Intelligent assisted teacher lesson preparation system and method based on education big model

Abstract

The present invention discloses an intelligent auxiliary teacher lesson preparation system and method based on an education macro model, comprising the following steps: S1. generating a standardized teacher lesson preparation data set; S2. constructing a teaching knowledge graph; S3. generating preliminary intelligent lesson preparation content; S4. presenting the preliminary generated intelligent lesson preparation content and providing an interactive interface combining natural language and speech recognition; S5. generating adjusted intelligent lesson preparation content; S6. adaptively learning and optimizing the intelligent lesson preparation content based on the teacher's historical lesson preparation data and the interactive feedback data, adjusting the weights of key knowledge points in the teaching knowledge graph, and updating the content generation strategy of the education macro model to generate a final intelligent lesson preparation plan. The present invention significantly improves the responsiveness of intelligently generated content to the teacher's teaching intentions, supports dynamic coordination of teaching content and test question style with the teaching rhythm and classroom feedback, and provides an intelligent decision-making support foundation for large-scale personalized teaching.

Description

Technical Field

[0001] The present invention relates to the field of educational technology, and in particular to an intelligent-assisted teacher lesson preparation system and method based on an education macro model. Background Art

[0002] With the rapid development of artificial intelligence and educational information technology, intelligent test question generation has gradually become an important direction of teaching assistance in the field of basic education. In primary and secondary schools, it has shown broad application potential in reducing the burden of teacher preparation, improving the consistency of question quality, and enhancing personalized teaching support. At present, mainstream primary and secondary school test question generation methods mostly rely on manual writing or question bank calls based on fixed templates. Teachers usually compile questions according to the teaching syllabus, textbook content and teaching experience. Although this traditional model has certain teaching pertinence, it faces obvious bottlenecks when dealing with large-scale, cross-disciplinary and personalized test question generation needs.

[0003] In the existing technology, some systems have introduced natural language processing technology to try to achieve automatic question generation. However, most of them are limited to the extraction and template matching of textbook keywords, and lack an in-depth understanding of the teaching context, course objectives and knowledge point structure. The generated questions are often simple in content and simple in structure, and cannot meet the high standards of teaching evaluation for logical rigor, semantic accuracy and knowledge coverage. At the same time, the existing technology generally lacks an effective teacher interaction feedback mechanism, and cannot dynamically feed back the teacher's modification opinions on the content, structure and difficulty of the questions in actual teaching to the question generation model, resulting in the inability to continuously optimize the generation quality.

[0004] In terms of teaching content assistance, although some platforms have the ability to intelligently push courseware or teaching resources, their underlying logic is mostly based on statistical content recommendation. The lack of support for structured teaching knowledge graphs makes it difficult to establish a semantic closed loop between teaching objectives, knowledge points, and topics, resulting in a disconnect between content push and classroom objectives. Furthermore, the expression of teachers' lesson preparation intentions in the current system mostly relies on static parameter selection, and lacks the ability to semantically understand natural language or voice input intentions, which limits the efficiency and intelligence of interaction between teachers and intelligent systems.

[0005] Therefore, there is an urgent need to propose an intelligent generation method for primary and secondary school test questions that can integrate deep semantic understanding, structured teaching knowledge modeling, and a human-computer feedback closed-loop mechanism to effectively make up for the shortcomings of existing technologies. Summary of the Invention

[0006] One purpose of the present invention is to propose an intelligent assisted teacher preparation system and method based on an educational macro model. The present invention significantly improves the responsiveness of intelligently generated content to teachers' teaching intentions, supports dynamic coordination of teaching content and test question style with teaching rhythm and classroom feedback, and provides an intelligent decision-making support foundation for large-scale personalized teaching.

