time, and calculating the completion time index and the correctness index according to the completion time and the correctness of the exercises, and calculating the comprehensive difficulty representation coefficient according to the correctness index and the completion time index;
[0061] Step S4, based on the comprehensive difficulty characterization coefficient, the preset satisfaction range, the learning time and the standard time, determine whether the difficulty needs to be adjusted, and when it is determined that the difficulty needs to be adjusted, select existing standard exercises of the same difficulty from the preset question bank, and based on the accuracy and completion time of the exercises after completing training for several goals, as well as the calculated standard comprehensive difficulty characterization coefficient and the comprehensive difficulty characterization coefficient, determine whether the model conversion coefficient needs to be adjusted, and adjust the model conversion coefficient based on the adjustment ratio calculated from the satisfaction range and the comprehensive difficulty characterization coefficient.
[0062] First, the target's learning performance is monitored in real time, including knowledge level, learning speed, accuracy of exercises and learning time. Then, based on the collected data and preset standards, the time difficulty adjustment factor and the target difficulty level are calculated, and then the input difficulty level is generated to guide the generative large model to output exercises of corresponding difficulty. Then, the accuracy and completion time of the exercises generated by the target are collected, and the completion time index and accuracy index are calculated based on this. These indicators are combined to obtain the comprehensive difficulty representation coefficient. Then, based on the comprehensive difficulty representation coefficient and the preset satisfaction range, it is determined whether the difficulty of the exercises needs to be adjusted, and existing exercises are selected or adjusted from the question bank. Finally, based on the target's exercise performance and the comprehensive difficulty representation coefficient, it is determined whether the model conversion coefficient needs to be adjusted, and it is adjusted according to the satisfaction range to optimize the output of the generative large model and better meet the target's learning needs.
[0063] Assume that the target is learning mathematics and the adaptive teaching method described above is used to generate exercises suitable for him.
[0064] First, data collection:
[0065] The target knowledge level K is 70 points, the learning speed L is 1, the correct rate of practice questions C is 80%, and the learning time T is 30 minutes;
[0066] Duration difficulty adjustment factor calculation:
[0067] Assume that the standard duration T' is 45 minutes, α is 0.01, β is 0.5, and the learning duration T is 30 minutes, which is less than the standard duration T', so the duration difficulty adjustment factor F is: F = 0.01 × (30-45) = -0.15 (the absolute value is taken here, because the duration difficulty adjustment factor should be positive, indicating that the longer the learning time, the greater the difficulty);
[0068] Target difficulty level calculation:
[0069] Assume that the target difficulty level is calculated as: D = K + L × C + F × T;
[0070] Assume that the knowledge level K is 70, the learning speed L is 1, the accuracy C is 80%, the learning time T is 30 minutes, and the time difficulty adjustment factor F is 0.15;
[0071] Then the target difficulty level D is: D = 70 + 1 × 0.8 + 0.15 × 30 = 75.3;
[0072] Enter the difficulty level calculation:
[0073] Assume that the model conversion factor R is 1.2;
[0074] Input difficulty level D' is: D' = D × R = 75.3 × 1.2 = 90.36;
[0075] Exercise generation:
[0076] Using the input difficulty level D'=90.36, the generative large model generates several math exercises whose difficulty roughly corresponds to the target's current learning level and the required challenge;
[0077] Practice feedback collection:
[0078] After the target completes these exercises, his completion time and accuracy are collected;
[0079] Calculation of comprehensive difficulty representation coefficient:
[0080] Assume that the completion time index Tz and the accuracy index Cz are 95% and 70% respectively, the weight coefficient w1 is 0.6, and w2 is 0.4.
[0081] The comprehensive difficulty representation coefficient is:
[0082] Dz=w1×Cz+w2×Tz=0.6×0.7+0.4×0.95==1.01
[0083] Difficulty Adjustment Determination:
[0084] If the comprehensive difficulty representation coefficient of 1.01 exceeds the preset satisfactory range, the satisfactory range is assumed to be 0.95-1.0, it will be determined that the difficulty needs to be adjusted.
