Personalized teaching skill training system, device, medium and product based on large language model

Abstract

The application discloses a personalized teaching skill training system and device based on a large language model, a medium and a product, relates to the technical field of teaching skill training, and provides a solution meeting teaching requirements for a teaching problem proposed by a teacher through a teaching skill learning module in the system, so as to help the teacher learn teaching skills; secondly, a plurality of virtual student roles and a real scene teaching scene are generated through a real scene teaching simulation module, and the teacher is provided with online video mode for simulation, so as to exercise and improve actual teaching ability; thirdly, teaching behavior performance data and teacher-student interaction data are analyzed through a teaching feedback analysis module, a report is generated, and suggestions are proposed for deficiencies; finally, suitable teaching skill training resources are recommended through a teaching resource recommendation module, so as to help the teacher deeply learn and improve teaching methods. The scheme can effectively improve the teaching ability of the teacher, especially the response ability when facing diversified student behaviors and demands.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of teaching skill training, and particularly relates to a personalized teaching skill training system, device, medium and product based on a large language model. BACKGROUND

[0002] With the increasing diversification of the educational environment, the challenges faced by teachers are also increasing, especially novice teachers often have difficulty in quickly adapting to the variability of student behavior and teaching needs in the actual teaching process. Current teacher training methods are mainly divided into offline and online two kinds. Most of the online training modes are non-interactive, such as watching videos and theoretical explanations, but these methods cannot fully simulate real teaching scenarios, resulting in insufficient experience of teachers in actual teaching. Compared with online training mode, offline training mode can provide real classroom scenarios for teachers, but on the one hand, the cost of offline training is high, and it is difficult to apply and popularize at any time; on the other hand, although offline training facilitates the sharing of teaching experience between novice teachers and experienced teachers, it is often difficult to combine into actual teaching scenarios to exercise the teaching ability of teachers, and it is also difficult to provide personalized feedback and support, making it difficult for teachers to adjust according to the diversified needs of students.

[0003] Therefore, in the current situation that the teacher training system generally lacks real-time feedback and real teaching scenario simulation, it is particularly necessary to effectively improve the response ability of teachers, especially novice teachers, when facing diversified student behaviors and needs. SUMMARY

[0004] The purpose of the present application is to provide a personalized teaching skill training system, device, medium and product based on a large language model, which can effectively improve the teaching ability of teachers, especially the response ability when facing diversified student behaviors and needs.

[0005] To achieve the above purpose, the present application provides the following solutions:

[0006] In a first aspect, the present application provides a personalized teaching skill training system based on a large language model, comprising: a teaching skill learning module, a real scene teaching simulation module, a teaching feedback analysis module and a teaching resource recommendation module.

[0007] The teaching skill learning module is configured to answer the teaching questions raised by teachers based on a large language model, provide teaching theories, teaching strategies and / or teaching cases that meet the teaching needs of teachers, and help teachers learn teaching skills.

[0008] The real scene teaching simulation module is used to generate a plurality of virtual student roles and real scene teaching scenes based on a large language model, to provide a teacher with teaching simulation in the form of online video, and the virtual student roles simulate the behavior patterns of real students to generate feedback to the teacher during the teaching simulation process, so as to exercise and improve the actual teaching ability of the teacher.

[0009] The teaching feedback analysis module is used to obtain teaching behavior performance data and teacher-student interaction data of the teacher during the teaching simulation process, analyze the teaching behavior performance data and the teacher-student interaction data based on a large language model, generate a teaching analysis report, and generate improvement suggestions for deficiencies in the teaching simulation process.

[0010] The teaching resource recommendation module is used to recommend teaching skill training resources to the teacher according to the teaching analysis report based on a large language model, to help the teacher to further learn and improve the teaching method.

[0011] Optionally, a teaching question answering large model is used in the teaching skill learning module to answer the teaching questions raised by the teacher, and provide teaching theories, teaching strategies and / or teaching cases that meet the teaching needs of the teacher; the teaching question answering large model is constructed based on a large language model, and is obtained after the teaching question answering large model is trained using a teaching skill knowledge data set; the trained teaching question answering large model can comprehensively understand the teaching needs of the teacher and generate teaching theories, teaching strategies and / or teaching cases that meet the teaching needs of the teacher, to help the teacher to learn teaching skills.

[0012] The teaching question answering large model includes an input layer, a position encoding layer, an encoding layer, a multi-head self-attention mechanism, a feed-forward neural network layer, a decoding layer and an output layer; the input layer is used to accept the teaching questions input by the teacher and convert them into word embedding form that can be processed by the model; the position encoding layer is used to provide position information for each word in the teaching question, to ensure that the model can understand the order and context of words when processing text; the encoding layer is used to analyze and process the context of the teaching question through self-attention mechanism, to understand the teaching needs of the teacher; the multi-head self-attention mechanism is used to simultaneously focus on multiple parts of the teaching question, to capture the dependency relationship between different teaching concepts, to ensure that the generated answer covers multiple knowledge points; the feed-forward neural network layer is used to further process the context representation generated by the multi-head self-attention mechanism, to enhance the processing ability of the model for complex teaching questions; the decoding layer is used to generate teaching theories, teaching strategies and / or teaching cases that meet the teaching needs of the teacher according to the encoding results of the model; and the output layer is used to output a probability distribution through a softmax activation function, to output the answer of the model to the teaching question.