[0007] According to an embodiment of the present invention, an intelligent-assisted teacher lesson preparation method based on an education macro model includes the following steps:

[0008] S1. Collect and integrate teacher preparation data from multiple sources, pre-process the teacher preparation data, and generate a standardized teacher preparation dataset;

[0009] S2. Use semantic parsing of standardized teacher lesson preparation datasets to construct a teaching knowledge graph;

[0010] S3. Perform semantic understanding and contextual analysis on the lesson preparation demand data input by teachers based on the educational macro model, match the lesson preparation demand data with the teaching knowledge graph, and generate preliminary intelligent lesson preparation content;

[0011] S4. Present the initially generated intelligent lesson preparation content and provide an interactive interface that combines natural language and voice recognition, allowing teachers to preview, edit, and provide feedback on the intelligent lesson preparation content in real time, generating interactive feedback data;

[0012] S5. Analyze the interactive feedback data, extract the content modifications, key adjustments and personalized needs involved in the feedback, and dynamically optimize the initially generated intelligent lesson preparation content based on the feedback information to generate adjusted intelligent lesson preparation content;

[0013] S6. Based on the teacher's historical lesson preparation data and the interactive feedback data, the intelligent lesson preparation content is adaptively learned and optimized, the weights of key knowledge points in the teaching knowledge graph are adjusted, and the content generation strategy of the education model is updated to generate the final intelligent lesson preparation plan.

[0014] Optionally, the S1 includes the following steps:

[0015] S11. Obtain the original education data set D raw The original education data set includes the primary and secondary school textbook content set T m , syllabus content collection T s , Collection of past exam questions T e And the collection of primary and secondary school learning question bank content T q ;

[0016] S12. For the original education data set D raw Perform text cleaning to remove and standardize the garbled information, redundant symbols, non-teaching content, and format errors contained in the original education data set, and obtain the cleaned education text set D clean ,Among them, the cleaning process adopts a unified standard rule to ensure the ,consistency of the expression structure of the content from each source;

[0017] S13. Cleaned educational text set D clean Perform format standardization processing, standardize the paragraph structure, question type identification, knowledge point description and teaching terms, and obtain the standardized education data set D norm , wherein the format standardization process maintains the consistent expression form of all samples in terms of semantic structure and information level;

[0018] S14. Educational data set D based on standardized format norm Mark key knowledge points and associate each educational data sample with its corresponding teaching knowledge point set K = {k1, k2, ..., k n} to match and label a knowledge point in the dataset to form the labeled education data set D tag , where k i Represents the i-th knowledge point, and the annotation results are embedded into each sample data in a structured label manner;

[0019] S15. For the labeled educational data set D tag Re-index and group according to the subject, grade, course content and knowledge point category, organize each sample data into a structured form, and construct a structured education dataset D struct , where each piece of structured education data includes: education text sample x i , corresponding knowledge point label k i and its classification information i ,Classification information includes subject attributes, grade dimensions and question type categories.

[0020] Optionally, the S2 includes the following steps:

[0021] S21. Based on structured education dataset D struct For each educational text sample x i Divide the teaching semantic units and construct the teaching semantic block set B i ={b i1 ,b i2 ,...,b im}, the teaching semantic block represents the smallest information unit with consistent teaching logic, which is used to capture the semantic dependency structure of teaching objectives and knowledge points;

[0022] S22. Teaching semantic block set B i Each teaching semantic block b ij Apply the context embedding representation function to generate the semantic embedding vector E ij , and map the semantic embedding vector to the semantic compression space S sem , retaining semantic factors related to test question generation;

[0023] S23. Introducing the semantic activation function A(k t ,E ij )=cosine(φ k (k t ),E ij ) Calculate the teaching semantic block b ij For knowledge point k t The semantic support strength of k (k t ) represents the embedding representation of the knowledge point, which is used to guide the correlation between the knowledge point and the context semantics;

[0024] S24. Construct the teaching knowledge graph G based on the set of all knowledge point labels K, the set of teaching objectives G, and the set of teaching semantic blocks B. edu =(V,E,W), where the node set V includes the knowledge point node v k , teaching target node v g and teaching semantic block node v b , the edge set E includes semantic support edges, target dependency edges and chapter structure edges, and the weight set W is composed of the semantic activation function value and the context embedding representation function;

[0025] S25. Based on teaching knowledge graph G edu For a given teaching target node v g and knowledge point node v k Execute context-aware path reasoning algorithm to calculate the test question generation teaching path P inf :

[0026]

[0027] Among them, P inf is the teaching path that is ultimately used to generate test questions. For all teaching target nodes v g To knowledge point node v k Redundancy(p) is the path redundancy penalty function used to avoid repeated references to the same semantic block, λ div To control the balance coefficient between diversity rewards and structural overlap, the obtained teaching path is used to guide the educational macro-model to generate goal-oriented test questions;

[0028] S26. Set the teaching path P inf Encoded as a semantic path vector π, it serves as the input semantic control factor of the educational model generation module, guiding the model to generate a set of test questions with rigorous logic, consistent semantics, and clear knowledge points.

