Method and system for automatically generating teaching plan by using large language model

Abstract

The invention discloses a method and system for automatically generating a teaching plan by using a large language model, and relates to the technical field of artificial intelligence, and the method comprises the steps: extracting and organizing the knowledge of the education field, and constructing a teaching plan knowledge base; optimizing a large language model according to the teaching plan knowledge base to obtain an optimized large language model; and obtaining target information, retrieving matched teaching material contents and teaching plan examples in the teaching plan knowledge base according to the target information, providing the teaching material contents and the teaching plan examples to the optimized large language model in a sub-module manner, and generating a teaching plan meeting actual teaching requirements. Structured knowledge is combined, the accuracy and correlation of teaching plan generation of the large language model are improved, and teachers are helped to improve the lesson preparation efficiency.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and in particular relates to a method and system for automatically generating teaching plans by using a large language model. Background Art

[0002] In the modern education system, lesson plans are an important tool for teachers to conduct classroom teaching. A well-designed lesson plan can not only ensure the structure and coherence of teaching content, but also significantly improve the overall effectiveness of the teaching process. Traditional lesson plans usually include multiple components such as teaching objectives, teaching materials, detailed teaching activities, and evaluation methods. These elements together form a blueprint for what teachers should teach in class, how to teach, and how to evaluate teaching effectiveness. However, despite the self-evident importance of lesson plans, teachers often face challenges such as tight time and limited resources when preparing lesson plans.

[0003] The traditional method of designing lesson plans relies on manual editing, which is not only time-consuming but also requires teachers to have high professional knowledge. This is very challenging for novice teachers and those in areas with scarce educational resources.

[0004] With the remarkable performance of large language models (LLMs) in language understanding and text generation, some current methods attempt to use LLMs to automatically generate lesson plans. However, lesson plan generation based on LLMs still faces some challenges. First, lesson plans have a highly structured format, including clear teaching objectives, detailed teaching steps, and specific evaluation methods. Due to its openness and instability, the content generated by LLMs may not guarantee the structural integrity of all modules. Second, LLMs lack practical teaching experience and the ability to perceive the classroom environment. The generated lesson plans may not fully consider students' learning backgrounds, interests, and needs, and it is difficult to provide personalized teaching strategies. In addition, lesson plan generation is a knowledge-intensive problem that requires scientific teaching principles and subject expertise, while LLMs usually lack sufficient domain-specific educational knowledge and are prone to generating inaccurate teaching content.

[0005] In general, most existing methods usually adopt one of two "extreme" approaches when solving the lesson plan generation problem: either using rule-based lesson plan template designs that are difficult to scale, or relying on simple prompts to guide LLMs, often ignoring key contextual information such as grade level, textbook information, and individual teaching requirements. Summary of the invention

[0006] To solve the above technical problems, the present invention proposes a method and system for automatically generating lesson plans using a large language model, which combines structured knowledge to improve the accuracy and relevance of lesson plans generated by the large language model, helping teachers improve their lesson preparation efficiency.

[0007] On the one hand, to achieve the above-mentioned purpose, the present invention provides a method for automatically generating a teaching plan using a large language model, comprising:

[0008] Extract and organize knowledge in the field of education and build a teaching plan knowledge base;

[0009] Optimizing the large language model according to the teaching plan knowledge base to obtain an optimized large language model;

[0010] Obtain target information, retrieve matching teaching material content and lesson plan example modules from the lesson plan knowledge base according to the target information, and provide them to the optimized large language model. The large language model module generates lesson plans that meet actual teaching needs.

[0011] Optionally, building a teaching plan knowledge base includes:

[0012] Divide the teaching plan into several teaching modules, and unify the data structure of the divided teaching modules;

[0013] Extract knowledge structure information based on various textbooks in different regions;

[0014] Extract textbook content, and the extracted textbook content corresponds to the knowledge structure.