[0013] Optionally, the teaching question answering large model is trained using the teaching skill knowledge dataset, specifically including: taking the teaching skill knowledge dataset as the input of the teaching question answering large model, training the model by giving the input data and the expected output, adjusting the parameters of the teaching question answering large model, and making the teaching question answering large model more suitable for content generation in the education field; the teaching skill knowledge dataset includes teaching theories, teaching strategies and teaching cases.

[0014] Optionally, in the real scene teaching simulation module, a teaching scene generation large model is used to generate a plurality of virtual student roles and real scene teaching scenes in the form of online videos for teachers to perform teaching simulation; the teaching scene generation large model is constructed based on a large language model and obtained after the teaching scene generation large model is trained using a student behavior pattern dataset, and the trained teaching scene generation large model can generate a plurality of virtual student roles with different behavior patterns and real scene teaching scenes for teachers to perform teaching simulation. In the teaching simulation process, the virtual student role will simulate the behavior pattern of a real student to provide feedback to the teacher, thereby improving the actual teaching ability of the teacher.

[0015] Optionally, in the teaching feedback analysis module, a teaching feedback analysis large model is used to analyze the teaching behavior performance data and the teacher-student interaction data, generate a teaching analysis report, and generate improvement suggestions for the shortcomings in the teaching simulation process; the teaching feedback analysis large model is constructed based on a large language model and obtained after the teaching feedback analysis large model is trained using a teaching process evaluation dataset, and the trained teaching feedback analysis large model can use a teaching behavior analysis unit, an interaction data analysis unit, a teaching language sentiment analysis unit, and a teaching feedback strategy analysis unit to perform multi-dimensional analysis on the teaching behavior performance data and the teacher-student interaction data, accurately generate a teaching analysis report, and generate improvement suggestions for the shortcomings in the teaching simulation process.

[0016] Optionally, the teaching behavior analysis unit is used to automatically detect and classify the classroom language of the teacher through natural language processing technology, and analyze the type and applicability of the teaching behavior; the interaction data analysis unit is used to evaluate the interaction frequency, depth and quality of the teacher and the student reflected in the teacher-student interaction data by using social network analysis and multi-dimensional interaction analysis method; the teaching language sentiment analysis unit is used to evaluate the language sentiment expression of the teacher in the classroom based on a sentiment analysis algorithm; and the teaching feedback strategy analysis unit is used to evaluate the effectiveness of the feedback strategy by analyzing the feedback strategy of the teacher to the student.

[0017] Optionally, the teaching resource recommendation module adopts a teaching resource recommendation large model to recommend teaching skill training resources to the teacher according to the teaching analysis report; the teaching resource recommendation large model is obtained after a large language model is constructed and the teaching resource recommendation large model is trained using a teaching analysis report data set, so that the teaching resource recommendation large model can fully understand and analyze the language content in the teaching analysis report, analyze the specific aspects that the teacher needs to improve, accurately recommend teaching skill training resources to the teacher, and help the teacher to deeply learn and improve the teaching method.

[0018] In a second aspect, the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to realize the functions of the personalized teaching skill training system based on the large language model as described above.

[0019] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to realize the functions of the personalized teaching skill training system based on the large language model as described above.

[0020] In a fourth aspect, the present application provides a computer program product comprising a computer program, and the computer program is executed by a processor to realize the functions of the personalized teaching skill training system based on the large language model as described above.

[0021] According to the specific embodiments provided by the present application, the present application discloses the following technical effects:

[0022] The application provides a large language model-based personalized teaching skill training system, device, medium and product. In the system, first, a teaching skill learning module is used to answer the teaching problems raised by teachers, and teaching theories, teaching strategies and / or teaching cases meeting the teaching needs of the teachers are provided to help the teachers learn teaching skills. Second, a real scene teaching simulation module is used to generate a plurality of virtual student roles and real scene teaching scenes, and the teachers are provided with online videos for teaching simulation. In the teaching simulation process, the virtual student roles simulate the behavior patterns of real students to provide feedback to the teachers, so as to exercise and improve the actual teaching ability of the teachers. Third, a teaching feedback analysis module is used to analyze teaching behavior performance data and teacher-student interaction data, generate a teaching analysis report and generate improvement suggestions for the deficiencies in the teaching simulation process. Finally, a teaching resource recommendation module is used to recommend teaching skill training resources to the teachers according to the teaching analysis report, so as to help the teachers deeply learn and improve the teaching methods. The above scheme provided by the application can answer the teaching problems raised by the teachers, improve the basic teaching ability of the teachers, simulate the real scene teaching scene to exercise the actual teaching ability of the teachers, analyze and quantify the data in the teaching simulation process, obtain an analysis report and provide improvement suggestions for the deficiencies, help the teachers deeply learn and improve the deficiencies of the teaching methods, and improve the response ability of the teachers, especially the novice teachers, when facing diversified student behaviors and needs. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor based on these drawings.