[0029] S27. Generate a set of test questions Perform knowledge point call analysis to extract the set of knowledge points called in actual generation and teaching objectives set And compare it with the target set K, G in the path reasoning phase to calculate the semantic coverage deviation:

[0030]

[0031] Among them, Δ sem is the semantic coverage deviation of the semantic path of the current generated test question to the teaching objective, and β is the weight factor of the semantic coverage deviation;

[0032] S28. Based on the feedback results of semantic coverage deviation, adjust the edge weight set W in the teaching knowledge graph, increase the edge weights of semantic nodes that appear in the path but are not called, optimize the edge strength of the misactivated path, and form a closed loop of semantic graph optimization driven by test question generation:

[0033]

[0034] Among them, W ′ (u,v) is the edge weight set after feedback adjustment, η is the structure update learning rate, Indicates whether the edge participates in the current round of reasoning path.

[0035] Optionally, S3 includes the following steps:

[0036] S31. Receive the teacher's input lesson preparation demand data set D plan , the teacher's input lesson preparation demand data set D plan All text information in the text is semantically encoded and processed, and the teacher's lesson preparation vector representation Ψ is generated through the context understanding module in the education model. plan ,The teacher’s lesson preparation vector represents the information that integrates the teaching goal intention, course context, learning group characteristics and teaching method requirements;

[0037] S32. Calling the teaching knowledge graph G edu The teacher's lesson plan vector is represented by Ψ plan Match each of the teaching semantic path vector sets π in the teaching knowledge graph one by one, calculate their semantic relevance, and obtain the semantic relevance score between the teacher's lesson preparation vector representation and each semantic path;

[0038] S33. Based on the calculated semantic relevance score, filter out the vector representation of teacher lesson preparation Ψ plan The path set that meets the preset semantic relevance threshold is used as the semantic support path set π for generating intelligent lesson preparation content. * ;

[0039] S34. Semantic support path set Π obtained based on screening * And the teacher's lesson plan vector representation Ψ plan , the preliminary intelligent lesson preparation content set C is generated by the education model plan ={c1,c2,...,c k The initial intelligent lesson preparation content set includes the draft syllabus, the first draft of the courseware, classroom handouts, teaching cases, and personalized exercise suggestions. Each content is generated by the semantic generation function of the education model under the dual constraints of path semantics and teacher intention.

[0040] S35. Preliminary intelligent lesson preparation content set C plan Each content in the teaching consistency test and content structure integrity assessment is carried out, and the content adaptation score vector Ω={ω1,ω2,...,ω k}, the rating vector is used to represent the structural adaptation of each content relative to the teacher's input intention and the alignment with the teaching goal;

[0041] S36. The intelligent lesson preparation content with scores lower than the preset tolerance threshold τ in the fitness score vector Ω is returned to the education model for regeneration and correction, and the revision record in the form of structured tags is retained in the regenerated version, and finally a preliminary intelligent lesson preparation content set is generated. The preliminary intelligent lesson preparation content set meets the teaching objectives, contextual logic and individual characteristics requirements.

[0042] Optionally, the lesson preparation requirement data set includes a teaching objective description, a course theme keyword set, teaching stage information, teaching method preferences, and a student portrait parameter set.

[0043] Optionally, the S4 includes the following steps:

[0044] S41. Gather the preliminary intelligent lesson preparation content Input into the intelligent interactive presentation module for structured display. The intelligent interactive presentation module presents the teaching outline draft, courseware draft, classroom handouts, teaching cases and personalized exercise suggestions in a variety of ways, using document view, slide view and interactive component view respectively according to the content type.