[0085] Model conversion factor adjustment:
[0086] The system will determine whether the model conversion coefficient R needs to be adjusted based on the difference and the preset difference threshold, and make corresponding adjustments.
[0087] By collecting the target's learning data in real time, this method can provide each target with tailored learning materials to ensure that the teaching content matches the target's personal ability. By using the time-difficulty adjustment factor and the model conversion coefficient, the difficulty of the exercises can be dynamically adjusted to adapt to the target's learning progress and comprehension ability. By accurately matching the target's learning level, the target can reduce the time wasted on questions of inappropriate difficulty, thereby improving learning efficiency. According to the target's practice feedback, not only can the difficulty of the exercises be adjusted, but the model conversion coefficient can also be optimized, so that the teaching method can be continuously improved to better adapt to the target's learning needs. Through the comprehensive difficulty representation coefficient, the target's learning performance can be monitored and feedback and support can be provided in a timely manner. The use of the preset question bank and model conversion coefficient makes teaching resources more efficiently utilized and reduces resource waste.
[0088] The time difficulty adjustment factor is a value used in the adaptive teaching system to reflect the relationship between the time the target spends learning a specific content and the preset standard time. This factor is calculated dynamically and used to adjust the difficulty of the generated exercises to adapt to the target's learning speed and comprehension ability.
[0089] The standard duration is used as a benchmark for evaluating and adjusting the difficulty of learning. In teaching methods, for example, the design team of a mathematics course decides that the learning content of each chapter should be completed within 40 to 45 minutes based on the complexity of the course content. This embodiment selects 45 minutes, which is the best learning time that meets the goal.
[0090] The target difficulty level, in the adaptive teaching system, is a quantitative indicator used to indicate the difficulty level of exercises or learning materials. This value is usually calculated based on the target's learning performance and the standards set by the system. The purpose is to generate exercises that match the target's current knowledge level and ability. It reflects the target's current learning status and ability, and is a preliminary assessment of the target's learning difficulty.
[0091] Input difficulty level. In the adaptive teaching system, the difficulty index is obtained after the target difficulty level is converted by the generative big model, which is used to guide the generative big model to generate exercises. This conversion process takes into account multiple factors to ensure that the difficulty of the generated exercises matches the target learning progress and ability.
[0092] Generative big models are algorithms that can generate exercises based on model conversion coefficients. They can be constructed using recurrent neural networks (RNNs), long short-term memory networks (LSTMs), transformers, etc.
[0093] The satisfaction range, in the adaptive teaching system, is used to define the best performance range that the target achieves when completing exercises. The satisfaction range is usually set based on the target's learning goals, the requirements of the syllabus, and the target's cognitive development level. In this embodiment, it is set based on the comprehensive difficulty representation coefficient calculated based on the accuracy rate and completion time, which directly affects the personalized adjustment of the teaching content and the learning effect of the target. In this embodiment, it is set to 0.95-1.0. Setting a satisfaction range can help the system determine whether the target is in the best learning state and adjust the teaching content accordingly.
[0094] A preset question bank is a pre-prepared, structured collection of educational content that contains exercises or test questions of various types and difficulty levels. The purpose of the question bank is to provide sufficient resources for the teaching system to dynamically generate or select appropriate exercises based on the target's learning progress, ability, and performance.
[0095] Specifically, the target difficulty level of the exercises suitable for the target is calculated based on the knowledge level, learning speed, correctness of the exercises, learning time, the preset standard time, and the calculated time difficulty adjustment factor.
[0096]
[0097] Among them, D is the target difficulty level, K is the knowledge level, L is the learning speed, C is the correct rate of exercises, T is the learning time, and T' is the standard time. It is the duration difficulty adjustment factor.
[0098] The target's learning data is collected, including knowledge level, learning speed, accuracy of exercises and learning time. Then, based on these data, the preset standard time and the calculated time difficulty adjustment factor, a specific algorithm is used to combine these parameters to calculate the learning difficulty value suitable for the target, thereby generating or selecting exercises of appropriate difficulty.