[0024] Figure 1 A functional module schematic diagram of a large language model-based personalized teaching skill training system according to an embodiment of the present application.

[0025] Figure 2 A model structure schematic diagram of a teaching problem answering large model in a large language model-based personalized teaching skill training system according to an embodiment of the present application.

[0026] Figure 3 A structural schematic diagram of a computer device according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] With reference to the drawings and specific embodiments described below, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0028] The above purposes, features and advantages of the present application will be more apparent and understandable. The present application will be further described in detail below with reference to the drawings and specific embodiments.

[0029] In one exemplary embodiment, as shown in Figure 1 A large language model-based personalized teaching skill training system is provided, including a teaching skill learning module M1, a real scene teaching simulation module M2, a teaching feedback analysis module M3, and a teaching resource recommendation module M4. The functions implemented by each functional module will be described below.

[0030] The teaching skill learning module M1 is used to answer the teaching questions raised by teachers based on a large language model, and provide teaching theories, teaching strategies and / or teaching cases that meet the teaching needs of teachers to help teachers learn teaching skills. The teaching skill learning module M1 is used to provide professional knowledge about teaching theories, strategies and cases for teachers. In this embodiment, a teaching question answering large model is used in the teaching skill learning module M1 to answer the teaching questions raised by teachers, and provide teaching theories, teaching strategies and / or teaching cases that meet the teaching needs of teachers; the teaching question answering large model is constructed based on a large language model and obtained after training the teaching question answering large model using a teaching skill knowledge dataset, and the trained teaching question answering large model can comprehensively understand the teaching needs of teachers and generate teaching theories, teaching strategies and / or teaching cases that meet the teaching needs of teachers. For example, the teaching question answering large model can be obtained by adjusting an open source large language model (such as Llama).

[0031] In this embodiment, as shown in Figure 2As shown, the teaching question answering large model used in the teaching skill learning module M1 includes an input layer, a position encoding layer, an encoding layer, a multi-head self-attention mechanism, a feed-forward neural network layer, a decoding layer, and an output layer in structure; the input layer is used to accept the teaching question input by the teacher and convert it into a word embedding form that the model can process; the position encoding layer is used to provide position information for each word in the teaching question, ensuring that the model can understand the order and context of words when processing text; the encoding layer is used to analyze and process the context of the teaching question through self-attention mechanism, to understand the teaching needs of the teacher; the multi-head self-attention mechanism is used to focus on multiple parts of the teaching question at the same time, capture the dependency between different teaching concepts, and ensure that the generated answer covers multiple knowledge points; the feed-forward neural network layer is used to further process the context representation generated by the multi-head self-attention mechanism, enhancing the model's ability to handle complex teaching questions; the decoding layer is used to generate teaching theories, teaching strategies, and / or teaching cases that meet the teacher's teaching needs based on the model's encoding results; the output layer is used to output a probability distribution through a softmax activation function, outputting the model's answer to the teaching question.

[0032] The teaching question answering large model is trained using the teaching skill knowledge dataset, specifically including: taking the teaching skill knowledge dataset as the input of the teaching question answering large model, training the model by giving the input data and the expected output, adjusting the parameters of the teaching question answering large model, making the teaching question answering large model more suitable for content generation in the education field, for example, inputting "how to conduct differentiated teaching?", according to the output of the model and the expected output, gradually fine-tuning the model parameters to generate accurate suggestions on teaching strategies; In addition, the trained model needs to be verified and tested: test the model on the validation set to ensure that it can correctly understand and generate education-related content, and make necessary adjustments to the model. The teaching skill knowledge dataset includes teaching theories, teaching strategies, and teaching cases.

[0033] Specifically in this embodiment, teaching skills are just one of the multiple types of text data covered in the dataset, ensuring that the model can fully understand and generate recommendations about teaching. Specifically, (1) teaching theories, including classical education theories and learning theories, cover constructivism, behaviorism, cognitivism, etc., in the form of theoretical literature, research reviews, and teaching theory book summaries. (2) Teaching strategies cover various teaching methods (such as lecture method, discussion method, experimental method, etc.) and differentiated teaching strategies, in the form of teaching method guidelines, strategy explanations, and comparative analysis documents. (3) Teaching cases, successful and failed cases in actual classrooms, including strategy application and result analysis, in the form of case analysis reports, research papers, and classroom video text descriptions. (4) Course design and teaching plan, including course goal setting, course structure design, and teaching link planning, in the form of course outline, teaching plan template, and case study. (5) Teaching evaluation and feedback, collecting formative and summative evaluation strategies, combining student feedback for teaching adjustment, in the form of teacher evaluation reports and feedback data. (6) Subject professional knowledge, specific knowledge points and teaching methods of various subjects, especially for core courses (Chinese, mathematics, English, etc.), in the form of subject knowledge point explanations and teaching materials.

[0034] As an exemplary embodiment, when a teacher obtains teaching knowledge in the form of questions and answers through the teaching skill learning module M1, the execution process of the teaching skill learning module M1 is as follows:

[0035] Teaching question input: The teacher inputs a question related to teaching, such as "What are the classroom management skills for cooperative learning?"