[0045] S42. Based on the structured display interface, the natural language interaction submodule and the speech recognition interaction submodule are integrated into the interaction interface. The natural language interaction submodule receives the text instruction set Γ input by the teacher. text , the speech recognition interaction submodule receives the teacher speech input set Γ voice , where the text instruction set and the speech input set are uniformly mapped to the teaching feedback semantic space;

[0046] S43. Input natural language into set Γ text and the speech input set Γ voice The common encoding is the teacher interaction feedback representation set Γ feedback ,The teacher interaction feedback representation set is used to express the teacher’s ,editing and modification intentions, reconstruction suggestions, ,content preferences or teaching style adjustment requirements for the ,preliminary intelligent lesson preparation content;

[0047] S44. Representation of the teacher interaction feedback set Γ feedback With intelligent lesson preparation content collection Match each content one by one and identify the content index set I corresponding to the feedback target match , and extract the original semantic path vector set Π of the corresponding content segment match ;

[0048] S45. Perform semantic alignment on each teacher interaction feedback representation and its mapped content segment to extract the offset features and adjustment direction of the teacher's intention in the original path semantics, forming a feedback offset vector set Δ edit ;

[0049] S46. Generate interactive feedback dataset Each feedback data record includes the original semantic path vector π l , feedback offset vector δ l And the feedback target corresponding content index i l .

[0050] Optionally, the S5 includes the following steps:

[0051] S51. Receive and parse the interactive feedback dataset F interact , extracting the content modification information, teaching focus adjustment requirements and personalized teaching preferences involved in the feedback;

[0052] S52. Offset vector δ based on teacher feedback l The semantic features of the corresponding original semantic path vector π l Perform semantic reconstruction to generate the adjusted path semantic vector π′ l ,The semantic reconstruction process keeps the original structural logic of the path unchanged, and only performs minimal ,disturbance optimization on the node content, edge weight distribution, or knowledge point organization;

[0053] S53. Set all reconstructed path semantic vectors Π′={π′1,π′2,...,π′ s} is input as a semantic driving factor into the content regeneration module of the education model, linking the original teacher preparation vector representation Ψ plan Jointly generate an adjusted set of intelligent lesson preparation content

[0054] S54. In the interactive feedback dataset F interact Based on the teacher feedback offset vector delta l With the original semantic path vector pi l As input, the intelligent lesson preparation content collection Conduct semantic-guided reconstruction to generate a collection of intelligent lesson preparation content after fine-tuning the content The adjusted intelligent lesson preparation content collection retains the original semantic path logical structure and integrates the content modifications, teaching focus adjustments, and personalized needs expressed in teacher feedback. Specifically, it includes the draft teaching outline, preliminary courseware, class handouts, teaching cases, and personalized exercise suggestions.

[0055] S55. Final output of the adjusted intelligent lesson preparation content set The adjusted intelligent lesson preparation content collection is used to support teachers' teaching implementation in real time and dynamically support personalized teaching design.

[0056] Optionally, the draft teaching syllabus: Based on the original semantic path vector π l For the covered teaching objective nodes, if the teacher feedback offset vector contains information on teaching objective decomposition and knowledge point reordering, the priority of the teaching objective nodes in the path will be reordered, and the order of items and the level of objective expression in the draft teaching outline will be updated accordingly;

[0057] The first draft of the courseware: Based on the original semantic path vector π l If the teacher's feedback offset vector contains slide layout preferences, diagram suggestions, or content simplification requirements, the weights of nodes related to visual expression in the path are adjusted, and the courseware page structure and information distribution are reconstructed.

[0058] The lecture notes: Based on the original semantic path vector π l For the knowledge point coverage section in the teacher feedback, if the offset vector contains the need to change the example questions, expand the lecture structure, or adjust the image-text ratio, the corresponding knowledge point segment is partially updated and the exemplary content node in the path is replaced;

[0059] The teaching case: Based on the original semantic path vector π l If the teacher feedback offset vector contains the teaching context reconstruction, background replacement or style correction requirements, the dialogue framework and context setting of the path segment are regenerated, and the starting and ending points of the path remain unchanged.

[0060] The personalized exercise suggestion is based on the original semantic path vector π lThe associated knowledge point nodes and their graph edge weight information. If the teacher's feedback offset vector contains ability adaptation, difficulty control or test point reinforcement requirements, the node weights and edge structure in the path segment will be fine-tuned to reconstruct the matching degree of question type and difficulty while maintaining consistency with the original path logic.

