Language large model-based teaching plan generation method and system, and readable storage medium

Abstract

The invention provides a teaching plan generation method and system based on a language large model and a readable storage medium, and relates to the technical field of education, and the method comprises the steps: firstly presetting a teaching plan grading standard containing at least two hierarchical definitions; after obtaining course information, searching a preset number of reference teaching plans, and extracting first-level labels of the reference teaching plans according to a grading standard teaching plan grading standard; then, in combination with a preset first control prompt word and a teaching plan grading standard, calling the language large model to generate a first-edition teaching plan base point, and optimizing to obtain a final-edition teaching plan base point; and finally, according to a preset second control prompt word, the final-edition teaching plan base point and the first level label, calling the language large model comprehensive reference teaching plan to generate a first-edition teaching plan, and optimizing to obtain a final-edition teaching plan. By implementing the method, a high-quality teaching plan with professional depth and personalized requirements can be efficiently generated, the professionality and pertinence of teaching plan generation are improved, and the efficiency of teacher design is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of education, and in particular to a lesson plan generation method and system based on a language large model and a readable storage medium. BACKGROUND

[0002] Teaching design is the core work of pre-class preparation, and a good teaching design generates a personalized lesson plan required for the class, which is the key to a good class. However, designing a good lesson plan is not an easy task, and it takes a lot of time to search for reference lesson plans, read and understand them in depth, and then design a personalized lesson plan that meets the needs of the class.

[0003] In recent years, with the maturity of language large models such as ChatGPT and DeepSeek, many teachers have begun to try to use language large models to design lesson plans in order to improve the efficiency of lesson plan design. However, lesson plan design is a very professional work in the field of education and also needs to be designed specifically and individually for the class. The lesson plan generated by simple prompts often lacks content, is not specific enough, and is not professional enough. Teachers still need to spend a lot of time modifying the lesson plan to meet their needs, and the efficiency of work is limited.

[0004] In summary, the existing technology has the problem of insufficient specificity and lack of professionalism in the lesson plan generated based on the language large model. The present application provides a lesson plan generation method based on a language large model to improve the efficiency and quality of teachers designing lesson plans. SUMMARY

[0005] The present application provides a lesson plan generation method and system based on a language large model and a readable storage medium to solve the problem of low efficiency in the prior art that teachers rely on language large models to generate lesson plans, which leads to a lack of content, specificity, and professionalism in the lesson plan, and thus a lot of time is spent searching, understanding, and modifying the reference lesson plan to generate the lesson plan.

[0006] In a first aspect, the present application provides a lesson plan generation method based on a language large model, which comprises: presetting a lesson plan grading standard, the lesson plan grading standard comprising at least two level definitions; after obtaining course information, searching for a preset number of lesson plans as reference lesson plans according to the course information; extracting a first level label of the reference lesson plans according to the lesson plan grading standard; calling a language large model to generate a preliminary lesson plan base point according to a preset first control prompt word and the lesson plan grading standard; receiving a first optimization instruction for the preliminary lesson plan base point, optimizing the preliminary lesson plan base point according to the optimization instruction, and generating a final lesson plan base point; calling the language large model to comprehensively analyze the reference lesson plans according to a preset second control prompt word, the final lesson plan base point and the first level label, and generating a preliminary lesson plan; receiving a second optimization instruction for the preliminary lesson plan, optimizing the preliminary lesson plan, and generating a final lesson plan.

[0007] The above technical solution first presets a lesson plan grading standard comprising at least two level definitions, thereby providing a unified and clear basis for subsequent lesson plan generation, and then generates a final lesson plan through a two-stage method, first generates a lesson plan base point to determine the specific requirements of the course, and then generates a lesson plan to avoid the generated lesson plan content being too general and not specific enough. Specifically, after obtaining the course information, the reference lesson plans are searched to provide basic material support for the language large model to fit the course, thereby reducing the disconnection between the generated content of the large model and the course. Then, the first level label of the reference lesson plans is extracted according to the lesson plan grading standard, so that the characteristics of the reference lesson plans are clearer and more accurate for subsequent fine-tuning. Then, the language large model is called to generate a preliminary lesson plan base point to determine the specific teaching requirements or teaching goals of the course, and the lesson plan grading standard is relied on to ensure the professionalism and content integrity of the generated preliminary lesson plan base point. The first control prompt word is also used to constrain the generation direction of the large model. Then, the first optimization instruction is received to generate a final lesson plan base point, which integrates the teaching experience of teachers and compensates for the cognitive bias of the language large model for specific courses. Then, the second control prompt word, the final lesson plan base point and the first level label are combined to generate a preliminary lesson plan, the effective content related to the course in the reference lesson plans is extracted according to the final lesson plan base point and the first level label, and the preliminary lesson plan is generated based on these effective contents, so that the content of the preliminary lesson plan is comprehensive, professional and specific to the course. Finally, the final lesson plan is obtained through the optimization of the teacher, which not only takes advantage of the efficient generation capability of the language large model, but also compensates for the possible shortcomings of the model through the intervention of the teacher, and the man-machine cooperation is organic, which greatly improves the specificity, professionalism and efficiency of the lesson plan generation, reduces the teacher's preparation time cost, and at the same time ensures that the quality of the lesson plan can meet the actual teaching needs.

[0008] In combination with some embodiments of the first aspect, in some embodiments, the lesson plan grading standard comprises two or more grading dimensions, and each grading dimension comprises at least two level definitions.

[0009] By adopting the above technical solution, the multiple analysis dimensions under the grading standard of the teaching plan can cover key elements of teaching plan design from different angles, such as student learning situation, teaching objectives, teaching methods, and the like, avoiding one-sidedness caused by single-dimensional evaluation. Each grading dimension is further divided into different levels, which can accurately locate the specific level of students in different aspects. It can distinguish the subtle differences in different teaching scenarios and correspond to the teaching needs of different student groups. This refined grading standard can enable the language large model to more accurately match the actual teaching scenario when generating a teaching plan in the future, and can also more clearly guide teachers to clarify course requirements and more accurately and comprehensively describe course requirements, avoiding content deviation caused by ambiguous standards.

[0010] In some embodiments in combination with some embodiments of the first aspect, the teaching plan grading standard at least includes a student learning situation grading dimension and / or a teaching objective grading dimension, and the initial version of the teaching plan base point at least includes student learning situation analysis design and / or teaching objective design.

[0011] By adopting the above technical solution, student learning situation analysis is the key to "teaching according to learning", and the student learning situation grading dimension can accurately analyze the student situation from multiple dimensions in the subsequent language large model, avoiding the general description of student learning situation in traditional teaching plans, and making the subsequent teaching activity design more in line with the actual ability of students. The teaching objective grading dimension sets a clear teaching achievement orientation for the teaching plan, which can avoid the problem of ambiguous or student-level target setting, and ensure that the teaching activities are carried out around clear and reasonable objectives. The initial version of the teaching plan base point generates content for at least one of the two core links, effectively solving the problem of disconnection between teaching links and student needs and / or teaching objectives in existing teaching plans, making the teaching plan more practical and instructive, and improving the teaching effect.

[0012] In some embodiments in combination with some embodiments of the first aspect, according to the teaching plan grading standard, the step of extracting the first level label of the reference teaching plan specifically includes: presetting a third control prompt word; controlling the language large model to extract the first teaching plan content of the reference teaching plan according to the third control prompt word, the first teaching plan content being content related to the teaching plan grading standard; determining the level of the first teaching plan content corresponding to the teaching plan grading standard; and determining the first level label of the reference teaching plan according to the level of the first teaching plan content.

[0013] By adopting the technical solution, the third control prompt word can accurately guide the extraction behavior of the language large model, avoid the model from extracting irrelevant content when processing the reference teaching plan, and ensure that the first teaching plan content related to the teaching plan grading standard in the reference teaching plan is accurately extracted. Further judging the level of the corresponding teaching plan grading standard can convert the implicit information (such as the implied learning situation adaptation degree and teaching goal level) of the reference teaching plan into explicit level labels, so that the value attributes of the reference teaching plan are clearer. Finally, the level label of the reference teaching plan is determined according to the levels of all the first teaching plan content, which can establish a clear association between the reference teaching plan and the teaching plan grading standard, facilitating accurate screening and calling of the reference teaching plan in subsequent steps. This process solves the problem of disordered reference teaching plan information and difficulty in accurately utilizing beneficial components in the traditional method. Through the coherent steps of prompt word guidance, effective content extraction, level judgment, and level label determination, the scattered reference resources become orderly and can be accurately matched, and the required reference content can be quickly called when generating a teaching plan, which not only improves the utilization efficiency of reference resources, but also avoids the decline of the quality of the teaching plan caused by the misuse of mismatched reference content.

[0014] In combination with some embodiments of the first aspect, in some embodiments: the step of calling the language large model to generate the preliminary teaching plan base point according to the preset first control prompt word and the teaching plan grading standard specifically includes: the first control prompt word includes a first built-in prompt word and a first user prompt word; according to the teaching plan grading standard, the language large model is called to identify the first instruction sentence in the first user prompt word related to the teaching plan grading standard, and the level requirement of the preliminary teaching plan base point to be generated is determined according to the first instruction sentence, the level requirement including the number of layers and the level of each layer; the language large model is guided to generate the preliminary teaching plan base point according to the level requirement and the first built-in prompt word.