[0036] Semantic analysis: The input layer of the model first receives the question and performs semantic analysis on the question through the multi-head self-attention mechanism in the encoding layer, understanding the core needs of the question.

[0037] Knowledge retrieval: In the knowledge base fine-tuned by the model, retrieve teaching theories, strategies, or cases related to the question.

[0038] Answer generation: The decoding layer generates specific answers based on the encoded information, providing the teacher with teaching recommendations that meet their needs.

[0039] Result output: The output layer converts the generated answer into natural language output, for example, "In cooperative learning, you can improve classroom order management by assigning clear roles and tasks."

[0040] In the above teaching skill learning module M1, the teaching question answering large model generates answers through the encoding and decoding layers, ultimately providing detailed recommendations on teaching theories, teaching strategies, or cases. This not only helps teachers quickly obtain the required professional knowledge, but also provides personalized and precise teaching support.

[0041] The live teaching simulation module M2 is used to generate a plurality of virtual student roles and live teaching scenes based on a large language model, to provide online video for teachers to conduct teaching simulation. The virtual student roles will simulate the behavior patterns of real students to produce feedback to the teachers during the teaching simulation process, to exercise and improve the actual teaching ability of the teachers. Specifically, the live teaching simulation module M2 allows teachers to access a highly simulated simulated classroom environment through online video. In the simulated classroom environment, the large language model not only plays different types of student roles, but also simulates various situations in a real classroom through dynamic interaction.

[0042] In the present embodiment, a teaching scene generation large model is used in the live teaching simulation module M2 to generate a plurality of virtual student roles and live teaching scenes in the form of online video for teachers to conduct teaching simulation. The teaching scene generation large model is constructed based on a large language model and obtained after the teaching scene generation large model is trained using a student behavior pattern dataset. The trained teaching scene generation large model can generate a plurality of virtual student roles with different behavior patterns and live teaching scenes for teachers to conduct teaching simulation. The virtual student roles will simulate the behavior patterns of real students to produce feedback to the teachers during the teaching simulation process, to exercise and improve the actual teaching ability of the teachers.

[0043] Specifically, the teaching scene generation large model can generate virtual student roles with different learning levels, learning styles, and personality characteristics through a pre-tuned dataset. For example, the model can create student roles with weak basic knowledge, proactive students, introverted and quiet students, etc. Each virtual student has different behavior patterns in the classroom, which can present a diverse student group.

[0044] For this function, the teaching scene generation large model inputs content mainly including multi-dimensional data related to student characteristics and teaching interaction during the training process, so that the model can generate diverse virtual student roles. The expected output is that the model can learn to generate text descriptions that meet the input characteristics, i.e., for each different type of virtual student, the model can give its behavior pattern and corresponding classroom interaction response.

[0045] Specifically, the input during model training can include student feature description data, classroom interaction behavior data, and scenario simulation and teaching cases.

[0046] The student characteristic description data includes the learning level, learning style, personality characteristics, etc. of different students. The input data will describe the specific performance of these characteristics in detail, for example: learning level (beginner, intermediate, advanced), learning style (visual learning, auditory learning, hands-on operation), personality (introverted, extroverted, positive initiative, shy, etc.). The classroom interaction behavior data contains a large number of descriptions of actual teacher-student interactions in the classroom, involving how students respond to teachers' questions, cooperate with classmates, and the specific problems encountered by different types of students in the learning process. These interaction behavior descriptions can help the model learn the behavior patterns of different students in the classroom. The scenario simulation and teaching case contain scene descriptions in the classroom and students' reactions in different teaching situations, such as performance in a certain classroom task, reaction to different teaching methods, etc. These data can help the model understand the diverse ways students react.

[0047] The expected output during model training includes: virtual student images that meet the characteristic description, virtual student classroom interaction behaviors, and student reactions in scenario simulation

[0048] The virtual student image that meets the characteristic description has the following specific requirements: the model should be able to generate detailed virtual student images that meet the characteristics based on the input student characteristics. For example, for a student with "weak basic knowledge, introverted and quiet", the model should be able to generate the typical reactions of this student in the classroom, such as "hesitating to answer when the teacher asks a question, preferring to complete tasks independently", etc. The specific requirements for virtual student classroom interaction behavior are the description of the behavior and reaction of students in different classroom situations. For example, in the face of a certain problem or task, an active student may answer the question or actively seek help, while an introverted student may take a long time to think before answering. The description of these behavior patterns is the expected output of the training, used to simulate different reactions of students. The specific requirements for student reactions in scenario simulation are that the model should be able to generate specific reactions of virtual students based on the input classroom scene in scenario simulation tasks. For example, in a team cooperation task, the model can generate the behavior dialogue of the student, reflecting the individual differences in cooperation, such as a student who is good at leading the team, while other students may show follow-up team actions.