[0015] Optionally, optimizing the large language model according to the teaching plan knowledge base includes:

[0016] Constructing a teaching plan instruction data set according to the teaching plan knowledge base;

[0017] Through the teaching plan instruction data set, the supervised optimization large language model is used to perform context relevance learning to complete the optimization of the large language model.

[0018] Optionally, generate lesson plans that meet actual teaching needs including:

[0019] Acquire target information, and correspond the acquired target information to the knowledge structure information;

[0020] Retrieve the corresponding teaching material content and teaching plan examples in the constructed teaching plan knowledge base according to the specific structural information;

[0021] The teaching material content and lesson plan examples retrieved according to the target information are provided to the large language model in modules, and the large language model generates lesson plans that meet actual teaching needs in modules.

[0022] On the other hand, to achieve the above-mentioned purpose, the present invention also provides a system for automatically generating teaching plans using a large language model, comprising: a teaching plan knowledge base construction module, a model optimization module and a teaching plan generation module;

[0023] The teaching plan knowledge base construction module is used to extract and organize knowledge in the field of education and construct a teaching plan knowledge base;

[0024] The model optimization module is used to optimize the large language model according to the teaching plan knowledge base to obtain an optimized large language model;

[0025] The lesson plan generation module is used to obtain target information, retrieve matching teaching material content and lesson plan examples in the lesson plan knowledge base according to the target information, and provide them to the optimized large language model in modules. The large language model generates lesson plans that meet actual teaching needs in modules.

[0026] Optionally, the teaching plan knowledge base construction module includes a teaching plan module extraction unit, a knowledge structure extraction unit and a teaching material content extraction unit;

[0027] The teaching plan module extraction unit is used to divide the teaching plan into a plurality of teaching modules and unify the divided teaching modules into a unified data structure;

[0028] The knowledge structure extraction unit is used to extract knowledge structure information based on various teaching materials in different regions;

[0029] The teaching material content extraction unit is used to extract teaching material content, and the extracted teaching material content corresponds to the knowledge structure.

[0030] Optionally, optimizing the large language model according to the teaching plan knowledge base includes:

[0031] Constructing a teaching plan instruction data set according to the teaching plan knowledge base;

[0032] Through the teaching plan instruction data set, the supervised optimization large language model is used to perform context relevance learning to complete the optimization of the large language model.

[0033] Optionally, the teaching plan generation module includes an information input unit, a knowledge base retrieval enhancement generation unit and a module teaching plan generation unit;

[0034] The information input unit is used to obtain target information, and the target information corresponds to the knowledge structure information;

[0035] The knowledge base retrieval enhancement generation unit is used to retrieve corresponding teaching material content and teaching plan examples in the constructed teaching plan knowledge base according to specific structural information;

[0036] The module-based lesson plan generation unit is used to provide the relevant teaching material content and lesson plan examples retrieved according to the target information to the large language model in modules, and the large language model generates lesson plans that meet actual teaching needs in modules.

[0037] Technical effect of the present invention: The present invention discloses a method and system for automatically generating lesson plans using a large language model, which realizes the automatic generation of high-quality lesson plans by constructing a lesson plan knowledge base and optimization technology, using retrieval enhancement generation (RAG) technology and a modular generation method. Specifically, the present invention can systematically extract and organize knowledge in the field of education, enhance the teaching theory and subject expertise of the large language model, and thus generate lesson plans that meet actual teaching needs. Through the information input module, teachers can quickly and accurately input necessary teaching information, reducing the tediousness and errors in the input process. The retrieval enhancement generation technology ensures the accuracy and relevance of the generated lesson plan content, while the modular generation method ensures the consistency and coherence of the content of each module of the lesson plan. Finally, the present invention can generate detailed, specific and practical lesson plans, significantly improving the teacher's lesson preparation efficiency and teaching effect. The present invention has broad application prospects in multiple educational fields. First, it can be applied to daily teaching activities in schools of all levels and types, helping teachers to quickly generate high-quality lesson plans, reduce the burden of lesson preparation, and improve teaching quality. Secondly, the present invention can also be applied to educational training institutions to provide trainers with standardized and personalized teaching plans and improve training effects. In addition, the present invention can also be applied to online education platforms, by automatically generating teaching plans, enriching teaching resources, and improving the quality and attractiveness of online courses. In short, the present invention can play an important role in various application scenarios in the field of education and promote the development of educational informatization and intelligence. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0039] Figure 1 A flow chart of a method for automatically generating a teaching plan using a large language model according to an embodiment of the present invention;