[0015] By adopting the technical solution, the built-in prompt word can provide a standardized guiding framework, ensuring that the model generates behaviors in accordance with the basic logic of the teaching plan design and avoiding deviation of the generated content from the specification; the user prompt word can integrate the personalized needs of teachers, making the generated preliminary teaching plan base point more in line with the teaching style and actual scene of teachers. The model can accurately capture the hierarchical requirements of teachers for the preliminary teaching plan base point according to the first instruction sentence in the user prompt word according to the teaching plan grading standard, avoiding deviation of the generated content from the expected result due to instruction understanding bias. Then, the preliminary teaching plan base point is generated in combination with the hierarchical requirements and the built-in prompt word, which can meet the personalized needs of teachers and follow the standardized grading framework, making the generated content both personalized and professional. This process solves the problem of inaccurate prompt word guidance and difficulty in balancing standardization and individuality in the prior art, and through the synergistic effect of built-in prompt words ensuring standardization, user prompt words reflecting individuality, and the model accurately identifying instructions, the preliminary teaching plan base point can meet the teaching plan grading standard and the specific needs of teachers, reducing the workload of subsequent optimization and improving the efficiency and quality of the teaching plan generation.

[0016] In combination with some embodiments of the first aspect, in some embodiments, according to the preset second control prompt word, the final teaching plan base point and the first hierarchical label, the step of calling the language large model to comprehensively analyze the reference teaching plan to generate a preliminary teaching plan includes: guiding the language large model to analyze the final teaching plan base point through the second control prompt word to generate a second hierarchical label of the final teaching plan base point; determining the hierarchical matching degree of each reference teaching plan according to the first hierarchical label of the reference teaching plan and the second hierarchical label; screening the reference teaching plan according to a preset first screening strategy, the decision factor of the first screening strategy including the hierarchical matching degree; and guiding the language large model to generate the preliminary teaching plan based on the screened reference teaching plan according to the second control prompt word.

[0017] By adopting the technical solution, the second control prompt word guide model analyzes the final version of the teaching plan base point and generates its two-level labels, which can convert the teacher's optimized core needs into quantifiable and matchable labels, providing clear basis for subsequent reference teaching plan screening. The model determines the matching degree by comparing the first-level label and the second-level label, which can quickly identify the reference content that best matches the core needs, avoiding blind calling of reference resources. According to the preset first screening strategy, the reference teaching plan is screened, which can eliminate the teaching plans that do not match or have low matching degree, ensuring that the reference teaching plans used in the subsequent generation of the preliminary version of the teaching plan are high-quality and high-matching-degree materials. Finally, the preliminary version of the teaching plan is generated based on the screened reference teaching plan, which can make the model efficiently integrate high-quality reference resources under the guidance of standardized core needs, generating a teaching plan with complete structure and content that meets the needs. This process solves the problem of disordered reference resource screening and the disconnection between generated teaching plans and core needs in existing methods. Through the above steps, the preliminary version of the teaching plan can not only reflect the teacher's individualized optimization needs, but also fully absorb the essence of high-quality reference resources.

[0018] In some embodiments in combination with some embodiments of the first aspect, the step of guiding the language large model to generate the preliminary version of the teaching plan based on the screened reference teaching plan according to the second control prompt word specifically includes: extracting second teaching plan content of the reference teaching plan according to the hierarchical matching degree of the reference teaching plan; and generating the preliminary version of the teaching plan by synthesizing the second teaching plan content.

[0019] By adopting the technical solution, the second teaching plan content is extracted according to the hierarchical matching degree of the reference teaching plan, which can accurately locate the part of the reference teaching plan that best matches the needs of the final version of the teaching plan base point, avoiding the inclusion of irrelevant or low-matching-degree content in the reference teaching plan in the generation process. Then, the preliminary version of the teaching plan is generated by synthesizing these high-matching-degree second teaching plan content, which can make the language large model focus on key information when integrating reference resources, avoiding the generation of a logically chaotic or unemphatic teaching plan due to information overload. This process further improves the utilization efficiency of reference resources. Compared with directly integrating the complete reference teaching plan, extracting the core content and then integrating it can reduce the impact of redundant information on the quality of the teaching plan, making the generated preliminary version of the teaching plan more refined and more in line with the core needs. At the same time, extracting content based on hierarchical matching degree can ensure that the core information of different reference teaching plans remains consistent in integration, avoiding content conflicts or disconnections.

[0020] In some embodiments in combination with some embodiments of the first aspect, a preset teaching plan sorting strategy is determined; a display interface is set, and a reference teaching plan list is displayed on the display interface, wherein the corresponding referenceability information is displayed for each reference teaching plan in the reference teaching plan list, and the referenceability information at least includes the first-level label; the order of each reference teaching plan in the reference teaching plan list is calculated according to the teaching plan sorting strategy, and the reference teaching plans are arranged according to the order.

[0021] By adopting the technical solution, the efficiency and accuracy of the teacher in viewing the reference teaching plan when optimizing the initial teaching plan can be effectively improved. The preset teaching plan sorting strategy and the generated ordered reference teaching plan list can enable the teacher to quickly locate the teaching plan to be viewed without blind searching one by one. The reference information containing the first-level label in the list intuitively presents the key hierarchical attributes of each reference teaching plan, helping the teacher quickly judge the reference value of the current teaching plan optimization, saving the tediousness of reading each article, and then more efficiently selecting appropriate reference content to facilitate the smooth development of the teaching plan optimization work.

[0022] In a second aspect, the present application provides a teaching plan generation system, characterized in that the teaching plan generation system comprises one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is configured to store computer program code, the computer program code comprising computer instructions, and the one or more processors are configured to invoke the computer instructions to enable the teaching plan generation system to perform the method of any one of the first aspect.

[0023] In a third aspect, the present application provides a computer-readable storage medium comprising instructions that, when executed on a teaching plan generation system, cause the teaching plan generation system to perform the method of any one of the first aspect.

[0024] The embodiments of the present application provide one or more technical solutions. Based on the discussion of these embodiments, an intelligent agent for assisting teaching plan design can be implemented, which has at least the following technical effects or advantages: 1. The concept of teaching plan hierarchical standard is proposed. The teaching plan hierarchical standard clearly defines the basis for describing teaching requirements, determines the framework and basis for language large model to generate teaching plan base points, and can guide teachers to accurately and comprehensively excavate and control course requirements, and quickly complete the review and optimization of the teaching plan base points generated by the language large model. The teaching plan hierarchical standard can also enable the language large model to more accurately match the actual teaching scene when generating teaching plans in the future, and avoid deviations in generated content due to the lack of standard guidance.

[0025] 2. A two-stage teaching plan generation method is proposed, which takes the teaching plan grading standard as the core. Based on the teaching plan grading standard, the method combines reference teaching plan search and hierarchical label extraction, first generates a teaching plan base point describing the teaching requirements, then generates a complete teaching plan, and finally forms a closed-loop human-computer collaborative technical solution through manual optimization. The core is to define boundaries by grading standards, associate reference resources and teaching plan requirements by hierarchical labels, first generate a teaching plan base point in the first stage, then generate a complete teaching plan in the second stage, and the content generated in two stages is optimized by manual optimization to ensure accuracy. The teaching plan base point in the first stage has less content, and the teacher can spend less time to complete the optimization. The complete teaching plan in the second stage has a large amount of content, but it is generated based on the teaching plan base point, and most of the content has a high degree of compliance, so only part of the content needs to be optimized. Therefore, the method can effectively solve the problems of insufficient pertinence and professionalism of the existing large model teaching plan generation and the need for a large amount of time for optimization by teachers, and improve the teaching plan generation efficiency and reduce the cost of teacher preparation.

[0026] 3. Some implementation recommendations are made for the content of the teaching plan grading standard. The teaching plan grading standard can include two or more grading dimensions, and each grading dimension is defined by at least two hierarchical levels. In terms of specific content, the teaching plan grading standard can include a learning situation grading dimension and / or a teaching goal grading dimension. The teaching plan grading standard is defined in multiple dimensions, and is refined into hierarchical differences in different dimensions, so that the hierarchical characteristics of each dimension can be clearly reflected when grading the teaching plan, making it easier to accurately grade based on multiple dimensions and multiple levels, and more efficiently and reasonably filter and integrate reference teaching plans to provide more targeted basis for teaching plan optimization and other work.

[0027] 4. A method for referencing effective content of a reference teaching plan is proposed. A pre-set third control prompt word is used to guide the language large model to extract the first teaching plan content related to the teaching plan grading standard in the reference teaching plan, determine the hierarchical label of the reference teaching plan after judging the hierarchical level of the content corresponding to the grading standard, and the core is to accurately constrain the extraction range of the large model through the third control prompt word, convert the implicit content related to the teaching plan grading standard in the reference teaching plan into explicit hierarchical labels, and establish a clear association between the reference teaching plan and the teaching plan grading standard. In the subsequent generation of teaching plans, the teaching requirements described by the teaching plan base point and the hierarchical labels of the reference teaching plan can be used to quickly filter out reference teaching plans with high compatibility for the current course, and extract the part of the content that meets the requirements, thereby providing accurate materials for generating high-quality teaching plans, so that the large model can generate a preliminary teaching plan based on high-quality content. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 is a flowchart of the teaching plan generation method based on a language large model in the embodiments of the present application; Figure 2is a specific flowchart of extracting the first-level label of the reference teaching plan in the embodiment of the present application; Figure 3 is a schematic diagram of an entity device structure of a teaching plan generation system in the embodiment of the present application. DETAILED DESCRIPTION

[0029] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments and are not intended to be limiting to the present application. As used in the specification and the appended claims of the present application, the singular forms "a," "an," and "the" are intended to include both singular and plural forms, unless the context clearly indicates otherwise. It will be further understood that the terms "and / or" used in the present application, mean any or all possible combinations of one or more of the listed items.