[0061] An intelligent-assisted teacher preparation system based on an educational macro model is applied to an intelligent-assisted teacher preparation method based on an educational macro model, including the following modules:

[0062] The data processing module is used to collect primary and secondary school textbooks, teaching syllabuses, past examination questions and exercise banks, and to perform text cleaning, format standardization and key knowledge point annotation on the collected education data to generate structured education data sets;

[0063] The educational model building module is used to pre-train and fine-tune structured educational datasets to build educational models for primary and secondary school scenarios;

[0064] The teaching knowledge graph construction module is used to construct the teaching knowledge graph based on the teaching semantic blocks, teaching objectives and knowledge point relationships, forming a semantic path reasoning mechanism;

[0065] The intelligent lesson preparation content generation module is used to receive the lesson preparation demand data input by teachers, combine the education big model with the teaching knowledge map, and generate preliminary intelligent lesson preparation content such as the teaching outline draft, courseware draft, classroom handouts, teaching cases and personalized exercise suggestions;

[0066] The interactive feedback processing module is used to collect teachers' edits, style preferences, and teaching adjustment requirements for intelligent lesson preparation content through the natural language and speech recognition interactive interface, and generate semantic offset representation and interactive feedback datasets;

[0067] The dynamic optimization module is used to reconstruct the original semantic path and update the graph path based on teacher feedback, generate dynamically adjusted intelligent lesson preparation content, and drive the subsequent test question generation process.

[0068] The beneficial effects of the present invention are:

[0069] (1) The present invention constructs a multi-level graph structure including teaching target nodes, knowledge point nodes and teaching semantic block nodes, introduces semantic activation function and path redundancy penalty function, uses path inference algorithm to obtain the optimal teaching path vector, and dynamically adjusts the graph edge weight based on the semantic coverage deviation between the generated test questions and the target knowledge points after the test questions are generated, thereby realizing a generation-driven self-learning update mechanism of the graph structure, breaking through the limitations of the static use of traditional knowledge graphs, making the semantic path generated by the test questions highly consistent with the teaching structure, and significantly improving the path generation accuracy and semantic coverage.

[0070] (2) The present invention uniformly maps the teacher's natural language input and voice instructions into interactive feedback representations, generates feedback offset vectors through semantic alignment analysis, and acts on the original path semantic vectors for minimum disturbance reconstruction, thereby achieving structured modification of multiple types of content such as teaching outlines, handouts, cases, and test question suggestions. On this basis, feedback data synchronously updates the teaching knowledge graph and generation model, forming a closed-loop content optimization mechanism, significantly improving the responsiveness of intelligently generated content to the teacher's teaching intentions, supporting the dynamic coordination of teaching content and test question style with the teaching rhythm and classroom feedback, and providing an intelligent decision-making support foundation for large-scale personalized teaching. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0072] Figure 1 This is a flowchart of an intelligent assisted teacher lesson preparation method based on an educational macro model proposed by the present invention. DETAILED DESCRIPTION

[0073] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0074] refer to Figure 1 , an intelligent assisted teacher lesson preparation method based on an education macro model, comprising the following steps:

[0075] S1. Collect and integrate teacher preparation data from multiple sources, pre-process the teacher preparation data, and generate a standardized teacher preparation dataset;

[0076] S2. Use semantic parsing of standardized teacher lesson preparation datasets to construct a teaching knowledge graph;

[0077] S3. Based on the educational macro model, perform semantic understanding and contextual analysis on the lesson preparation demand data input by teachers, match the lesson preparation demand data with the teaching knowledge graph, and generate preliminary intelligent lesson preparation content;

[0078] S4. Present the initially generated intelligent lesson preparation content and provide an interactive interface that combines natural language and voice recognition, allowing teachers to preview, edit, and provide feedback on the intelligent lesson preparation content in real time, generating interactive feedback data;

[0079] S5. Analyze the interactive feedback data to extract content modifications, key adjustments, and personalized needs involved in the feedback. Based on the feedback information, dynamically optimize the initially generated intelligent lesson preparation content to generate adjusted intelligent lesson preparation content.