[0040] Figure 2 The present invention is a schematic diagram of the structure of a system for automatically generating teaching plans using a large language model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0042] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0043] like Figure 1 As shown, this embodiment provides a method for automatically generating teaching plans using a large language model, including: extracting and organizing knowledge in the field of education, and constructing a teaching plan knowledge base;

[0044] Optimizing the large language model according to the teaching plan knowledge base to obtain an optimized large language model;

[0045] Obtain target information, retrieve matching teaching material content and lesson plan examples in the lesson plan knowledge base according to the target information, and provide them to the optimized large language model in modules. The large language model generates lesson plans that meet actual teaching needs in modules.

[0046] Furthermore, building a teaching plan knowledge base includes:

[0047] Divide the teaching plan into several teaching modules, and unify the data structure of the divided teaching modules;

[0048] Extract knowledge structure information based on various textbooks in different regions;

[0049] Extract textbook content, and the extracted textbook content corresponds to the knowledge structure.

[0050] Specifically, the construction of the teaching plan knowledge base aims to systematically extract and organize knowledge in the field of education to enhance the teaching theory and subject expertise of LLMs, thereby generating high-quality teaching plans. This module mainly includes the following three parts:

[0051] Since each lesson plan is usually a long text sequence, these lesson plans are divided into smaller, manageable teaching modules, including teaching content analysis, student situation analysis, teaching objectives, teaching difficulties, teaching preparation, teaching process, and teaching evaluation, so as to extract text information and solve the context length limit of LLMs. Since these lesson plans are written by teachers of different majors, there are great differences in writing habits and styles. In order to standardize the structure of generated lesson plans, a unified data structure is designed for each teaching module.

[0052] Lesson plan generation is a knowledge-intensive problem, and the generated content is highly relevant to the given inputs (such as grade, subject, and knowledge unit). The knowledge information provided is organized in a large number of textbooks in real educational scenarios. In order to generate in-domain lesson plans that are consistent with the knowledge structure information, knowledge structure information including grade, subject, textbook version, and chapter list was extracted from various textbooks in different regions (such as elementary school, mathematics, People's Education Edition (2024), first grade, recognition of numbers within 5 and addition and subtraction, recognition of 1 to 5, and comparison of size). Detailed structural information enables the model to generate lesson plans that are highly relevant and meet the personalized teaching needs of teachers. In order to ensure the consistency between the lesson plan content and the course knowledge structure, each lesson plan is mapped to a unique knowledge structure information.

[0053] Lesson plan resources are limited, and not all knowledge structures have corresponding lesson plan resources. In order to make up for this deficiency and avoid the illusion problem of LLMs, and ensure that the generated lesson plan content is comprehensive and accurate, in addition to corresponding relevant lesson plan resources on each knowledge structure, the previous textbook content is also matched to provide the model with more reliable and comprehensive reference information, and the generated lesson plan content is also more accurate. Textbook content usually contains a variety of content such as text, pictures, tables, formulas, etc. In order to ensure the comprehensiveness and accuracy of textbook information, a multimodal large language model is used to perform structured extraction of textbook content, and the textbook content of each page is converted into pictures and input into the multimodal large language model, requiring the multimodal large language model to extract the corresponding textbook content. For text information (text, tables, formulas), it needs to be completely consistent with the textbook, and the formula is required to use the latex format. For picture information, the multimodal large language model is required to describe the picture content with text. An example of extracted textbook content is shown in Table 1, in which some content is omitted with ellipsis.