[0049] In specific applications, the large language model within the system is built with a general template framework and some basic and common virtual student role designs, such as a mischievous boy in the first grade and a quiet girl in the third grade. Teachers only need the number of students and the corresponding template roles to quickly generate virtual students in the classroom. In addition to basic student roles, teachers can also customize student roles. Teachers can input simple descriptions, such as "a math-achieving, group-discussion-loving, but occasionally distracted fourth-grader," and the system will automatically expand the details and fill them into the general template framework to generate more detailed virtual student images. In addition, the template framework has a parameterized structure that allows the student's interaction mode, question type, participation frequency, and other parameters to be adjusted according to the description, making the generated role more in line with the teacher's specific needs and the actual classroom situation that may occur. The general template framework also supports the generation of different types of student groups. For example, teachers can set up a classroom with different types of student roles, such as active question-asking students, shy but smart students, and students with slightly weaker understanding abilities. The model will generate a classroom scene with rich interaction patterns based on different template combinations in the framework to simulate a variety of classroom situations.

[0050] In addition, the real scene teaching simulation module M2 also has a question mechanism and a feedback mechanism.

[0051] Question mechanism: Virtual students can actively ask questions based on course content, and the type of question will be adjusted according to the characteristics of the student role. For example, students with weak foundations may ask simple clarifying questions, while advanced students may ask challenging questions. This question mechanism can help teachers develop the ability to handle questions at different levels.

[0052] Feedback mechanism: During the teacher's teaching process, virtual students can give real-time feedback in the form of nodding, doubts, questions, or hand signals. Through this interaction, teachers can feel the students' sense of participation and adjust their teaching strategies based on feedback.

[0053] As an exemplary embodiment, the teaching scene generation large model should also simulate classroom situations: the large language model can generate complex classroom situations according to different teaching situations. For example, some students in the classroom may be distracted, interrupted, or have difficulty understanding. Teachers need to adjust their explanation methods and management strategies according to these situations to maintain classroom order and improve teaching efficiency.

[0054] The teaching scenario generation large model should also simulate the complexity of role interaction: as the teacher's performance in the simulated classroom improves, the student role generated by the large language model will dynamically adjust, and the complexity and difficulty of interaction will gradually increase. For example, student questions may be simple at the beginning, but as the teacher's teaching skills improve, student questions will become more challenging, and even interdisciplinary questions or complex thinking challenges may appear to test the teacher's adaptability.

[0055] The teaching scenario generation large model should also simulate customized scenarios: depending on different teaching goals, the practical simulation module also supports teachers choosing specific scenarios for training, such as large-class teaching, group discussion, or one-on-one tutoring. The virtual student interaction patterns and question feedback methods in each scenario will be different to meet the practical needs of teachers in various teaching environments.

[0056] The teaching feedback analysis module M3 is used to obtain the teaching behavior performance data and teacher-student interaction data in the teaching simulation process, analyze the teaching behavior performance data and teacher-student interaction data based on a large language model, generate a teaching analysis report, and generate improvement suggestions for deficiencies in the teaching simulation process.

[0057] In this embodiment, the teaching feedback analysis large model is used in the teaching feedback analysis module M3 to analyze the teaching behavior performance data and teacher-student interaction data, generate a teaching analysis report, and generate improvement suggestions for deficiencies in the teaching simulation process; the teaching feedback analysis large model is constructed based on a large language model and obtained after training the teaching feedback analysis large model using a teaching process evaluation data set. The trained teaching feedback analysis large model can use the teaching behavior analysis unit, the interaction data analysis unit, the teaching language sentiment analysis unit, and the teaching feedback strategy analysis unit to perform multi-dimensional analysis on the teaching behavior performance data and teacher-student interaction data, accurately generate a teaching analysis report, and generate improvement suggestions for deficiencies in the teaching simulation process.

[0058] Specifically, the teaching behavior analysis unit is used to automatically detect and classify the teacher's classroom language through natural language processing technology, analyze the types of teaching behavior and their applicability; in this teaching behavior analysis unit, a data set containing actual classroom teaching interaction data can be used to fine-tune the large language model so that it can recognize and classify teaching behaviors (such as explanation, questioning, feedback, management, etc.). For example, data using the Flanders Interaction Analysis Method can be used for labeling and training, and the model needs to learn the characteristics and classification standards of different teaching behaviors during the training process.

[0059] Several common teaching behaviors are explained and described:

[0060] Explanation, refers to the behavior of teachers imparting knowledge to students in the classroom, usually manifested as teachers explaining concepts, theories or knowledge points to students unilaterally. It is a major part of teaching, aiming to help students understand the teaching content. Example: The teacher writes down the formula on the blackboard and then explains the derivation process of the formula in detail. Explain the background, development process and impact of historical events, such as explaining "the causes of World War II and its impact on the world situation". Help students understand the phenomena in physical experiments, such as the principles of refraction and reflection of light.

[0061] Questioning, is the behavior of teachers asking questions to students in the classroom, to stimulate students' thinking, understand students' understanding level or guide them to understand knowledge in depth. Questioning can be divided into open questions and closed questions. Example: Open question: The teacher asks, "What are the advantages of constructivism in actual classroom teaching?" This question has no unique correct answer, aiming to encourage students to express their own ideas. Closed question: The teacher asks, "What is 2 plus 2?" This question usually has only one correct answer, aiming to detect students' mastery of basic knowledge. In a classroom discussion, the teacher asks, "Who can give an example to illustrate the application of the law of conservation of energy in life?"