[0099] By utilizing the learning data collected in real time, such as knowledge level, learning speed, correctness of exercises and learning time, combined with the preset standard time and time difficulty adjustment factor, the learning difficulty value suitable for the target is dynamically calculated. It can provide a personalized learning path for each target, dynamically adjust the learning difficulty to match the target's learning progress and ability, thereby improving learning efficiency and results.
[0100] Specifically, the duration difficulty adjustment factor is calculated based on the learning duration and the preset standard duration.
[0101]
[0102] Among them, α is a constant that controls the degree of influence, and β is the upper limit of the duration difficulty adjustment factor.
[0103] The time difficulty adjustment factor is calculated using the difference between the learning time and the preset standard time, where α is a constant that controls the degree of impact of the time, and β sets the maximum value of the time difficulty adjustment factor to ensure that the adjustment factor is within a reasonable range and avoid excessive fluctuations in difficulty that affect the target learning effect.
[0104] By calculating the duration difficulty adjustment factor, the adaptive teaching system can adjust the teaching difficulty more finely to adapt to the target's current learning status and progress. This method helps maintain the target's learning motivation and ensures that they will not be frustrated by questions that are too difficult, nor lose interest because the questions are too easy.
[0105] Specifically, the target difficulty level is multiplied by the preset model conversion coefficient to calculate the input difficulty level for the model to generate the corresponding target difficulty level.
[0106] D'=D×R,
[0107] Where D' is the input difficulty level and R is the model conversion coefficient.
[0108] The target difficulty level D of the target is multiplied by a preset model conversion coefficient R to generate an input difficulty level D' of the exercise suitable for the target exercise. This process allows the system to convert the original difficulty value into a value suitable for use by the generative large model.
[0109] This conversion mechanism provides a flexible way to adjust the difficulty of the exercises to match the target's ability and learning needs. By precisely controlling the conversion of difficulty values, a series of exercises can be generated that are neither too difficult nor too easy for the target, thereby promoting the target's learning progress and cognitive development.
[0110] Specifically, the completion time indicator is calculated based on the ratio of the calculated completion time to the preset minimum completion time.
[0111]
[0112] Where Tz is the completion time indicator, Tw is the completion time, and Twmin is the minimum completion time.
[0113] The actual time Tw taken by the target to complete the exercise is compared with the minimum completion time Twmin preset by the system to calculate a completion time index Tz, which reflects the efficiency and suitability of the time taken by the target to complete a specific exercise.
[0114] By calculating the completion time indicator, the teaching system can evaluate the target's learning speed and understanding. If the completion time indicator shows that the target takes too long, it means that the question is too difficult or the target has difficulty understanding certain concepts; if the time is too short, it means that the question is too difficult or the target has mastered the knowledge points well. Such feedback can help the adaptive teaching system adjust the difficulty of subsequent exercises to ensure that the target learns at an appropriate challenge level.
[0115] Specifically, the accuracy index is calculated by calculating the ratio of the accuracy of the exercise questions to the preset maximum accuracy.
[0116]
[0117] Among them, Cz is the accuracy index and Cmax is the maximum accuracy.
[0118] By comparing the actual accuracy of the target with the maximum accuracy Cmax preset by the system, an accuracy index Cz is calculated.
[0119] The accuracy index Cz provides a key parameter for the adaptive teaching system to judge the target's mastery of the learning content. If the accuracy index is high, it means that the target has a good understanding and mastery of the relevant concepts, and the system can appropriately increase the difficulty of the questions to provide further challenges. Conversely, if the accuracy index is lower than expected, the system can adjust the teaching strategy, such as reviewing relevant knowledge points or providing additional exercises, to help the target strengthen understanding and learning.
[0120] Specifically, the comprehensive difficulty representation coefficient is calculated based on the accuracy rate and completion time indicators of the exercises.