[0030] Hereinafter, the terms "first" and "second" are only for the purpose of description, and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first" and "second" can explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, the meaning of "a plurality of" is two or more, unless otherwise specified.

[0031] Since the embodiments of the present application involve the application of classroom learning situation, in order to facilitate understanding, the related terms and concepts involved in the embodiments of the present application will be introduced first.

[0032] (1) Language large model: an artificial intelligence model based on deep learning, which has the ability of natural language understanding. Roughly, language large models can be divided into two categories: classification language large models and generation language large models. Classification language large models have the ability to understand and classify a piece of natural language, such as the BERT model. Generation language large models have the ability of natural language understanding and generation, which can realize intelligent content generation, intelligent dialogue, information retrieval and summary, and other natural language related functions, such as ChatGPT, DeepSeek, Qwen and other large models.

[0033] (2) Prompt word: structured or unstructured instructions, information or context provided by the user to guide the generation of the language large model to output the expected results when interacting with the generation language large model.

[0034] (3) Lesson Plan: A practical teaching document prepared by the teacher to ensure the smooth and effective conduct of teaching activities, with each lesson or topic as a unit, detailing the teaching content, steps, and methods. It is a teaching framework and implementation plan that the teacher systematically plans and carefully designs based on the course syllabus and textbook content, combined with the students' actual situation. It reflects the teacher's teaching philosophy, experience, and organizational skills.

[0035] (4) Learning Situation Analysis: Users comprehensively, objectively, and deeply identify, analyze, and sort out factors such as students' current knowledge base, skill level, cognitive patterns (e.g., thinking style, attention characteristics), learning motivation, learning habits, emotional attitudes, and individual differences (e.g., cognitive style, learning needs, life experience background). Its core purpose is to provide empirical evidence for the accurate calibration of teaching objectives, the reasonable selection of teaching content, the appropriate selection of teaching methods, the scientific design of teaching activities, and the personalized implementation of teaching evaluation, ultimately achieving "teaching based on learning" and promoting targeted development of all students based on their original level.

[0036] (5) Teaching Objectives: Based on the overall requirements of the national curriculum standards (or curriculum syllabus), combined with the characteristics of the specific teaching content, students' learning foundation (such as cognitive starting point and developmental needs), and teaching conditions, teachers clearly define the observable, measurable, and assessable expected learning outcomes that students should achieve in the dimensions of cognition, skills, and affective attitudes and values ​​after completing a specific class period, unit, or course. It is both the "starting point" for selecting teaching content, designing teaching methods, and organizing teaching activities, and the "end point" for controlling the teaching process, evaluating learning outcomes, and inspecting teaching quality, running through the entire teaching process.

[0037] The subject of this application can be a terminal device that runs lesson plan design software with a lesson plan generation system, such as a mobile phone, tablet computer, desktop computer, smart TV, etc., without limitation.

[0038] The teaching plan generation method based on a large language model proposed in this application does not limit the specific technical means of UI (user interface), and can be any implementation method. Common UI implementation methods include mobile terminal APP, web page, client software, etc., which are not limited here.

[0039] The following describes the process of the lesson plan generation method based on a large language model in this embodiment. Please refer to [link / reference]. Figure 1 ,like Figure 1 As shown, Figure 1 A flowchart illustrating the language-based large-scale model-based lesson plan generation method provided in this application.

[0040] S101. Pre-defined lesson plan grading standards, which include at least two levels of definition; The course plan grading standard defines a basic method and guideline for analyzing the individualized characteristics or teaching requirements of a course. The course plan grading standard is the basis for designing a course plan. Under the guidance of the course plan grading standard, a teacher deeply understands the individualized characteristics and requirements of the course, determines the level of the course plan to be designed corresponding to the course plan grading standard, and thus designs a targeted course plan. The course plan grading standard includes which aspects of content, the present application does not limit, and users (usually school research groups) can determine by their own research. Generally, for different subjects or school stages, users need to develop different course plan grading standards. Guiding the design of a course plan based on the course plan grading standard reduces the theoretical knowledge reserve requirements of individual teachers, thereby reducing the difficulty of designing a course plan for teachers and avoiding the design of a course plan without course targeting.

[0041] In an embodiment of the present application, the course plan grading standard includes two or more grading dimensions, and each grading dimension includes at least two level definitions. The course plan grading standard includes multiple dimensions, and each dimension is defined. Preferably, the dimensions are as orthogonal as possible (i.e., the aspects concerned by two different dimensions do not intersect or intersect very little). The dimensions of the course plan grading standard can be hierarchically defined, first divided into several large dimensions, and the large dimensions are further subdivided into several sub-dimensions, which can be further subdivided. Subdividing the course plan grading standard into multiple dimensions can make the definition of the course plan grading standard clearer and more complete, facilitate the understandability and applicability of the course plan grading standard, and reduce the difficulty for teachers to master.

[0042] In some embodiments, in order to accurately match the course plan to the actual situation of the students and avoid the generated course plan from deviating from the student's learning situation or the teaching goal being ambiguous, to improve the targeting of the course plan, the course plan grading standard in step S101 can include a learning situation grading dimension and / or a teaching goal grading dimension.

[0043] In an embodiment, the course plan grading standard includes a learning situation grading dimension, which describes the grading content related to the learning situation of students, and focuses on in-depth analysis of the learning situation of students to guide the design of a course plan that meets the current learning situation of students. Further, the learning situation grading dimension can be subdivided into multiple grading sub-dimensions. For example, it can be subdivided into three sub-dimensions: cognitive basis, learning ability and habit, and learning interest and value dimension, and several levels are defined for each grading sub-dimension. This embodiment only exemplarily describes possible subdivided sub-dimensions, possible levels, and possible level definitions of the learning situation grading dimension, and does not make specific limitations thereon.

[0044] Specifically, the cognitive basis refers to the "knowledge reserve" (such as concepts, formulas, theorems), "cognitive ability" (such as memory, understanding, application, analysis, evaluation, and creation of thinking levels), and "knowledge structure integrity" (the degree of correlation between knowledge) that students already have and are directly related to the current learning content. Based on the definition of this sub-dimension, it can be divided into four levels: preparatory level, basic level, development level, and proficiency level. Specifically, the four levels are defined as follows: 1. Preparatory level: completely unaware of the core concepts and basic terminology of the current learning content, only able to complete low-level memory tasks (such as reciting scattered knowledge points), unable to understand the essence of knowledge, no relevant pre-knowledge reserve, and knowledge system is broken; 2. Basic level: can recall the definition of core concepts and basic terminology, but understanding is not deep, can complete basic understanding tasks (such as explaining the surface meaning of knowledge points), but cannot be flexibly applied, has a small amount of pre-knowledge, and can establish simple knowledge associations; 3. Development level: can deeply understand the essence of core concepts, distinguish between confusing concepts, can complete application and analysis tasks (such as using knowledge points to solve typical problems and disassemble problem logic), has solid pre-knowledge, and can establish associations between multiple knowledge points; 4. Proficiency level: can deeply master the knowledge system, understand the logic and rules behind the knowledge, can complete evaluation and creation tasks (such as judging the pros and cons of solution, designing new problem-solving approaches), has a complete knowledge system, and can establish knowledge associations across modules and domains.

[0045] Specifically, student ability and habit refers to the "methods and skills" (such as preview, review, notes, reflection, cooperative learning, etc.) and "stable learning behavior patterns" that students actively adopt during the learning process, with the core being "whether they can actively regulate the learning process". Based on the definition of this sub-dimension, it can be divided into four levels: passive adaptation type, preliminary development type, systematic application type, and autonomous optimization type. The definitions of the four levels are as follows: 1. Passive adaptation type: no active learning strategies, completely dependent on teacher instructions, no preview, review, only completes the basic tasks arranged by the teacher, no note-taking or reflection habits, and no self-regulation in the learning process; 2. Preliminary development type: has simple strategy awareness, occasionally uses basic learning methods; can perform shallow preview (such as browsing content), basic review (such as reciting key points), but not in-depth; has preliminary note-taking and reflection habits, which are relatively simple; 3. Systematic application type: can actively use multiple learning strategies and form stable habits; can perform in-depth preview (marking questions, relating to old knowledge), systematic review (combing knowledge framework); note-taking is structured, reflection is in-depth, and can regulate the learning process; 4. Autonomous optimization level: can adjust strategies and optimize methods according to learning results; can choose appropriate strategies based on learning content characteristics (such as different disciplines and different tasks); can independently innovate strategies and evaluate the effectiveness of strategies.

[0046] Specifically, learning interest and value refers to the "intensity of interest", "type of learning motivation" (such as love for the subject itself, score / reward, etc.), "learning self-confidence" (such as whether to dare to try, not afraid of mistakes) and "subject value identification" (such as whether to think that the subject is useful and meaningful) of the students to the learned subject. Based on the definition of this sub-dimension, it can be divided into four levels: 1. Negative resistance type: completely uninterested in learning content, avoiding learning; the learning motivation is basically "avoiding criticism / punishment", and there is no active learning intention; no self-confidence, afraid to try, thinking "I can't learn"; 2. Passive acceptance type: general interest in learning content, not rejecting but not actively; the learning motivation is basically "complete task / obtain basic recognition" (such as "just hand in homework"); have preliminary self-confidence, only participate in learning under encouragement; 3. Active participation type: interested in learning content, actively pay attention to relevant information; the learning motivation is basically "curiosity / driven by the desire to know" (such as "want to know why"); self-confidence, actively participate in learning, not afraid of mistakes; 4. Active identification type: love learning content, regard learning as a pleasure; the learning motivation is basically "internal value identification" (such as "think this part of knowledge is useful / significant"); have self-confidence and tenacity, can actively overcome learning difficulties.