[0062] Feedback, is the reaction of teachers to students' performance, answers or behavior, aiming to encourage students to continue to work hard, or to correct their mistakes. Feedback can be positive (praise, encouragement) or negative (correction, criticism). Example: When the student correctly answered a question, the teacher said, "Good! Your answer is very comprehensive, keep it up." The student made a mistake in solving a math problem, the teacher said, "Your calculation here is a little wrong, let's take a look together and see where the mistake is, okay?" The student used a very vivid metaphor in his writing, the teacher commented, "Your metaphor is used very well, making the entire description vivid and specific."

[0063] Classroom Management, refers to the behavior of teachers maintaining order in the classroom and organizing teaching activities. It includes maintaining classroom discipline, managing student activities and time scheduling, etc. Effective classroom management is a key factor to ensure the smooth progress of teaching. Example: The teacher reminds the students to pay attention to the class time: "There are 10 minutes left, we need to finish this part as soon as possible"; Assign group tasks among students to ensure that each group member has clear responsibilities: "Your group is responsible for collecting experimental data, the other group is responsible for analyzing the results"; When students start to disperse their attention, the teacher says, "Classmates, please focus here, we are discussing a very important concept."

[0064] The interaction data analysis unit is used to assess the frequency, depth, and quality of teacher-student interactions using social network analysis and multidimensional interaction analysis. The interaction data analysis unit involves modeling interactions between teachers and students, which can be trained using graph data such as teacher-student interaction network graphs. Using social network analysis (SNA) tools, the graph model is trained in combination with classroom interaction data, enabling the model to learn to identify the frequency and depth of interactions between different students and teachers, and to determine whether any students are being ignored, etc.

[0065] The teaching language sentiment analysis unit is used to assess the emotional expression of teachers in the classroom based on sentiment analysis algorithms. The teaching language sentiment analysis unit can use pre-trained sentiment classification models, such as sentiment classifiers based on the Transformer architecture, combined with sentiment-labeled data sets in specific teaching scenarios for fine-tuning training. For example, by using a large amount of labeled classroom teaching language data (containing positive, negative, encouraging, and corrective labels), the model can accurately determine the emotions conveyed by the teacher during the teaching process.

[0066] The teaching feedback strategy analysis unit is used to assess the effectiveness of feedback strategies by analyzing the feedback strategies of teachers to students. In the teaching feedback strategy analysis unit, the content of the teacher's feedback can be used to train a large language model with a dataset labeled with "feedback types" (such as praise, correction, and suggestions) so that it can understand and generate different types of feedback content and evaluate its effectiveness.

[0067] The teaching feedback analysis large model is also used to summarize the analysis results. The system integrates the analysis results of each dimension to generate a comprehensive teaching analysis report, which includes:

[0068] Behavioral performance evaluation: detailed list of teacher performance in different teaching segments, such as the quality of explanation, questioning, and feedback, and points out areas for improvement.

[0069] Interaction chart: generates an interaction network graph or heat map between the teacher and the virtual student, showing the coverage and quality of the interaction.

[0070] Sentiment analysis feedback: feedback on the emotional expression of the teacher in the simulated teaching, helping the teacher improve classroom atmosphere management.

[0071] Improvement suggestions: based on the analysis results, the system provides teaching improvement suggestions, such as enhancing interaction strategies or adjusting question types.

[0072] After the analysis is complete, the system generates a visual report and provides data feedback to the teacher to help them improve their teaching performance.

[0073] The teaching resource recommendation module M4 is configured to recommend teaching skill training resources to teachers based on the teaching analysis report and according to a large language model, so as to help the teachers to further learn and improve teaching methods.

[0074] In this embodiment, the teaching resource recommendation module M4 is configured to recommend teaching skill training resources to teachers based on the teaching analysis report and according to a large language model. The teaching resource recommendation model is constructed based on a large language model, and is obtained after the teaching resource recommendation model is trained using a teaching analysis report data set. As a result, the teaching resource recommendation model can fully understand and analyze the language content in the teaching analysis report, analyze the specific aspects that need to be improved by the teachers, and accurately recommend teaching skill training resources, such as teaching videos, articles, online courses, etc., to the teachers, so as to help the teachers to further learn and improve teaching methods.

[0075] In the teaching resource recommendation module M4, the main function of the teaching resource recommendation model is to understand and analyze the language content in the analysis report, and to make reasonable recommendations based on the analysis results.

[0076] In the training process of the teaching resource recommendation model, the input of the model can include the following types of information: teacher feedback analysis report, teacher background and demand, and teaching resource library tags.

[0077] The text content of the teacher feedback analysis report is a detailed report generated by the teaching feedback analysis module M3, which describes the performance of the teacher in the simulated teaching, including teaching behavior, interaction data, emotional expression and feedback strategy, etc. The weak links or specific fields that need to be improved of the teacher are marked in the feedback report. For example: “insufficient classroom management”, “low frequency of interaction with students”, “non-diversified questioning strategy”, etc.