[0121] Dz=w1×Cz+w2×Tz,
[0122] Among them, Dz is the comprehensive difficulty representation coefficient, w1 is the accuracy weight coefficient, and w2 is the completion time weight coefficient.
[0123] The target's exercise accuracy and completion time indicators are combined, and a comprehensive difficulty representation coefficient is calculated by assigning different weight coefficients to the accuracy w1 and completion time w2. This coefficient comprehensively reflects the target's performance in the two dimensions of accuracy and completion time.
[0124] The comprehensive difficulty representation coefficient provides a multi-dimensional evaluation tool that enables the teaching system to more accurately grasp the target's learning status. By considering the two key factors of accuracy and completion time, the system can adjust the teaching difficulty more carefully to ensure that the exercises can effectively test the target's learning outcomes while maintaining the challenge of learning.
[0125] The accuracy weight coefficient reflects the importance of the accuracy of the exercise questions in evaluating the target learning performance. A higher weight means that the system attaches more importance to the accuracy of the target answers.
[0126] The completion time weight coefficient indicates the proportion of completion time in the evaluation. A higher weight means that the system pays more attention to the speed of answering the target questions.
[0127] In this embodiment, the accuracy weight coefficient is set to 0.6, and the completion time weight coefficient is set to 0.4. By emphasizing the accuracy rate, the target is encouraged to deeply understand the learning materials rather than just completing the exercises quickly, preventing the target from neglecting careful thinking and accurate answers to questions in pursuit of speed.
[0128] Please continue reading Figure 2 As shown, it is a decision logic diagram for determining whether to compare the learning time in this embodiment;
[0129] Specifically, by comparing the comprehensive difficulty representation coefficient with the preset satisfactory range,
[0130] If the comprehensive difficulty representation coefficient is not within a satisfactory range, it is determined that it is necessary to compare the learning time with the standard time.
[0131] Please continue reading Figure 3 As shown, it is a decision logic diagram for determining whether to adjust the difficulty in this embodiment;
[0132] Compare the study time with the standard time.
[0133] If the learning time is less than the standard learning time, it is determined that there is no need to adjust the difficulty;
[0134] If the learning time is greater than or equal to the standard learning time, it is determined that the difficulty needs to be adjusted.
[0135] The actual comprehensive difficulty representation coefficient of the target is compared with the satisfactory range preset by the system. If the actual performance of the target is not within the satisfactory range, the system will further analyze the difference between the learning time and the preset standard time to determine whether the difficulty of the exercises needs to be adjusted. The actual learning time used by the target to complete the exercises is compared with the preset standard time. If the target time is less than the standard time, the system will consider the current difficulty appropriate and no adjustment is required; if the time reaches or exceeds the standard time, the system will determine that the difficulty may need to be adjusted.
[0136] If the target finds the exercises too difficult or too easy, the system can optimize the learning experience by adjusting the difficulty to ensure that the target remains in the optimal learning zone. This helps maintain the target's learning motivation, improve learning efficiency, and promote cognitive development. By considering the length of study, the system can evaluate the target's adaptability to the difficulty of exercises. Shorter study time may indicate that the target has a good grasp of the material, while longer study time means that the target is having difficulty. This approach helps the system make more accurate difficulty adjustment decisions, thereby improving the target's learning efficiency and quality.
[0137] Please continue reading Figure 4 As shown, it is a decision logic diagram for determining whether to adjust the model conversion coefficient in this embodiment;
[0138] Specifically, when it is determined that the difficulty needs to be adjusted, the difference between the standard comprehensive difficulty characterization coefficient and the comprehensive difficulty characterization coefficient is calculated, and based on the difference and the preset difference threshold, it is determined whether the model conversion coefficient needs to be adjusted.
[0139] If the difference is greater than the preset difference threshold, it is determined that the model conversion coefficient needs to be adjusted, the adjustment ratio is calculated according to the minimum value of the satisfactory range and the comprehensive difficulty characterization coefficient, and the model conversion coefficient is adjusted according to the adjustment ratio.