[0054] Table 1

[0055]

[0056]

[0057] Further, optimizing the large language model according to the teaching plan knowledge base includes:

[0058] Constructing a teaching plan instruction data set according to the teaching plan knowledge base;

[0059] Through the teaching plan instruction data set, the supervised optimization large language model is used to perform context relevance learning to complete the optimization of the large language model.

[0060] Specifically, in the present invention, it is intended to adapt LLMs to the field of education to generate acceptable lesson plans and help teachers improve teaching efficiency. At present, supervised optimization (SFT) for contextual learning is widely used to enhance the knowledge of specific fields of LLMs, and the quality of optimization data is a key factor in determining the effect of model generation. In order to ensure that the model can generate lesson plans that meet the needs of the education field, a high-quality lesson plan instruction dataset is constructed using the constructed lesson plan knowledge base, and the instructions are shown in Table 2.

[0061] Table 2

[0062]

[0063] Through the instructions, the training data for optimizing the large language model was constructed, and the training examples for teaching content analysis are shown in Table 3.

[0064] Table 3

[0065]

[0066] In the above optimization instruction example, not only the knowledge structure information of the lesson plan is provided, but also the teaching material content is added according to the knowledge structure. The constructed teaching plan instruction data set is input into the large language model. Through multiple rounds of iterative training strategies, the large language model not only learns how to generate teaching plans according to the provided knowledge structure, but also learns how to process and integrate the retrieved teaching material content to improve the accuracy and context relevance of the generated teaching plan.

[0067] Furthermore, the lesson plans generated to meet actual teaching needs include:

[0068] Acquire target information, and correspond the acquired target information to the knowledge structure information;

[0069] Retrieve the corresponding teaching material content and teaching plan examples in the constructed teaching plan knowledge base according to the specific structural information;

[0070] The teaching material content and lesson plan examples retrieved according to the target information are provided to the large language model in modules, and the large language model generates lesson plans that meet actual teaching needs in modules.

[0071] Specifically, lesson plan generation aims to use the textbook knowledge base and retrieval-augmented generation (RAG) technology to enable LLMs to generate high-quality lesson plans that meet actual teaching needs. This module mainly includes the following three parts:

[0072] Before LLMs generates lesson plans, it is a prerequisite to provide complete and necessary information, such as grade, subject, and course title. In order to avoid cumbersome input process and reduce errors, the knowledge structure information is connected in the information input module, providing a drop-down input method. The specific method is as follows:

[0073] First, when using the system, teachers can select necessary information such as grade, subject, and course title through the drop-down menu. These options are automatically generated based on pre-built knowledge structure information, ensuring that teachers can quickly and accurately enter the required information. For example, teachers can select "First Grade" in the "Grade" drop-down menu, then select "Mathematics" in the "Subject" drop-down menu, and finally select "1 to 5 Knowledge and Addition and Subtraction" in the "Chapter" drop-down menu.

[0074] Secondly, in order to further simplify the input process, the system integrates an intelligent prompt function. When the teacher selects a certain information, the system will automatically provide relevant next options based on the selected information. For example, when the teacher selects the subject of "mathematics", the system will automatically display the course topics related to mathematics, reducing the teacher's selection range and improving input efficiency.

[0075] In this way, information input not only improves the accuracy and efficiency of input, but also reduces the burden on teachers during the input process, ensuring that LLMs can obtain complete and accurate input information before generating lesson plans.

[0076] The corresponding teaching material content is retrieved according to the specific structural information. This process is realized through the retrieval enhancement generation (RAG) technology. Based on the information such as grade, subject and course title input by the teacher, the system retrieves the matching teaching material content and examples from the teaching plan knowledge base and provides them to LLMs as a reference for generating teaching plans.