[0047] It should be noted that in the specific implementation, the names of the four levels of the above three learning situation classification sub-dimensions can be uniformly taken as: first level, second level, third level, fourth level, or the level names are named in two ways, for example, the names of the four levels of the "cognitive foundation" sub-dimension are: preparatory level (first level), basic level (second level), development level (third level), and proficiency level (fourth level). The advantage of this is that the naming "preparatory level, basic level, development level, proficiency level" ensures readability and understandability, and the naming "first level, second level, third level, fourth level" can simplify the writing of prompts, for example, the user only needs to input in the prompt: "Please generate an initial teaching plan base point of a student learning situation with all the classification sub-dimensions at the fourth level", which can generate an initial teaching plan base point of a student learning situation at the fourth level, without the need to specify the required level for each learning situation classification sub-dimension in the prompt.

[0048] The above embodiment defines the teaching plan grading standard according to the student learning situation, which has three beneficial effects: 1. Reducing the threshold of learning situation analysis, especially helping new teachers or cross-disciplinary teachers to avoid the problem of “missing dimension and misjudgment”, without needing to start from zero, teachers can systematically master the true level of students in key dimensions such as cognitive basis, learning strategy, and emotional attitude; 2. Providing a practical teaching basis, the specific description of different levels (such as “cognitive basis level: can list simple equations but cannot solve them”), which allows teachers to quickly locate the students in their class, and then accurately design layered goals, adapt teaching methods, and arrange gradient assignments, avoiding “one-size-fits-all”; 3. Improving the teaching pertinence, clear dimension and level division can help teachers identify the weaknesses of students (such as “most students are in the passive adaptation level of learning strategy”), so as to focus on weaknesses in teaching plan design and classroom guidance, truly realize “teaching according to learning”, make teaching more in line with student needs, and improve classroom efficiency and learning effect.

[0049] In another embodiment, the teaching plan grading standard includes a teaching goal grading dimension, which describes the grading content of the teaching goal, focusing on in-depth analysis of the teaching goal to guide users to design a clear teaching plan for the teaching goal. Further, the teaching goal grading dimension can be subdivided into multiple grading sub-dimensions for definition. For example, the teaching goal grading dimension can be subdivided into: knowledge and skills, process and method, emotional attitude and value, and several levels can be defined for each grading sub-dimension. This embodiment only exemplarily describes possible teaching goal subdivision dimensions, possible levels and level definitions, without making specific limitations.

[0050] Specifically, knowledge and skills are basic cognitive content and key operational abilities that students need to master through learning, which are the premise of forming process and method, developing emotional attitude and value, and are the core support for students to adapt to society and lifelong learning. The knowledge and skills dimension focuses on knowledge structuring and skill automation. Knowledge and skills can be divided into three levels: basic mastery, application integration, and innovation migration, which are defined as follows: 1. Basic mastery: can memorize and reproduce basic knowledge / skills, i.e., can accurately recite, identify basic concepts, formulas, rules, etc. of the subject, and complete standardized operations in imitation practice; 2. Application integration: can flexibly use knowledge / skills in new situations, i.e., can transfer knowledge to unfamiliar situations to solve problems, integrate multiple knowledge points to complete complex tasks, and skill operation reaches proficiency; 3. Innovation migration: can generate new knowledge / skill systems, i.e., can creatively reorganize knowledge, design new methods or solutions, and solve open-ended problems, with skill application reflecting individualization and efficiency.

[0051] Specifically, the process and method are the cognitive process and practical process experienced by students in the learning of science, as well as the learning strategies and thinking methods mastered by students. They are the key link connecting "knowledge and skills" and "emotional attitudes and values", and are the core approach to cultivating students' "self-directed learning, cooperative inquiry, and innovative thinking". Process and method focus on the internalization of learning strategies and the advancement of thinking methods. The grading reflects the process characteristics of "from passive acceptance to active creation". Process and method can be divided into three levels: imitation experience, independent application, and innovative exploration. The three levels are defined as follows: 1. Imitation experience: can experience the learning process and initially master the method, that is, can complete the learning process under the guidance of the teacher, imitate the use of observation, experiment, data analysis and other methods, record the process but lack reflection; 2. Independent application: can independently select and optimize methods, that is, can independently design learning paths, flexibly select methods to solve problems, adjust strategies through reflection, and form individualized learning patterns; 3. Innovative exploration: can create new methods or research paradigms, that is, can integrate interdisciplinary methods, propose new research ideas or technical routes, verify and improve methods through practice, and form systematic achievements.

[0052] Specifically, emotional attitudes and values are the inner experience, value judgment, and behavior tendency formed by students in the learning of science. They are the embodiment of the fundamental task of "moral education" in subject education, and are related to students' "sound personality, social responsibility, and ideal and belief". They are the "soul dimension" of the three-dimensional goal. Emotional attitudes and values focus on the deepening of emotional experience and the behaviorization of values. Emotional attitudes and values are divided into three levels: participation and identification, internalization and practice, and character shaping. The three levels are defined as follows: 1. Participation and identification: passively accept and initially identify values, that is, can participate in relevant activities, express basic understanding of values, and occasionally show initiative. 2. Internalization and practice: can actively integrate values into behavior, can consciously use value criteria to guide action, and can stick to correct behavior in complex situations, forming stable attitude tendency. 3. Character shaping: values are internalized into individual character, that is, can influence others through value judgment, innovate value realization methods in practice, and form unique beliefs and social responsibility.

[0053] It should be noted that in specific implementation, similar to the grading dimension of learning situation, the three levels of each of the above teaching goal grading dimensions can be uniformly named as: first level, second level, and third level. The advantage of this is that users can simplify the preparation of prompt words. Both naming methods can also be used.

[0054] The above embodiments define lesson plan grading standards based on teaching objectives. The beneficial effects are as follows: From a teaching planning perspective, it makes objectives clearer and more systematic, avoiding vagueness and generality. It helps teachers accurately grasp the teaching focus in different dimensions (such as knowledge and skills, processes and methods, etc.), and the grading aligns with students' cognitive patterns, allowing for differentiated design based on learning levels and catering to the needs of students at different levels. From the perspective of teaching implementation and evaluation, clear objectives provide direction for designing teaching activities and offer concrete evidence for evaluating teaching effectiveness, reducing the randomness of teaching. It also provides teachers with reference materials, saving preparation time, improving the efficiency and quality of lesson plan writing, and contributing to the implementation of the teaching objectives of "teaching according to aptitude" and "cultivating morality and character."

[0055] In another embodiment, the lesson plan grading standard includes a student learning level grading dimension and a teaching objective grading dimension, guiding teachers to design lesson plans by comprehensively considering student learning and teaching objectives, thus creating more targeted lesson plans. It should be noted that the student learning level grading dimension and the teaching objective grading dimension can be considered as two major dimensions of the lesson plan grading standard, and these can be further subdivided into several sub-dimensions. Of course, these sub-dimensions can be further subdivided.

[0056] The lesson plan grading standards proposed in the above embodiments are exemplary and can be used as design guidelines to further refine or simplify them in terms of dimensions and levels, designing lesson plan grading standards that better suit their own needs. Specific details of the lesson plan grading standards are not limited in this application. In addition to the student learning situation and teaching objectives mentioned in the embodiments of this application, the lesson plan grading standards can also include other aspects, such as teacher-student interaction, to guide lesson plans to have better teacher-student interaction design. It is worth noting that schools can incorporate content related to specific aspects of the teaching process that they want to guide teachers to strengthen into the lesson plan grading standards, thereby providing targeted guidance for lesson plan design and improving classroom teaching effectiveness.

[0057] S102. After obtaining the course information, search for a preset number of lesson plans as reference lesson plans based on the course information. The course information refers to the course information corresponding to the lesson plans to be designed, including but not limited to grade level, subject, chapter, theme, and textbook version. The purpose of searching for lesson plans based on the course information is to find lesson plans highly relevant to this course as reference plans. This search can be for lesson plans that fully match the course information or only partially match it. This application does not limit the degree of matching between the searched lesson plans and the course information. Furthermore, more features can be added to narrow the search scope, such as strong teacher-student interaction or a large amount of homework content. This application does not limit the use of additional search features.

[0058] The source of the reference lesson plan can be various. In an exemplary embodiment, the lesson plan is searched from a user resource library, which is a private resource pool built in the user application system and stores the lesson plan resources uploaded by users and institutions in the past. In another embodiment, the lesson plan is searched from a public website, preferably a website with high education professionalism or high document quality, such as the National Elementary and Secondary Wisdom Education Platform, the National Education Resource Public Service Platform, the Subject Network, the Teaching Network, the Internet Public Resource Website, the Baidu Library, the DouDing Network, the industry forum, etc. The reference lesson plan can also be searched from the user resource library and the public website. It should be noted that the search range of the reference lesson plan is not limited in the present application.

[0059] The number of lesson plans to be searched can be preset by the user, such as 10. If the preset value is large, the search time and the subsequent processing time will be longer. However, the number of reference lesson plans is large, and the quality of the generated lesson plan can also be high. This is a trade-off for the user, and the present application does not limit the number of searched lesson plans.

[0060] S103, extracting a first-level label of the reference lesson plan according to the lesson plan grading standard; The first-level label of the reference lesson plan reflects the level of the reference lesson plan corresponding to the lesson plan grading standard. If the lesson plan grading standard contains multiple grading dimensions, and if the grading dimensions are further divided into multiple grading sub-dimensions, the first-level label can contain the levels of each grading sub-dimension, or only contain a total level obtained by combining the levels of all grading sub-dimensions, or contain both the levels of each grading sub-dimension and the total level. After extracting the first-level label of the reference lesson plan, the reference lesson plan can be referenced according to the first-level label in the subsequent process.