[0140]
[0141] Among them, P is the adjustment ratio, Mmin is the minimum value of the satisfactory range, and R' is the adjusted model conversion coefficient.
[0142] After determining that the difficulty needs to be adjusted, the difference between the standard comprehensive difficulty characterization coefficient and the actual comprehensive difficulty characterization coefficient is calculated and compared with the preset difference threshold. If this difference exceeds the threshold, the system will determine that the model conversion coefficient needs to be adjusted. Subsequently, the system will calculate an adjustment ratio P based on the minimum value Mmin of the satisfaction range and the current comprehensive difficulty characterization coefficient, and adjust the model conversion coefficient R' accordingly to ensure that the difficulty of the generated exercises is closer to the target optimal learning range.
[0143] This adjustment mechanism enables the adaptive teaching system to respond more accurately to the target's learning performance and adjust the teaching difficulty in a timely manner to meet the target's personalized learning needs. By optimizing the model conversion coefficient, the system can generate exercises that are more in line with the target's current learning level, avoiding problems of too high or too low difficulty, thereby improving the target's learning efficiency and learning outcomes.
[0144] The difference threshold is a parameter used in the adaptive teaching system to determine whether the model conversion coefficient needs to be adjusted. It represents the maximum acceptable difference between the actual performance of the target and the expected performance. The setting of the difference threshold usually depends on the learning objectives of the target and the requirements of the syllabus. In this embodiment, a difference threshold is set, which is assumed to be 0.05.
[0145] This setting ensures that the teaching content matches the target's actual ability, provides a personalized learning experience, and avoids frequent difficulty adjustments, allowing the target to focus more on learning rather than constantly adapting to new difficulties.
[0146] Please continue reading Figure 5 As shown, it is a schematic diagram of the structure of the adaptive teaching system of the generative large model of this embodiment;
[0147] An adaptive teaching system based on a generative large model, based on the above-mentioned adaptive teaching method based on a generative large model, comprises:
[0148] The data collection module is used to collect the knowledge level, learning speed, correctness of exercises and learning time of several targets in real time;
[0149] A large model generation module, connected to the data acquisition module, is used to calculate a time difficulty adjustment factor according to the learning time and the preset standard time, and calculate a target difficulty level of exercises suitable for the target to practice according to the knowledge level, learning speed, correct rate of exercises, learning time, the preset standard time and the calculated time difficulty adjustment factor, and calculate an input difficulty level for model generation corresponding to the target difficulty level by multiplying the target difficulty level and the preset model conversion coefficient, and generate a number of corresponding large model exercises by inputting the input difficulty level into the generative large model;
[0150] The data collection module is also used to collect in real time the correctness and completion time of the large model exercises corresponding to the several targets;
[0151] The determination module is connected to the data acquisition module and the large model generation module respectively, and is used to calculate the completion time index and the accuracy index according to the completion time and the accuracy of the exercises respectively, and calculate the comprehensive difficulty characterization coefficient according to the accuracy index and the completion time index, and determine whether the difficulty needs to be adjusted according to the comprehensive difficulty characterization coefficient, the preset satisfaction range, the learning time and the standard time.
[0152] The adjustment module is connected to the determination module and is used to select existing standard exercises of the same difficulty from a preset question bank when it is determined that the difficulty needs to be adjusted, and determine whether the model conversion coefficient needs to be adjusted based on the accuracy and completion time of the exercises after completing training for several goals, as well as the calculated standard comprehensive difficulty characterization coefficient and the comprehensive difficulty characterization coefficient, and adjust the model conversion coefficient based on the satisfaction range and the adjustment ratio calculated by the comprehensive difficulty characterization coefficient when it is determined that the model conversion coefficient needs to be adjusted.