[0078] The teacher background and demand record the historical performance, course subject, teaching style preference, etc. of the teacher, which are helpful for personalized recommendation.

[0079] The teaching resource library tags include: resource library description: description information and related tags of each resource in the resource library, such as “interactive teaching strategy”, “classroom management video”, “teaching theory article”, etc. Key words and classification tags: tags related to teaching resources and their classification, including teaching stage (such as junior high school, high school), teaching theme (such as classroom management, teaching method), teaching type (video, article, course, etc.) and other information.

[0080] In the training process of the teaching resource recommendation model, the output of the model mainly includes the following contents: generated recommended content, resource summary and applicable scenarios, and personalized adjusted recommended content.

[0081] The generated recommendation content is output in the form of a list of recommended resources. The model outputs a list of recommended resources for the identified problems or weaknesses in the feedback analysis report. Each resource recommendation includes relevant description information to help teachers understand the purpose and applicable scenarios of the recommended resources. The model also generates a brief explanation of why a certain resource is recommended. For example, "due to the teacher's lack of classroom management, a classroom management teaching video is recommended to learn effective classroom control strategies."

[0082] Resource summary and applicable scenarios generate a concise summary for each recommended resource, helping teachers quickly understand the main content and value of the resource. For example, if a video on interactive teaching is recommended, the model generates a brief introduction to the video, such as "this video introduces three different interactive teaching methods suitable for classroom discussion and team cooperation activities." The applicable scenarios of the recommended resources are also generated to ensure that teachers know when to use the recommended resources. For example, "the course design template is suitable for experimental teaching scenarios in junior high school science courses."

[0083] Personalized adjustment of recommended content refers to generating resources for individual needs based on the teacher's background, teaching preferences, and historical usage records. For example, if a teacher is interested in a certain teaching method, the model will prioritize recommending more in-depth learning materials related to that teaching method. For continuous multiple training simulations and feedback of teachers, the model will dynamically adjust the recommended content based on the latest feedback results and the teacher's improvement performance. For example, if a teacher has improved in a certain area, the model can recommend advanced resources to further enhance their skills.

[0084] In addition, it also includes multi-level resource recommendation. First, for basic problems in teacher performance, the system first recommends some basic teaching resources, such as teaching skill videos, case studies, and teaching articles. These resources focus on helping teachers master core teaching skills. For teachers with some teaching experience but need to further improve in certain areas, the system will recommend more in-depth resources, such as online courses, teaching strategy seminars, and teaching research articles. These resources can help teachers gain more in-depth theoretical knowledge and practical experience. If the analysis results show that the teacher performs outstandingly in a specific area but has room for improvement, the system will also recommend some personalized coaching resources, such as academic articles or professional research on specific issues, to help the teacher delve deeper into a certain field.

[0085] It can be understood that, whether in answering the teaching problems raised by the teachers by using the teaching skill learning module, or in recommending the teaching skill training resources to the teachers by using the teaching resource recommendation module, or in adjusting the parameters of the large language model through training to enable it to achieve the predetermined functions, the knowledge resources used can come from Internet search or self-built knowledge base. Of course, under the condition of being legal and compliant, the source of the resources used is not limited thereto, and the present application does not make additional limitations.

[0086] The scheme provided by the above embodiments of the present application can, on the one hand, answer the teaching problems raised by the teachers through the teaching skill learning module M1, and on the other hand, simulate the real teaching scene through the real teaching simulation module M2 to exercise the actual teaching ability of the teachers, in addition, the data in the teaching simulation process are analyzed and quantified through the teaching feedback analysis module M3 to obtain an analysis report and make improvement suggestions for the deficiencies through the teaching resource recommendation module M4, which helps the teachers, especially the novice teachers, to deeply learn and improve the deficiencies of the teaching methods, and improves the response capability of the teachers, especially the novice teachers, when facing diversified student behaviors and demands.

[0087] Of course, Figure 1 The architecture shown is only exemplary, and when implementing different functions, according to actual needs, one or at least two components in the system shown can be omitted. Figure 1 The architecture shown is only exemplary, and when implementing different functions, according to actual needs, one or at least two components in the system shown can be omitted.

[0088] In an exemplary embodiment, a computer device is provided, which can be a server or a terminal, and an internal structure diagram thereof can be as shown. Figure 3 The computer device includes a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through network connection. The computer program is executed by the processor to realize the functions of the above-mentioned embodiment of the individualized teaching skill training system based on the large language model.

[0089] Those skilled in the art can understand that, Figure 3The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0090] In an exemplary embodiment, a computer device is also provided, including a memory and a processor, the memory storing a computer program, and the processor implementing the steps in the above method embodiments when executing the computer program.

[0091] In an exemplary embodiment, a computer readable storage medium is provided, storing a computer program, and the computer program implements the steps in the above method embodiments when executed by a processor.

[0092] In an exemplary embodiment, a computer program product is provided, including a computer program, and the computer program implements the steps in the above method embodiments when executed by a processor.