[0061] For example, if the lesson plan grading standard adopts the learning situation grading dimension described in S101, the first-level label extracted from a reference lesson plan can be: "cognitive basis: level three, learning ability and habit: level four, learning interest and value: level three, and total: between level three and level four". The meaning is that the level characteristics of the reference lesson plan are "cognitive basis" corresponding to level three, "learning ability and habit" corresponding to level four, "learning interest and value" corresponding to level three, and the total level is between level three and level four.

[0062] It is to be explained that the first-level label extracted from the reference lesson plan can contain the hierarchy of all hierarchical sub-dimensions, or only contain the hierarchy of part of the hierarchical sub-dimensions, for example, the above example can not have the dimension of "learning interest and value", and some hierarchical sub-dimensions are not extracted because of the design flaws of the lesson plan, or these dimensions are not concerned for the lesson plan. Further, the first-level label extracted from the reference lesson plan can be cross-level, and still taking the above example, the first-level label extracted can be: "cognitive basis: level two~level three, learning ability and habit: level three~level four, learning interest and value: level three~level four", which means that the level of the dimension of cognitive basis is between level two~level three, the level of the dimension of learning ability and habit is between level three~level four, and the level of the dimension of learning interest and value is between level three~level four. The specific form of the first-level label is not limited by the present application. Whether to include the overall level and the method of summarizing the overall level according to the level of each hierarchical sub-dimension are also not limited by the present application.

[0063] The lesson plan searched in the user resource library can have a structured attribute description, and through these attribute descriptions, the first-level label can be obtained, especially, the lesson plan designed by the method of the present application includes the level label in its structured attribute description, which can be directly extracted.

[0064] In an embodiment, the first-level label of the reference lesson plan can be extracted by using a language large model, please refer to Figure 2 , Figure 2 The specific flowchart of extracting the first-level label of the reference lesson plan in the embodiment method of the present application is as follows: S1031, presetting a third control prompt word; The third control prompt word can be a prompt word for guiding the language large model to interpret the content related to the lesson plan grading standard, and the specific content and setting method are not limited here.

[0065] S1032, controlling the language large model to extract the first lesson content of the reference lesson plan according to the third control prompt word, the first lesson content being the content related to the lesson plan grading standard; Specifically, the content of the reference lesson plan and the third control prompt word are input into the language large model together, and the third control prompt word guides the language large model to perform text analysis on the reference lesson plan. For the content in the text, the language large model will judge whether this part of the content is related to the lesson plan grading standard, if it is related, then this part of the content is identified as the first lesson content. If the lesson plan grading standard includes multiple hierarchical dimensions, then the first lesson content is the content related to multiple hierarchical dimensions. If the lesson plan grading standard includes the level definition under the hierarchical dimension, then the first lesson content is the content related to the level definition under the hierarchical dimension.

[0066] S1033, determining the level corresponding to the teaching plan grading standard of the first teaching plan content; If the first teaching plan content is identified, the model will continue to determine the specific level of this part of content according to the level definition of the teaching plan grading standard (if the teaching plan grading standard contains multiple dimensions, according to the level definition of the teaching plan grading dimension). According to the description and definition of the teaching plan grading standard, the third control prompt word guides the language model to determine whether the content of the reference teaching plan is the first teaching plan content according to the teaching plan sentence by sentence. If it is, it is determined that it corresponds to the level of the teaching plan grading standard. The level determined can strictly correspond to one level of the teaching plan grading standard, or can be across several levels of the teaching plan grading standard.

[0067] In the reference teaching plan, it may be possible to extract multiple first teaching plan contents, or it may not be possible to extract the first teaching plan content.

[0068] The level of this step is determined by which level of the teaching plan grading standard the first teaching plan content can cover. If the first teaching plan content can only reflect the overall characteristics of a certain grading dimension and cannot be disassembled to a specific sub-dimension, only the overall level of the grading dimension is determined. If the first teaching plan content can explicitly reflect the characteristics of multiple grading sub-dimensions under a certain grading dimension, the level of each grading sub-dimension is determined, and then the overall level of the grading dimension can be further summarized.

[0069] S1034, determining the first level label of the reference teaching plan according to the level of the first teaching plan content.

[0070] If no first teaching plan content is extracted, the first level label is empty. If the first teaching plan content is extracted, the first level label of the reference teaching plan can be generated by summarizing the level of all the first teaching plan content using the generative language model, or a traditional algorithm can be designed to determine the first level label of the reference teaching plan. The specific method is not limited in the present application.

[0071] It should be noted that there are other ways to extract the first level label of the reference teaching plan. In another embodiment, instead of using a generative language model, a special classification language model is trained to identify the content related to the teaching plan grading standard in the reference teaching plan and determine its level. In another embodiment, it is completed in combination with the classification language model and the generative language model. There are other feasible ways to extract the first level label of the reference teaching plan. The specific technical implementation means are not limited in the present application. With the "first level label", the core value of each reference teaching plan can be accurately identified, which can facilitate subsequent accurate matching based on the specific requirements set by the teacher, and improve the accuracy and efficiency of the reference teaching plan.

[0072] S104, according to the preset first control prompt word and the teaching plan grading standard, calling a language large model to generate a preliminary version of the teaching plan base point; The preset first control prompt word is used to guide the language large model to generate the preliminary version of the teaching plan base point according to the teaching plan grading standard. The teaching plan base point is the content generated according to the teaching plan grading standard, which is related to the teaching requirements of the course. The preliminary version of the teaching plan base point refers to the teaching plan base point generated by the language large model, which is still in the initial version and needs to be optimized by the teacher before it becomes the final version of the teaching plan base point.

[0073] The preliminary version of the teaching plan base point is generated according to the teaching plan grading standard. If the teaching plan grading standard includes the content of the learning situation grading dimension, the preliminary version of the teaching plan base point contains the design content of the learning situation analysis. If the teaching plan grading standard includes the content of the teaching goal grading dimension, the preliminary version of the teaching plan base point contains the design content of the teaching goal. If the teaching plan grading standard includes both the learning situation grading and the teaching goal grading, the preliminary version of the teaching plan base point contains the design content of the learning situation analysis and the teaching goal. If the teaching plan grading standard contains multiple grading dimensions, the preliminary version of the teaching plan base point contains the design content of all or part of the grading dimensions. Corresponding to the definition of multiple levels in the teaching plan grading standard, the preliminary version of the teaching plan base point can contain the design of one or more levels, or it can be a cross-level design. For new teachers, they may not be able to accurately grasp the teaching requirements of the course, so the language large model generates multiple level designs and provides multiple options. With the comparison of options, the teacher can select the level design that best fits the course and optimize it. The teacher does not need to build from scratch, but can directly select the appropriate version for optimization, reducing the difficulty of designing the teaching plan.

[0074] For example, if the teaching plan grading standard uses the teaching goal grading dimension described in step S101, the following is a preliminary version of the teaching plan base point generated by the language large model based on this teaching plan grading standard. The initial teaching plan base point contains the design of all levels of the three sub-dimensions of the teaching goal grading dimension, and organizes the design content according to the levels.

[0075] Level 3: Knowledge and skills: guide students to comprehensively use the concept of quadratic equation and the properties of equations to solve practical problems across contexts such as geometric area, growth rate, etc., and achieve knowledge transfer; provide light-weight assistance for a small number of students with weak foundation.

[0076] Process and method: through independent inquiry and group innovation modeling display, let students master the thinking method of "practical problem-equation modeling-solution verification-scheme optimization", and improve critical thinking.

[0077] Emotional attitude and values: stimulate students' deep interest in mathematical modeling, cultivate rigorous attitude in achievement display, appreciate the value of mathematical solution to real-world problems, and meet the high-level requirements of academic quality.

[0078] Level 2: Knowledge and Skills: Ensure students accurately grasp the definition, general form, and meaning of solutions of quadratic equations, and be able to skillfully formulate equations for simple practical problems. Focus on explaining common and easily mistaken points such as "coefficients of terms" and "verification of solutions to equations".

[0079] Process and Methods: Through the analysis of typical cases and step-by-step modeling practice, students are guided to form a learning path of "understanding concepts - establishing equations - reflecting and verifying", which enhances their ability to apply knowledge to new situations. Teachers provide regular guidance to weaker groups.

[0080] Emotional attitudes and values: By achieving the goals of basic modeling tasks, students will build confidence in learning, cultivate rigorous problem-solving habits, appreciate the logic of mathematical knowledge, and enhance their willingness to learn proactively.

[0081] Level 1: Knowledge and Skills: Help students accurately memorize the definition and general form (ax+bx+c=0, a≠0) of a quadratic equation, be able to collectively identify equation types, imitate and list simple equations, and understand the connection with linear equations in one variable.

[0082] Process and Methods: Through animated demonstrations, step-by-step demonstrations, and peer support in small groups, students will master basic modeling methods, complete the learning process step by step using task sheets, and cultivate the habit of "imitation-practice-consolidation".

[0083] Emotional attitudes and values: By combining real-life examples such as calculating classroom area and pricing goods, students can experience the connection between mathematics and life, gain a sense of accomplishment by achieving basic tasks, and overcome their fear of difficulty. Based on the initial lesson plan framework with the above three levels, teachers can compare and select the most suitable one to optimize and generate the final lesson plan framework.