[0153] A personalized learning experience is achieved through a series of carefully designed modules working together. First, the data acquisition module monitors and collects various learning indicators of learners in real time, including knowledge level, learning speed, accuracy of exercises and learning time. Then, the large model generation module calculates the target difficulty level of exercises suitable for learners based on these data and the preset standard time, and generates corresponding exercises accordingly. The system also tracks the accuracy and completion time of learners in completing these exercises in real time. These data are used by the judgment module to calculate the comprehensive difficulty representation coefficient to evaluate whether the current difficulty is within the preset satisfactory range. If the judgment result indicates that the difficulty needs to be adjusted, the adjustment module will intervene, select exercises of appropriate difficulty and adjust the model conversion coefficient to ensure that the difficulty of the exercises matches the actual performance of the learners.
[0154] The system adapts teaching content in real time to meet the needs of different learners. The system dynamically generates exercises to ensure that learners always learn at an appropriate level of challenge, thereby improving learning efficiency and effectiveness. In addition, the system can reduce the workload of teachers, allowing them to focus more on teaching design and other creative tasks. As learners progress, the system will continue to optimize the learning path to help learners master knowledge at a pace that suits them. Ultimately, this system helps to improve the overall quality of education, achieve optimal allocation of educational resources, and bring learners a richer and deeper learning experience.
[0155] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.
[0156] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. An adaptive teaching method based on a generative large model, characterized in that: include: Step S1, real-time collection of the knowledge level, learning speed, correct rate of exercises and learning time of several targets; Step S2, calculating a duration difficulty adjustment factor according to the learning duration and the preset standard duration, and calculating a target difficulty level of exercises suitable for the target to practice according to the knowledge level, learning speed, correct rate of exercises, learning duration, the preset standard duration and the calculated duration difficulty adjustment factor, and calculating an input difficulty level for model generation corresponding to the target difficulty level according to the target difficulty level and the preset model conversion coefficient, and generating a number of corresponding large model exercises by inputting the input difficulty level into the generative large model; Step S3, collecting the correctness and completion time of the large model exercises corresponding to the completion of several goals in real time, and calculating the completion time index and the correctness index according to the completion time and the correctness of the exercises, and calculating the comprehensive difficulty representation coefficient according to the correctness index and the completion time index; Step S4, based on the comprehensive difficulty characterization coefficient, the preset satisfaction range, the learning time and the standard time, determine whether the difficulty needs to be adjusted, and when it is determined that the difficulty needs to be adjusted, select existing standard exercises of the same difficulty from the preset question bank, and based on the accuracy and completion time of the exercises after completing training for several goals, as well as the calculated standard comprehensive difficulty characterization coefficient and the comprehensive difficulty characterization coefficient, determine whether the model conversion coefficient needs to be adjusted, and adjust the model conversion coefficient based on the adjustment ratio calculated from the satisfaction range and the comprehensive difficulty characterization coefficient.
2. The adaptive teaching method based on generative large model according to claim 1 is characterized in that: Calculate the target difficulty level of exercises suitable for the target according to the knowledge level, learning speed, accuracy of exercises, learning time, preset standard time and the calculated time difficulty adjustment factor. Among them, D is the target difficulty level, K is the knowledge level, L is the learning speed, C is the correct rate of exercises, T is the learning time, and T' is the standard time. It is the duration difficulty adjustment factor.
3. The adaptive teaching method based on generative large model according to claim 2 is characterized in that: Calculate the duration difficulty adjustment factor based on the learning duration and the preset standard duration. Among them, α is a constant that controls the degree of influence, and β is the upper limit of the duration difficulty adjustment factor.
4. The adaptive teaching method based on generative large model according to claim 3 is characterized in that: The target difficulty level is multiplied by a preset model conversion coefficient to calculate the input difficulty level used to generate the model corresponding to the target difficulty level.
5. The adaptive teaching method based on generative large model according to claim 4 is characterized in that: The completion time indicator is calculated based on the ratio of the calculated completion time and the preset minimum completion time.
6. The adaptive teaching method based on generative large model according to claim 5 is characterized in that: The accuracy index is calculated by calculating the ratio of the accuracy of the exercise questions to the preset maximum accuracy.