[0077] First, the system will use the pre-built knowledge structure information to search based on the grade, subject, and course title information entered by the teacher. For example, when the teacher selects the course title "Growth of Science Plants in Grade 3 of Primary School", the system will retrieve the matching teaching material content from the lesson plan knowledge base. Secondly, in order to ensure the quality of generation, the system will also retrieve examples that fully match or are related to the input information. Finally, the system will input the retrieved teaching material content and high-quality examples into LLMs as references to assist it in generating lesson plan content.

[0078] The lesson plan is composed of different modules, such as teaching content analysis, teaching objectives, teaching process, etc. These modules do not exist independently, and different modules have premise and dependency management. For example, teaching objectives and teaching difficulties are usually the guidance for designing the teaching process. In addition, corresponding teaching preparations will be prepared according to the teaching process. At the same time, the lesson plan is a long text. Generating a complete lesson plan at one time may result in the generated content being too brief and not specific enough due to the word limit, affecting the overall quality and practicality.

[0079] In order to ensure the quality and consistency of the teaching plan content, the teaching plan content is divided into modules, and the separate module content is generated each time. For modules with dependencies, the content of the prerequisite module is provided to the pre-generated module to ensure the consistency of the content. For example, the teaching objectives and teaching key points are generated first, and then when generating the teaching process, the generated teaching objectives and teaching key points are provided to LLMs, and the teaching preparation content is generated according to the generated teaching process.

[0080] Through lesson plan generation, the lesson plan knowledge base and retrieval enhancement generation technology can be used to allow LLMs to generate high-quality lesson plans that meet actual teaching needs. This not only improves the accuracy and relevance of the lesson plans, but also enhances the consistency and practicality of the content of each module of the lesson plans, thereby helping teachers improve lesson preparation efficiency and teaching effectiveness.

[0081] like Figure 2 As shown, this embodiment also provides a system for automatically generating teaching plans using a large language model, including: a teaching plan knowledge base construction module, a model optimization module and a teaching plan generation module;

[0082] The teaching plan knowledge base construction module is used to extract and organize knowledge in the field of education and construct a teaching plan knowledge base;

[0083] The model optimization module is used to optimize the large language model according to the teaching plan knowledge base to obtain an optimized large language model;

[0084] The lesson plan generation module is used to obtain target information, retrieve matching teaching material content and lesson plan examples in the lesson plan knowledge base according to the target information, and provide them to the optimized large language model in modules. The large language model generates lesson plans that meet actual teaching needs in modules.

[0085] Furthermore, the teaching plan knowledge base construction module includes a teaching plan module extraction unit, a knowledge structure extraction unit and a teaching material content extraction unit;

[0086] The teaching plan module extraction unit is used to divide the teaching plan into a plurality of teaching modules and unify the divided teaching modules into a unified data structure;

[0087] The knowledge structure extraction unit is used to extract knowledge structure information based on various teaching materials in different regions;

[0088] The teaching material content extraction unit is used to extract teaching material content, and the extracted teaching material content corresponds to the knowledge structure.

[0089] Specifically, the lesson plan knowledge base construction module aims to systematically extract and organize knowledge in the field of education to enhance the teaching theory and subject expertise of LLMs, thereby generating high-quality lesson plans. This module mainly includes the following three parts:

[0090] Since each lesson plan is usually a long text sequence, these lesson plans are divided into smaller, manageable teaching modules, including teaching content analysis, student situation analysis, teaching objectives, teaching difficulties, teaching preparation, teaching process, and teaching evaluation, so as to extract text information and solve the context length limit of LLMs. Since these lesson plans are written by teachers of different majors, there are great differences in writing habits and styles. In order to standardize the structure of generated lesson plans, a unified data structure is designed for each teaching module.