[0084] In the initial lesson plan baseline example generated in this application, the presentation format of its hierarchical sub-dimensions is flexible and configurable. Users can choose the presentation method of explicit or implicit sub-dimensions according to their teaching experience and usage needs, as detailed below: 1. Explicit sub-dimension presentation: that is, the presentation form currently adopted in the example, before the content of the teaching goal design at each level (three levels, two levels, one level), the hierarchical sub-dimension names such as “knowledge and skills”, “process and method”, “emotion attitude and value” are clearly marked, and the content of each part is clearly defined. The teaching goal hierarchical dimension. Under this form, the dimension information is intuitive and visible, and the user does not need to deduce or judge additionally. For users such as new teachers, cross-disciplinary teachers, or those who are not familiar with the hierarchical standards of teaching goals, they can choose the explicit sub-dimension presentation form by themselves. By clearly marking the hierarchical sub-dimension names, it can help such users quickly locate the content of each part, avoid understanding deviation or confusion of optimization direction caused by unclear dimension definition, and reduce the use threshold of the initial teaching plan base.

[0085] 2. Implicit sub-dimension presentation: refers to deleting the hierarchical sub-dimension names such as “knowledge and skills”, “process and method”, “emotion attitude and value” before the content of the teaching goal design at each level in the example, and only keeping the specific design content corresponding to each sub-dimension. At this time, the hierarchical sub-dimension information is not directly marked by words, but is implied in the specific content - according to the preset hierarchical standards of teaching goals, each part of the content still corresponds to a clear hierarchical sub-dimension, and only the hierarchical sub-dimension name is not reflected in the presentation form. For teachers with rich teaching experience, because they have formed a clear understanding of the dimension definition of the hierarchical standards of teaching goals, they can quickly and accurately determine the dimension attribution implied by the specific content, and at this time they can choose the implicit sub-dimension presentation form. This form can simplify the presentation content of the initial teaching plan base, reduce redundant information interference, enable teachers to focus more on the rationality and adaptability of the specific design content at each level, improve the optimization efficiency of the initial teaching plan base, and meet the usage habits and professional needs of such users.

[0086] It should be noted that the present application supports users to switch the sub-dimension presentation form according to their own needs, and the specific configuration method is not limited. Whether it is the intuitive presentation of explicit sub-dimension or the concise presentation of implicit sub-dimension, it does not affect the corresponding relationship between the initial teaching plan base and the preset hierarchical standards of teaching goals - under the two forms, the content of the initial teaching plan base is strictly generated according to the dimension requirements of the hierarchical standards of teaching goals, and only the information presentation level is different.

[0087] The first control prompt word includes the prompt word required to complete the task of the present step, which can include but is not limited to the hierarchical standards of teaching plans, system built-in prompt words, user input prompt words, classroom information, reference teaching plans, course standards, teaching materials, etc. It can also include other aspects of content. The specific content and form of the first control prompt word are not limited in the present application.

[0088] The first control prompt word guides the language large model to generate the initial version of the teaching plan base point, which can have various implementation manners. For example, in one embodiment, the first control prompt word includes system built-in prompt words, teaching plan grading standards, and classroom information, and the language large model generates the initial version of the teaching plan base point according to the system built-in prompt words. The system built-in prompt words require the language large model to have a design for each level of the teaching plan grading standards in the generated initial version of the teaching plan base point.

[0089] In another embodiment, the first control prompt word includes system built-in prompt words, teaching plan grading standards, and classroom information, and further includes user input prompt words, which further refine the initial version of the teaching plan base point.

[0090] In another embodiment, the first control prompt word includes system built-in prompt words, teaching plan grading standards, and searched reference teaching plans, and the language large model generates the initial version of the teaching plan base point by summarizing the corresponding content of the reference teaching plans.

[0091] In another embodiment, the first control prompt word further includes teacher input prompt words, which supplement the personalized needs of the teacher for the initial version of the teaching plan base point, while ensuring that the generated content strictly meets the preset teaching plan grading standards. The first control prompt word guides the language large model to work in the following manner: S1041, the preset first control prompt word includes a first built-in prompt word and a first user prompt word; The first built-in prompt word is a system built-in prompt word, which can be pre-written by a human or dynamically generated during system operation. The first user prompt word is a teacher input prompt word.

[0092] S1042, according to the teaching plan grading standards, the language large model is called to identify a first instruction sentence in the first user prompt word related to the teaching plan grading standards, and the level requirement of the initial version of the teaching plan base point to be generated is determined according to the first instruction sentence, the level requirement including the number of layers and the level of each layer; The sentence related to the teaching plan grading standards in the first user prompt word includes the level requirement of the initial version of the teaching plan base point to be generated, which is referred to as the first instruction sentence hereinafter. For example, if the teaching plan grading standards include a student learning situation grading dimension, the first instruction sentence is a sentence related to the student learning situation. Further, if the teaching plan grading standards include multiple grading sub-dimensions, the first instruction sentence is a sentence related to the teaching plan grading sub-dimensions. There can be multiple first instruction sentences identified from the first user prompt word, or there can be no valid first instruction sentence identified.

[0093] The hierarchical requirements of all the first instruction sentences are summarized to determine the hierarchical requirements of the user on the generated preliminary lesson plan base point. The hierarchical requirements include the number of layers and the level of each layer. If the lesson plan grading standard contains multiple grading dimensions and sub-dimensions, the identified hierarchical requirements contain the number of layers and the level of each layer for each grading dimension and sub-dimension. It should be noted that the first instruction sentence can require the generation of designs across multiple levels, and can also require not to generate the design of a certain grading dimension, that is, the preliminary lesson plan base point does not necessarily contain the design of all grading dimensions.

[0094] For example, if the lesson plan grading standard adopts the learning situation grading dimensions described in step S101, the hierarchical requirements identified from the first user prompt can be: "Cognitive Foundation: Two levels (three levels, four levels), Learning Ability and Habit: Three levels ~ Four levels, Learning Interest and Values: NO". The meaning is that the "Cognitive Foundation" dimension requires the generation of two levels of design, which are three levels and four levels of this dimension, respectively, the "Learning Ability and Habit" dimension requires the generation of a design between three levels and four levels, and the "Learning Interest and Values" dimension does not require the generation of design.

[0095] It should be noted that there can be no indication of hierarchical requirements for some grading dimensions in the first user prompt, and there can be no indication of not needing to generate. For this case, there can be multiple processing methods, such as prompting the user to remind whether the first user prompt needs to be rewritten, or directly generating the design of all levels of the grading dimension, or directly not generating the design of the grading dimension.

[0096] It should be noted that in addition to the above-mentioned first user prompt without hierarchical requirements for some grading dimensions, the teacher can not input the first user prompt, or there can be no valid instructions in the first user prompt. For this case, preferably, all levels of design can be generated by default. For new teachers, this situation can be because the teacher does not know how to write the correct first user prompt, or the teacher can not be clear enough about the requirements of the course. At this time, the teacher does not need to input any user prompt, and the system provides all levels of design, gives multiple options, and the teacher selects the most matched one for modification, which reduces the difficulty of use for the teacher.

[0097] For example, in one embodiment, the first built-in prompt contains the following simple sentence guide to guide the generation of the language model to implement the function of this step: "According to the lesson plan grading standard, first identify the sentences in the first user prompt related to the lesson plan grading standard, and then determine the hierarchical requirements of the generated preliminary lesson plan base point, which includes the number of layers and the level of each layer corresponding to each grading dimension".

[0098] In another embodiment, the function of this step is implemented in a programmed manner, and the basic process is as follows: first, use a language large model to analyze whether the sentence of the first user prompt word is a first instruction sentence sentence by sentence. Then, determine the level requirement of each first instruction sentence. Finally, aggregate the level requirements of all first instruction sentences to determine the level requirement of the preliminary course outline base point to be generated. In this embodiment, the language large model used can be a generative language large model, a specially trained classification language large model, or a combination of the two. Aggregating the level requirements of all first instruction sentences can be implemented by a language large model or a traditional rule algorithm. The specific technical means is not limited in the present application.

[0099] S1043, guiding the language large model to generate a preliminary course outline base point according to the level requirement and the first built-in prompt word.

[0100] If only one level requirement is extracted from the first user prompt word, the preliminary course outline base point contains a one-level design; if multiple level requirements are extracted, the preliminary course outline base point contains a multi-level design. If no level requirement is extracted (may be that the user does not input the level requirement, or may be that the user inputs incorrectly), there can be multiple processing methods, which can not generate the preliminary course outline base point, give the user a prompt information, or output all level designs, or take other processing methods, which are not limited in the present application.

[0101] There are various specific ways to generate the preliminary course outline base point. In one embodiment, the language large model directly generates according to the description and definition of the course outline grading standard; in another embodiment, first, according to the first level label of the reference course outline, filter out the reference course outline that meets the level requirement, and then use the first built-in prompt word to guide the language large model to synthesize the content related to the course outline base point in the selected reference course outline to generate the preliminary course outline base point.

[0102] According to the above embodiment, the corresponding first control prompt word can be easily written. The first control prompt word can be fixed by the system, or dynamically generated according to the preset rules, or even dynamically generated by the language large model. The writing and generation of prompt words are mature technologies, and will not be repeated here. It should be noted that the steps in the above embodiments do not necessarily be completed by calling the language large model once. The entire task can be completed by calling the language large model multiple times. The first control prompt word can be composed of multiple independent sub-prompt words, which are used in the corresponding steps of the language large model call.