7. The adaptive teaching method based on generative large model according to claim 6 is characterized in that: The comprehensive difficulty representation coefficient is calculated based on the accuracy rate and completion time indicators of the exercises. Dz=w1×Cz+w2×Tz, Among them, Dz is the comprehensive difficulty representation coefficient, w1 is the accuracy weight coefficient, and w2 is the completion time weight coefficient.
8. The adaptive teaching method based on generative large model according to claim 7 is characterized in that: Compare the comprehensive difficulty characterization coefficient with the preset satisfactory range, If the comprehensive difficulty representation coefficient is not within a satisfactory range, it is determined that the learning time needs to be compared with the standard time. If the learning time is less than the standard time, it is determined that there is no need to adjust the difficulty; If the learning time is greater than or equal to the standard time, it is determined that the difficulty needs to be adjusted.
9. The adaptive teaching method based on generative large model according to claim 8 is characterized in that: When it is determined that the difficulty needs to be adjusted, the difference between the standard comprehensive difficulty characterization coefficient and the comprehensive difficulty characterization coefficient is calculated, and based on the difference and the preset difference threshold, it is determined whether the model conversion coefficient needs to be adjusted. If the difference is greater than the preset difference threshold, it is determined that the model conversion coefficient needs to be adjusted, the adjustment ratio is calculated according to the minimum value of the satisfactory range and the comprehensive difficulty characterization coefficient, and the model conversion coefficient is adjusted according to the adjustment ratio. Among them, P is the adjustment ratio, Mmin is the minimum value of the satisfactory range, and R' is the adjusted model conversion coefficient.
10. An adaptive teaching system based on a generative large model, based on the adaptive teaching method based on a generative large model according to any one of claims 1 to 9, comprising: The data collection module is used to collect the knowledge level, learning speed, correctness of exercises and learning time of several targets in real time; A large model generation module, connected to the data acquisition module, is used to calculate a time difficulty adjustment factor according to the learning time and the preset standard time, and calculate a target difficulty level of exercises suitable for the target to practice according to the knowledge level, learning speed, the correct rate of exercises, the learning time, the preset standard time and the calculated time difficulty adjustment factor, and calculate an input difficulty level for model generation corresponding to the target difficulty level according to the target difficulty level and the preset model conversion coefficient, and generate a number of corresponding large model exercises by inputting the input difficulty level into the generative large model; The data collection module is also used to collect in real time the correctness and completion time of the large model exercises corresponding to the several targets; The determination module is connected to the data acquisition module and the large model generation module respectively, and is used to calculate the completion time index and the accuracy index according to the completion time and the accuracy of the exercises respectively, and calculate the comprehensive difficulty characterization coefficient according to the accuracy index and the completion time index, and determine whether the difficulty needs to be adjusted according to the comprehensive difficulty characterization coefficient, the preset satisfaction range, the learning time and the standard time. The adjustment module is connected to the determination module and is used to select existing standard exercises of the same difficulty from a preset question bank when it is determined that the difficulty needs to be adjusted, and determine whether the model conversion coefficient needs to be adjusted based on the accuracy and completion time of the exercises after completing training for several goals, as well as the calculated standard comprehensive difficulty characterization coefficient and the comprehensive difficulty characterization coefficient, and adjust the model conversion coefficient based on the satisfaction range and the adjustment ratio calculated by the comprehensive difficulty characterization coefficient when it is determined that the model conversion coefficient needs to be adjusted.
Citation Information
Patent Citations
Intelligent homework assignment method and device
CN106409034A
Title push method, storage medium, and application system
CN109360457A
An Online Education Platform
CN109461103A
Exercise question recommendation method and device, equipment and storage medium
CN110413728A
Personalized adaptive learning recommendation method based on big data analysis of education platform
CN111368182A
Cited By
Multi-scene switching driving simulator control method and device and storage medium
CN119649670A
Multi-scene switching driving simulator control method, device and storage medium
CN119649670B
Learning scheme adjustment method and device, equipment and storage medium
CN120409602A