[0091] Lesson plan generation is a knowledge-intensive problem, and the generated content is highly relevant to the given inputs (such as grade, subject, and knowledge unit). The knowledge information provided is organized in a large number of textbooks in real educational scenarios. In order to generate in-domain lesson plans that are consistent with the knowledge structure information, knowledge structure information including grade, subject, textbook version, and chapter list was extracted from various textbooks in different regions (such as elementary school, mathematics, People's Education Edition (2024), first grade, recognition of numbers within 5 and addition and subtraction, recognition of 1 to 5, and comparison of size). Detailed structural information enables the model to generate lesson plans that are highly relevant and meet the personalized teaching needs of teachers. In order to ensure the consistency between the lesson plan content and the course knowledge structure, each lesson plan is mapped to a unique knowledge structure information.

[0092] Lesson plan resources are limited, and not all knowledge structures have corresponding lesson plan resources. In order to make up for this deficiency and avoid the illusion problem of LLMs, and ensure that the generated lesson plan content is comprehensive and accurate, in addition to corresponding relevant lesson plan resources for each knowledge structure, the previous textbook content is also matched to provide more reliable and comprehensive reference information for the model, and the generated lesson plan content is also more accurate.

[0093] Further, optimizing the large language model according to the teaching plan knowledge base includes:

[0094] Constructing a teaching plan instruction data set according to the teaching plan knowledge base;

[0095] Through the teaching plan instruction data set, the supervised optimization large language model is used to perform context relevance learning to complete the optimization of the large language model.

[0096] Specifically, in the present invention, it is intended to adapt LLMs to the field of education to generate acceptable lesson plans and help teachers improve teaching efficiency. At present, supervised optimization (SFT) for contextual learning is widely used to enhance the knowledge of specific fields of LLMs, and the quality of optimization data is a key factor in determining the effect of model generation. In order to ensure that the model can generate lesson plans that meet the needs of the education field, a high-quality lesson plan instruction dataset is constructed using the constructed lesson plan knowledge base, and the instructions are shown in Table 4.

[0097] Table 4

[0098]

[0099]

[0100] In the above optimization instruction example, not only the knowledge structure information of the lesson plan is provided, but also the teaching material content is added according to the knowledge structure. Through the lesson plan instruction dataset, the model not only learns how to generate lesson plans according to the provided knowledge structure, but also learns how to process and integrate the retrieved teaching material content to improve the accuracy and contextual relevance of the generated lesson plans.

[0101] Furthermore, the teaching plan generation module includes an information input unit, a knowledge base retrieval enhancement generation unit and a module teaching plan generation unit;

[0102] The information input unit is used to obtain target information, and the target information corresponds to the knowledge structure information;

[0103] The knowledge base retrieval enhancement generation unit is used to retrieve corresponding teaching material content and teaching plan examples in the constructed teaching plan knowledge base according to specific structural information;

[0104] The module-based lesson plan generation unit is used to provide the relevant teaching material content and lesson plan examples retrieved according to the target information to the large language model in modules, and the large language model generates lesson plans that meet actual teaching needs in modules.

[0105] Specifically, the lesson plan generation module aims to use the textbook knowledge base and retrieval-augmented generation (RAG) technology to enable LLMs to generate high-quality lesson plans that meet actual teaching needs. This module mainly includes the following three parts:

[0106] Before LLMs generates lesson plans, it is a prerequisite to provide complete and necessary information, such as grade, subject, and course title. In order to avoid cumbersome input process and reduce errors, the knowledge structure information is connected in the information input module, providing a drop-down input method. The specific method is as follows:

[0107] First, when using the system, teachers can select necessary information such as grade, subject, and course title through the drop-down menu. These options are automatically generated based on pre-built knowledge structure information, ensuring that teachers can quickly and accurately enter the required information. For example, teachers can select "First Grade" in the "Grade" drop-down menu, then select "Mathematics" in the "Subject" drop-down menu, and finally select "1 to 5 Knowledge and Addition and Subtraction" in the "Chapter" drop-down menu.

[0108] Secondly, in order to further simplify the input process, the system integrates an intelligent prompt function. When the teacher selects a certain information, the system will automatically provide relevant next options based on the selected information. For example, when the teacher selects the subject of "mathematics", the system will automatically display the course topics related to mathematics, reducing the teacher's selection range and improving input efficiency.