[0103] In addition to the above-mentioned content in the first control prompt word, other content such as curriculum standards, teaching materials, output format requirements, etc. can also be included, and the present application is not limited. In addition, more detailed control statements can be added in the first built-in prompt word to enable the language large model to make more detailed analysis; in the first user prompt word, the first user prompt word teacher can also supplement personalized requirements, which can enable the model to capture personalized needs beyond the grading standards, so that the initial draft of the teaching plan not only meets the general standards, but also incorporates the teacher's teaching style and special needs, guiding the language large model to generate an initial draft of the teaching plan that better meets their own needs. The content and form of the specific first control prompt word, first built-in prompt word and first user prompt word are not limited by the present application.

[0104] S105, receiving a first optimization instruction for the initial draft of the teaching plan, optimizing the initial draft of the teaching plan according to the optimization instruction, and generating a final draft of the teaching plan; The initial draft of the teaching plan is generated by the language large model and may not fully meet the user's expectations, and the user may need to modify and optimize it to obtain the final draft of the teaching plan. If the initial draft of the teaching plan contains multiple levels of design, the user first selects a level closest to the requirements, and then modifies it based on this. The initial draft of the teaching plan is displayed in an editor and can be modified completely by manual editing or assisted by the language large model. The terminal device responds to the user's operation on the initial draft of the teaching plan and generates a corresponding first optimization instruction according to the user's operation and sends it to the teaching plan generation system, for example, expanding, abbreviating, adjusting the tone, etc. using the language large model, or selecting a paragraph of text and inputting detailed requirements according to the first optimization instruction to let the language large model regenerate. Manual editing and question-and-answer assisted editing using the language large model for a file are mature technologies, and the present application is not limited.

[0105] S106, according to the preset second control prompt word, the final draft of the teaching plan and the first level label, calling the language large model to comprehensively analyze the reference teaching plan and generate an initial draft of the teaching plan; The final draft of the teaching plan determines the teaching requirements and teaching objectives of the course teaching plan. Based on the final draft of the teaching plan, the second control prompt word is preset, which guides the language large model to generate a complete teaching plan according to the requirements of the final draft of the teaching plan. The teaching plan generated in this step is the initial draft of the teaching plan, which is the initial version of the teaching plan, and usually needs to be modified and optimized by the user before a usable final version of the teaching plan can be obtained.

[0106] In one embodiment, the second control prompt word contains the following sentence: "According to the requirements and goals set by the final version of the teaching plan base point, generate a complete teaching plan, and the format of the teaching plan must meet the requirements of the teaching plan template." This prompt word sentence can guide the language large model to generate the preliminary version of the teaching plan according to the final version of the teaching plan base point. Among them, the teaching plan template defines the paradigm of the teaching plan content, which mainly includes the definition of the dimensions of the teaching plan, such as: curriculum standard analysis, student situation analysis, teaching goals, teaching difficulties, teaching strategies and methods, teaching process, teaching evaluation methods, classroom blackboard writing, homework, etc. The specific teaching plan template is not limited in the present application.

[0107] In another embodiment, the second control prompt word contains the following sentence: "According to the requirements and goals set by the final version of the teaching plan base point, refer to the content of the reference teaching plan, and generate a complete teaching plan, and the format of the teaching plan must meet the requirements of the teaching plan template." In this embodiment, the language large model generates the preliminary version of the teaching plan by referring to the content of the reference teaching plan, which is more practical than the preliminary version of the teaching plan generated entirely by the language large model in the previous embodiment.

[0108] In another embodiment, the second control prompt word realizes programmed and more detailed guidance, and the basic process is as follows: S1061, guiding the language large model to analyze the final version of the teaching plan base point by the second control prompt word to generate the second level label of the final version of the teaching plan base point; The second level label is the same in concept as the first level label of the reference teaching plan described above, which determines the requirements and goals of the teaching plan to be designed.

[0109] S1062, determining the hierarchical matching degree of each reference teaching plan according to the first level label and the second level label; Among them, the hierarchical matching degree is defined as the matching degree of the requirements or goals of the reference teaching plan and the teaching plan to be designed. The hierarchical matching degree can be a score, or it can be rated as "high, medium, low". If the teaching plan grading standard contains multiple grading dimensions, the hierarchical matching degree can be a hierarchical matching degree array, and each element of the array corresponds to the matching degree of a grading dimension; the hierarchical matching degree can also include the matching degree array corresponding to each dimension and the comprehensive matching degree calculated by comprehensively matching the matching degrees of each dimension. According to the first level label of the reference teaching plan and the second level label of the final version of the teaching plan base point, a hierarchical matching degree calculation method can be easily designed, which is not limited in the present application.

[0110] S1063, filtering the reference teaching plan according to the first filtering strategy, and the decision factors of the first filtering strategy include the hierarchical matching degree; The first screening strategy can be to not filter any reference lesson plans or to filter out reference lesson plans with low hierarchical matching degrees according to the hierarchical matching degrees. Generating a lesson plan based on a reference lesson plan with a high matching degree can improve the generation speed of the language large model and can avoid referencing the content of a reference lesson plan with a low matching degree, thereby ensuring the quality of the initial version of the lesson plan. The first screening strategy is designed based on the hierarchical matching degree as the main decision factor. For example, the first screening strategy can be to select the top 10 lesson plans with high hierarchical matching degrees. Further, the first screening strategy can also include other decision factors, such as the completeness of the dimensions of the lesson plan, the content richness of each dimension of the lesson plan, the logical rigor of the content of the lesson plan, the user attributes of designing the lesson plan, the school attributes of designing the lesson plan, and the like. The specific selection of additional factors is not limited in the present application. Selecting appropriate decision factors can easily design the first screening strategy, and the specific first screening strategy is not limited in the present application.

[0111] S1064, guiding the language large model to generate the initial version of the lesson plan based on the screened reference lesson plans according to the second control prompt word.

[0112] Based on the selected reference lesson plan, the second control prompt word can guide the language large model to refer to the content of the selected reference lesson plan to generate the initial version of the lesson plan.

[0113] In another embodiment, the step S1064 adopts a more detailed guidance manner as follows: 1. First, guide the language large model to extract the second lesson plan content in the reference lesson plan according to the hierarchical matching degree of the reference lesson plan. The second lesson plan content is the content in the reference lesson plan that matches the hierarchical level of the final version of the lesson plan (i.e., the requirement or target matching), which can be completely matched content or partially matched content (when the hierarchical label spans multiple levels). For each piece of second lesson plan content, a matching score can also be generated. If the lesson plan grading standard includes multiple grading dimensions, the second control prompt word can further be refined to guide the language large model to extract the second lesson plan content of the reference lesson plan according to the hierarchical matching degree of each grading dimension.

[0114] 2. Then, generate the initial version of the lesson plan by synthesizing the second lesson plan content. This step can guide the language large model to synthesize the extracted second lesson plan content to generate the initial version of the lesson plan, or only synthesize the second lesson plan content with a high matching score according to the matching score. Extracting the part of the content with a high matching score from the reference lesson plan and generating the initial version of the lesson plan based on it can ensure the content quality of the initial version of the lesson plan and reduce the input of the teacher for modification.

[0115] According to the above embodiments, the corresponding second control prompt word can be easily written. The second control prompt word can be fixed by the system, dynamically generated according to a preset rule, or even dynamically generated by the large model. The writing and generation of the prompt word are mature technologies, and will not be described herein.

[0116] It is to be explained that each step in the above embodiment does not necessarily complete by one call of the language large model, the whole task can be completed by multiple calls of the language large model, the second control prompt word can be composed of multiple independent sub-prompt words, each sub-prompt word is used in the corresponding step of calling the language large model. In the second control prompt word, in addition to the content mentioned in the above embodiment, other content such as course-related background information, teaching plan template, etc. can also be included, and there can also be a supplementary prompt word input by the teacher. The complete content of the second control prompt word is not limited in the present application.

[0117] In summary, the method of the present application generates a preliminary teaching plan based on the teaching requirements determined by the teaching plan base point and the high-quality content of the teaching plan, fully combines the efficient generation capability of the large model and the precise target constraint of the teaching plan base point, so that most of the content of the preliminary teaching plan already meets the needs of the teacher, reduces the workload of subsequent optimization of the teacher, and ensures the completeness and accuracy of the content of the teaching plan. The course requirements are met.

[0118] S107, receiving a second optimization instruction for the preliminary teaching plan, optimizing the preliminary teaching plan to generate a final teaching plan.

[0119] The preliminary teaching plan is generated by the language large model, and the content may not completely meet the user's expectations, and the user may need to modify and optimize to obtain the final teaching plan. The preliminary teaching plan is presented in an editor, and the teacher can modify it completely by manual editing, or modify it in combination with the question and answer with the language large model. Similar to the first optimization instruction, the terminal device receives the user's operation on the preliminary teaching plan, and then determines the second optimization instruction and sends it to the system. For example, using the language large model to expand, abbreviate, adjust the tone, etc. You can also select a piece of text, determine and input detailed requirements according to the second optimization instruction, and let the language large model regenerate. Manually editing a file and using a language large model to assist in editing are mature technologies, and the present application does not limit them.

[0120] In one embodiment, in order to facilitate the user to view and refer to the reference teaching plan during modification, a teaching plan sorting strategy can be preset, the order of each reference teaching plan in the display list of the display interface is calculated according to the teaching plan sorting strategy, a list is set to display the reference teaching plan, and the reference teaching plan is arranged and displayed according to the order. Usually, what is displayed in the list is the link of the reference teaching plan, not the text.