[0093] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0094] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0095] The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0096] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described, but as long as the combinations of the technical features do not exist, they should be considered as the scope of the present application.

[0097] The principles and implementation modes of the present application are described by applying specific examples herein. The above description of the embodiments is only used to help understand the method and its core idea of the present application; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range can be changed. In conclusion, the content of the present application should not be understood as a limitation.

Claims

1. A personalized teaching skills training system based on a large language model, characterized in that, include: The module includes a teaching skills learning module, a real-world teaching simulation module, a teaching feedback analysis module, and a teaching resource recommendation module. The teaching skills learning module is used to answer teaching questions raised by teachers based on a large language model and a large teaching question-answering model. It provides teaching theories, teaching strategies and / or teaching cases that meet the teaching needs of the teachers to help them learn teaching skills. The large-scale teaching problem-solving model is built upon a large language model and trained using a teaching skills knowledge dataset. The model comprises an input layer, a positional encoding layer, an encoding layer, a multi-head self-attention mechanism, a feedforward neural network layer, a decoding layer, and an output layer. The input layer receives teaching problems from teachers and transforms them into word embeddings that the model can process. The positional encoding layer provides positional information for each word in the teaching problem, ensuring the model understands word order and context when processing text. The encoding layer analyzes and processes the context of the teaching problem through a self-attention mechanism to understand the teacher's teaching needs. The multi-head self-attention mechanism simultaneously focuses on multiple parts of the teaching problem, capturing dependencies between different teaching concepts to ensure the generated answer covers multiple knowledge points. The feedforward neural network layer further processes the contextual representation generated by the multi-head self-attention mechanism, enhancing the model's ability to handle complex teaching problems. The decoding layer generates teaching theories, strategies, and / or teaching cases that meet the teacher's teaching needs based on the model's encoding results. The output layer outputs a probability distribution through a softmax activation function, showing the model's solution to the teaching problem. The real-world teaching simulation module is used to generate several virtual student characters and real-world teaching scenarios based on a large language model and a large teaching scenario generation model. These scenarios are then provided to teachers for online video simulation. During the teaching simulation, the virtual student characters simulate the behavior patterns of real students and provide feedback to the teachers, thereby improving their actual teaching abilities. The large teaching scenario generation model is constructed based on the large language model and trained using a student behavior pattern dataset. The teaching feedback analysis module is used to acquire teaching behavior performance data and teacher-student interaction data during the teaching simulation. Based on the large language model, the teaching feedback analysis large model is used to analyze the teaching behavior performance data and teacher-student interaction data, generate a teaching analysis report, and generate improvement suggestions for the shortcomings in the teaching simulation process. The teaching feedback analysis model is built upon a large language model and trained using a teaching process evaluation dataset. This model can perform multi-dimensional analysis of teaching behavior performance data and teacher-student interaction data using teaching behavior analysis, interaction data analysis, teaching language sentiment analysis, and teaching feedback strategy analysis units. The teaching behavior analysis unit automatically detects and classifies teachers' classroom language using natural language processing technology, analyzing the types and applicability of teaching behaviors. The interaction data analysis unit uses social network analysis and multi-dimensional interaction analysis to evaluate the frequency, depth, and quality of teacher-student interactions reflected in the teacher-student interaction data. The teaching language emotion analysis unit is used to evaluate teachers' language emotion expression in the classroom based on emotion analysis algorithms; the teaching feedback strategy analysis unit is used to evaluate the effectiveness of feedback strategies by analyzing teachers' feedback strategies to students. The teaching resource recommendation module is used to recommend teaching skills training resources to teachers based on the teaching analysis report, using a large language model and a large teaching resource recommendation model, in order to help teachers learn more and improve their teaching methods. The teaching resource recommendation model is built on a large language model and trained using a teaching analysis report dataset. This enables the teaching resource recommendation model to fully understand and parse the language content in the teaching analysis reports, identify the specific aspects that teachers need to improve, and accurately recommend teaching skills training resources to teachers.

2. The personalized teaching skills training system based on a large language model according to claim 1, characterized in that, The teaching skills knowledge dataset is used to train the teaching problem-solving model. Specifically, this includes: using the teaching skills knowledge dataset as input to the teaching problem-solving model; training the model by providing input data and expected output; and adjusting the parameters of the teaching problem-solving model to make it more adaptable to content generation in the education field. The teaching skills knowledge dataset includes teaching theories, teaching strategies, and teaching cases.

3. The personalized teaching skills training system based on a large language model according to claim 1, characterized in that, The well-trained teaching scenario generation model can generate several virtual student characters and real-world teaching scenarios with different behavioral patterns, which can be used by teachers for teaching simulation. During the teaching simulation, the virtual student characters will simulate the behavior patterns of real students and provide feedback to the teachers, thereby improving the teachers' actual teaching abilities.

4. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the functions of the personalized teaching skills training system based on a large language model as described in any one of claims 1-3.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the functions of the personalized teaching skills training system based on a large language model as described in any one of claims 1-3.

6. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the functions of the personalized teaching skills training system based on a large language model as described in any one of claims 1-3.

Images