[0109] In this way, the information input unit not only improves the accuracy and efficiency of input, but also reduces the burden on teachers during the input process, ensuring that LLMs can obtain complete and accurate input information before generating lesson plans.

[0110] The corresponding teaching material content is retrieved according to the specific structural information. This process is realized through the retrieval enhancement generation (RAG) technology. Based on the information such as grade, subject and course title input by the teacher, the system retrieves the matching teaching material content and examples from the teaching plan knowledge base and provides them to LLMs as a reference for generating teaching plans.

[0111] First, the system will use the pre-built knowledge structure information to search based on the grade, subject, and course title information entered by the teacher. For example, when the teacher selects the course title "Growth of Science Plants in Grade 3 of Primary School", the system will retrieve the matching teaching material content from the lesson plan knowledge base. Secondly, in order to ensure the quality of generation, the system will also retrieve examples that fully match or are related to the input information. Finally, the system will input the retrieved teaching material content and high-quality examples into LLMs as references to assist it in generating lesson plan content.

[0112] The lesson plan is composed of different modules, such as teaching content analysis, teaching objectives, teaching process, etc. These modules do not exist independently, and different modules have premise and dependency management. For example, teaching objectives and teaching difficulties are usually the guidance for designing the teaching process. In addition, corresponding teaching preparations will be prepared according to the teaching process. At the same time, the lesson plan is a long text. Generating a complete lesson plan at one time may result in the generated content being too brief and not specific enough due to the word limit, affecting the overall quality and practicality.

[0113] In order to ensure the quality and consistency of the teaching plan content, the teaching plan content is divided into modules, and the separate module content is generated each time. For modules with dependencies, the content of the prerequisite module is provided to the pre-generated module to ensure the consistency of the content. For example, the teaching objectives and teaching key points are generated first, and then when generating the teaching process, the generated teaching objectives and teaching key points are provided to LLMs, and the teaching preparation content is generated according to the generated teaching process.

[0114] Through the lesson plan generation module, LLMs can use the lesson plan knowledge base and retrieval enhancement generation technology to generate high-quality lesson plans that meet actual teaching needs. The lesson plan generation module not only improves the accuracy and relevance of the lesson plan, but also enhances the consistency and practicality of the content of each module of the lesson plan, thereby helping teachers improve lesson preparation efficiency and teaching effectiveness.

[0115] The present invention realizes the automatic generation of high-quality lesson plans by constructing a lesson plan knowledge base and optimization technology, utilizing retrieval enhancement generation (RAG) technology and a modular generation method. Specifically, the present invention can systematically extract and organize knowledge in the field of education, enhance the teaching theory and subject expertise of LLMs, and thus generate lesson plans that meet actual teaching needs. Through the information input module, teachers can quickly and accurately input necessary teaching information, reducing the tediousness and errors in the input process. The retrieval enhancement generation technology ensures the accuracy and relevance of the generated lesson plan content, while the modular generation method ensures the consistency and coherence of the content of each module of the lesson plan. Finally, the present invention can generate detailed, specific and practical lesson plans, significantly improving the teacher's lesson preparation efficiency and teaching effect.

[0116] The present invention has a wide range of application prospects in multiple educational fields. First, the present application can be applied to daily teaching activities in schools of all levels and types, helping teachers to quickly generate high-quality lesson plans, reduce the burden of lesson preparation, and improve the quality of teaching. Secondly, the present invention can also be applied to educational training institutions to provide trainers with standardized and personalized teaching plans to improve training results. In addition, the present invention can also be applied to online education platforms, by automatically generating lesson plans, enriching teaching resources, and improving the quality and attractiveness of online courses. In short, the present invention can play an important role in a variety of application scenarios in the field of education, and promote the development of educational informatization and intelligence.