[0121] The referenceability information of each reference teaching plan is displayed, which is used for the teacher to conveniently select a reference teaching plan of interest for viewing. The referenceability information at least includes the first-level label, and the first-level label directly reflects the level of the reference teaching plan corresponding to the teaching plan grading standard, so that the user can accurately locate the reference teaching plan with high level matching degree, avoid selecting a wrong teaching plan due to ambiguous information and wasting time, and also can guarantee the accuracy of the teaching plan modification direction. When the teacher modifies the preliminary teaching plan by referring to the reference teaching plan, the first-level label in the list can be used to quickly distinguish the core adaptation attributes of different reference teaching plans, so that the reference process is smoother, and the teaching plan modification efficiency is indirectly improved.

[0122] The decision factor of the teaching plan sorting strategy can have multiple selection modes. In one embodiment, the reference degree of the reference teaching plan is selected as the decision factor of the teaching plan sorting strategy, which can be the proportion of the content of the reference teaching plan in the preliminary teaching plan, or the proportion of the content of the reference teaching plan being referenced into the preliminary teaching plan, or a combination of the two proportions. The decision factor of the teaching plan sorting strategy can also include the level matching degree of the teaching plan, and the decision factor of the teaching plan sorting strategy can also include other factors, such as some factors related to the quality of the teaching plan, such as the dimension completeness of the teaching plan, the content richness of each dimension, the logical rigor of the content description, the design time of the teaching plan, the attributes of the user who designs the teaching plan, the attributes of the school where the teaching plan is designed, and the like. The specific selection and specific implementation algorithm of the decision factor of the teaching plan sorting strategy are not limited in the present application. In one embodiment, multiple teaching plan sorting strategies are preset, and the user selects different strategy options to change the arrangement order of the reference teaching plans.

[0123] The referenceability information of each reference teaching plan is displayed in a preset position. The referenceability information reflects the description of the referenceability of the reference teaching plan, and in addition to the first-level label, can usually include information related to the decision factor of the teaching plan sorting strategy (such as the description of the decision factors related to the reference degree, the level matching degree, the design time, the design user, and the design school), which is beneficial to the user to select a reference teaching plan of interest for viewing and improve the work efficiency. The specific form of the referenceability information is not limited in the present application.

[0124] The teaching plan generation system in the embodiment of the present application will be described from the perspective of hardware processing. Please refer to Figure 3 , which is a schematic diagram of an entity device structure of the teaching plan generation system in the embodiment of the present application.

[0125] It should be noted that Figure 3 The structure of the teaching plan generation system shown is only one example, and should not bring any limitation to the function and use range of the embodiment of the present application.

[0126] As Figure 3As shown, the lesson plan generation system includes a central processing unit (CPU) 301 which can perform various appropriate actions and processes in accordance with a program stored in a read-only memory (ROM) 302 or a program loaded from a storage section 308 into a random access memory (RAM) 303, such as performing the methods described in the above embodiments. In the RAM 303, various programs and data required for operation of the system are also stored. The CPU 301, the ROM 302, and the RAM 303 are connected to each other through a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0127] Connected to the I / O interface 305 are an input section 306 including an audio input device, a push button switch, and the like; an output section 307 including a liquid crystal display (LCD), an audio output device, an indicator, and the like; a storage section 308 including a hard disk and the like; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, and the like. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as necessary. A removable media 311 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, and the like is mounted on the drive 310 as necessary, so that a computer program read therefrom is installed in the storage section 308 as necessary.

[0128] In particular, the processes described above with reference to the flowcharts can be implemented as a computer software program in accordance with embodiments of the present application. For example, embodiments of the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing a computer program for performing the methods illustrated by the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section 309, and / or installed from the removable media 311. When the computer program is executed by the central processing unit (CPU) 301, various functions defined in the present application are performed.

[0129] Note that specific examples of computer-readable storage media can include but are not limited to an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present disclosure, computer-readable storage media can be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

[0130] The flow diagrams and the block diagrams in the drawings are illustrations of possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present disclosure. It will be understood that each block of the flow diagrams and the block diagrams, and combinations of blocks in the flow diagrams and the block diagrams, can be implemented by a computer program instruction or code. Such instructions can be generated by a computer program product that causes a processor to perform functions described in the embodiments of the present disclosure. Also, it will be understood that each block of the flow diagrams and the block diagrams, and combinations of blocks in the flow diagrams and the block diagrams, can be implemented by special purpose hardware-based computer systems which perform the specified functions or operations.

[0131] Specifically, the lesson plan generation system of the embodiment includes a processor and a memory, and the memory stores a computer program. When the computer program is executed by the processor, the language large model-based lesson plan generation method provided in the above embodiment is implemented.

[0132] As another aspect, the present disclosure also provides a computer-readable storage medium. The storage medium can be included in the lesson plan generation system described in the above embodiments, or can exist independently without being assembled into the lesson plan generation system. The storage medium carries one or more computer programs. When the one or more computer programs are executed by a processor of the lesson plan generation system, the lesson plan generation system implements the language large model-based lesson plan generation method provided in the above embodiments.

[0133] The above embodiments are only used to illustrate the technical solutions of the present disclosure, but not limit the present disclosure; even though the present disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present disclosure.

[0134] In the above embodiments, the term "when" can be interpreted to mean "if" or "after" or "in response to determining" or "in response to detecting" depending on the context. Similarly, the phrase "on determining" or "if detecting (a stated condition or event)" can be interpreted to mean "if determining" or "in response to determining" or "on detecting (a stated condition or event)" or "in response to detecting (a stated condition or event)" depending on the context.

[0135] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by a computer program instructing the relevant hardware to complete, and the program can be stored in a computer readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments. The aforementioned storage medium includes ROM or random storage memory RAM, magnetic disk or optical disk and various storage program codes.

Claims

1. A method for generating lesson plans based on a large language model, characterized in that, The method includes: A pre-defined lesson plan grading standard is provided, which includes at least two levels of definition. After obtaining the course information, a preset number of lesson plans are searched for reference based on the course information; Based on the lesson plan grading criteria, extract the first-level tags of the reference lesson plans; Based on the preset first control prompt word and the lesson plan grading standard, the language big model is invoked to generate the initial lesson plan base points; Receive a first optimization instruction for the initial lesson plan base points, optimize the initial lesson plan base points according to the optimization instruction, and generate the final lesson plan base points; Based on the preset second control prompt word, the final lesson plan base point and the first level tag, the language big model is invoked to comprehensively analyze the reference lesson plan and generate the initial lesson plan; Receive a second optimization instruction for the initial version of the lesson plan, optimize the initial version of the lesson plan, and generate the final version of the lesson plan.

2. The method according to claim 1, characterized in that, The lesson plan grading standard includes two or more grading dimensions, and each grading dimension includes at least two level definitions.

3. The method according to claims 1-2, characterized in that, The lesson plan grading standard shall include at least the student learning grading dimension and / or the teaching objective grading dimension, and the initial lesson plan base shall include at least the student learning analysis design and / or the teaching objective design.

4. The method according to claim 1, characterized in that, The step of extracting the first-level tag of the reference lesson plan according to the lesson plan grading standard specifically includes: Preset third control prompt words; The first lesson plan content of the reference lesson plan is extracted based on the third control prompt word control language big model. The first lesson plan content is related to the lesson plan grading standard. Determine the level of the lesson plan content corresponding to the lesson plan grading standard. Based on the hierarchy of the first lesson plan content, determine the first-level label of the reference lesson plan.

5. The method according to claim 1, characterized in that: The step of generating the initial lesson plan baseline by calling the language big model based on the preset first control prompt word and the lesson plan grading standard specifically includes: The first control prompt includes a first built-in prompt and a first user prompt; According to the lesson plan grading standard, the language model is invoked to identify the first instruction statement related to the lesson plan grading standard in the first user prompt word, and the hierarchical requirements of the initial version of the lesson plan to be generated are determined according to the first instruction statement. The hierarchical requirements include the number of layers and the level of each layer. Based on the hierarchical requirements and the first built-in prompt word, the language model is guided to generate the initial lesson plan basis.

6. The method according to claim 1, characterized in that, The step of generating a preliminary lesson plan by calling the language big model to comprehensively analyze the reference lesson plan based on the preset second control prompt words, the final lesson plan base points, and the first level tags specifically includes: The second control prompt word guides the language model to analyze the core points of the final lesson plan and generate second-level tags for the core points of the final lesson plan. Based on the first level label and the second level label, determine the level matching degree of each reference lesson plan; The reference teaching materials are selected according to a preset first screening strategy, wherein the decision factors of the first screening strategy include the hierarchical matching degree. Guided by the second control prompt, the language model generates the initial version of the lesson plan based on the selected reference lesson plan.

7. The method according to claim 6, characterized in that, The step of guiding the language model to generate the initial lesson plan based on the selected reference lesson plan according to the second control prompt word specifically includes: Based on the hierarchical matching degree of the reference lesson plan, extract the content of the second lesson plan from the reference lesson plan; The initial version of the lesson plan was generated by combining the content of the second lesson plan.

8. The method according to claim 1, characterized in that, The method further includes: Determine the pre-set lesson plan sequencing strategy; Set up a display interface to display a list of reference lesson plans. The list of reference lesson plans displays corresponding reference information for each reference lesson plan. The reference information includes at least the first-level label. The order of each reference lesson plan in the reference lesson plan list is calculated according to the lesson plan sorting strategy, and the reference lesson plans are arranged in the order stated.

9. A lesson plan generation system, characterized in that, The lesson plan generation system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors call the computer instructions to cause the lesson plan generation system to perform the method as described in any one of claims 1-8.

10. A computer-readable storage medium comprising instructions, characterized in that, When the instruction is executed on the lesson plan generation system, the lesson plan generation system performs the method as described in any one of claims 1-8.

Images