[0117] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for automatically generating teaching plans using a large language model, characterized in that: include: Extract and organize knowledge in the field of education and build a teaching plan knowledge base; Optimizing the large language model according to the teaching plan knowledge base to obtain an optimized large language model; Obtain target information, retrieve matching teaching material content and lesson plan examples in the lesson plan knowledge base according to the target information, and provide them to the optimized large language model in modules. The large language model generates lesson plans that meet actual teaching needs in modules.

2. The method for automatically generating teaching plans using a large language model as claimed in claim 1, characterized in that: Building a teaching plan knowledge base includes: Divide the teaching plan into several teaching modules, and unify the data structure of the divided teaching modules; Extract knowledge structure information based on various textbooks in different regions; Extract textbook content, and the extracted textbook content corresponds to the knowledge structure.

3. The method for automatically generating teaching plans using a large language model as claimed in claim 1, characterized in that: Optimizing the large language model according to the teaching plan knowledge base includes: Constructing a teaching plan instruction data set according to the teaching plan knowledge base; Through the teaching plan instruction data set, the supervised optimization large language model is used to perform context relevance learning to complete the optimization of the large language model.

4. The method for automatically generating teaching plans using a large language model as claimed in claim 1, characterized in that: Generating lesson plans that meet actual teaching needs includes: Acquire target information, and correspond the acquired target information to the knowledge structure information; Retrieve the corresponding teaching material content and teaching plan examples in the constructed teaching plan knowledge base according to the specific structural information; The teaching material content and lesson plan examples retrieved according to the target information are provided to the large language model in modules, and the large language model generates lesson plans that meet actual teaching needs in modules.

5. A system for automatically generating teaching plans using a large language model, characterized in that: The system is used to implement the method for automatically generating teaching plans using a large language model as described in any one of claims 1 to 4, comprising: Lesson plan knowledge base construction module, model optimization module and lesson plan generation module; The teaching plan knowledge base construction module is used to extract and organize knowledge in the field of education and construct a teaching plan knowledge base; The model optimization module is used to optimize the large language model according to the teaching plan knowledge base to obtain an optimized large language model; The lesson plan generation module is used to obtain target information, retrieve matching teaching material content and lesson plan examples in the lesson plan knowledge base according to the target information, and provide them to the optimized large language model in modules. The large language model generates lesson plans that meet actual teaching needs in modules.

6. The system for automatically generating teaching plans using a large language model as claimed in claim 5, characterized in that: The teaching plan knowledge base construction module includes a teaching plan module extraction unit, a knowledge structure extraction unit and a teaching material content extraction unit; The teaching plan module extraction unit is used to divide the teaching plan into a plurality of teaching modules and unify the divided teaching modules into a unified data structure; The knowledge structure extraction unit is used to extract knowledge structure information based on various teaching materials in different regions; The teaching material content extraction unit is used to extract teaching material content, and the extracted teaching material content corresponds to the knowledge structure.

7. The system for automatically generating teaching plans using a large language model as claimed in claim 5, characterized in that: Optimizing the large language model according to the teaching plan knowledge base includes: Constructing a teaching plan instruction data set according to the teaching plan knowledge base; Through the teaching plan instruction data set, the supervised optimization large language model is used to perform context relevance learning to complete the optimization of the large language model.

8. The system for automatically generating teaching plans using a large language model as claimed in claim 5, characterized in that: The teaching plan generation module includes an information input unit, a knowledge base search enhancement generation unit and a module teaching plan generation unit; The information input unit is used to obtain target information, and the target information corresponds to the knowledge structure information; The knowledge base retrieval enhancement generation unit is used to retrieve corresponding teaching material content and teaching plan examples in the constructed teaching plan knowledge base according to specific structural information; The module-based lesson plan generation unit is used to provide the relevant teaching material content and lesson plan examples retrieved according to the target information to the large language model in modules, and the large language model generates lesson plans that meet actual teaching needs in modules.

Images