[0080] S6. Based on teachers’ historical lesson preparation data and interactive feedback data, adaptively learn and optimize the intelligent lesson preparation content, adjust the weights of key knowledge points in the teaching knowledge graph, and update the content generation strategy of the education model to generate the final intelligent lesson preparation plan.

[0081] In this embodiment, S1 includes the following steps:

[0082] S11. Obtain the original education data set D raw The original education data set includes the primary and secondary school textbook content set T m , syllabus content collection T s , Collection of past exam questions T e And the collection of primary and secondary school learning question bank content T q ;

[0083] S12. For the original education data set D raw Perform text cleaning to remove and standardize the garbled information, redundant symbols, non-teaching content, and format errors contained in the original education data set, and obtain the cleaned education text set D clean ,Among them, the cleaning process adopts a unified standard rule to ensure the ,consistency of the expression structure of the content from each source;

[0084] S13. Cleaned educational text set D clean Perform format standardization processing, standardize the paragraph structure, question type identification, knowledge point description and teaching terms, and obtain the standardized education data set D norm ,Among them, format standardization processing maintains the consistent expression form of all samples in ,the semantic structure and information level;

[0085] S14. Educational data set D based on standardized format norm Mark key knowledge points and associate each educational data sample with its corresponding teaching knowledge point set K = {k1, k2, ..., k n} to match and label a knowledge point in the dataset to form the labeled education data set D tag , where k i Represents the i-th knowledge point, and the annotation results are embedded into each sample data in a structured label manner;

[0086] S15. For the labeled educational data set D tagRe-index and group according to the subject, grade, course content and knowledge point category, organize each sample data into a structured form, and construct a structured education dataset D struct , where each piece of structured education data includes: education text sample x i , corresponding knowledge point label k i and its classification information i ,Classification information includes subject attributes, grade dimensions and question type categories.

[0087] In this embodiment, S2 includes the following steps:

[0088] S21. Based on structured education dataset D struct For each educational text sample x i Divide the teaching semantic units and construct the teaching semantic block set B i ={b i1 ,b i2 ,...,b im}, the teaching semantic block represents the smallest information unit with consistent teaching logic, which is used to capture the semantic dependency structure of teaching objectives and knowledge points;

[0089] S22. Teaching semantic block set B i Each teaching semantic block b ij Apply the context embedding representation function to generate the semantic embedding vector E ij , and map the semantic embedding vector to the semantic compression space S sem , retaining semantic factors related to test question generation;

[0090] S23. Introducing the semantic activation function A(k t ,E ij )=cosine(φ k (k t ),E ij ) Calculate the teaching semantic block b ij For knowledge point k t The semantic support strength of k (k t ) represents the embedding representation of the knowledge point, which is used to guide the correlation between the knowledge point and the context semantics;

[0091] S24. Construct the teaching knowledge graph G based on the set of all knowledge point labels K, the set of teaching objectives G, and the set of teaching semantic blocks B. edu =(V,E,W), where the node set V includes the knowledge point node v k , teaching target node v g and teaching semantic block node v b, the edge set E includes semantic support edges, target dependency edges and chapter structure edges, and the weight set W is composed of the semantic activation function value and the context embedding representation function;

[0092] S25. Based on teaching knowledge graph G edu For a given teaching target node v g and knowledge point node v k Execute context-aware path reasoning algorithm to calculate the test question generation teaching path P inf :

[0093]

[0094] Among them, P inf is the teaching path that is ultimately used to generate test questions. For all teaching target nodes v g To knowledge point node v k Redundancy(p) is the path redundancy penalty function used to avoid repeated references to the same semantic block, λ div To control the balance coefficient between diversity rewards and structural overlap, the obtained teaching path is used to guide the educational macro-model to generate goal-oriented test questions;

[0095] S26. Set the teaching path P inf Encoded as a semantic path vector π, it serves as the input semantic control factor of the educational model generation module, guiding the model to generate a set of test questions with rigorous logic, consistent semantics, and clear knowledge points.

[0096] S27. Generate a set of test questions Perform knowledge point call analysis to extract the set of knowledge points called in actual generation and teaching objectives set And compare it with the target set K, G in the path reasoning phase to calculate the semantic coverage deviation:

[0097]

[0098] Among them, Δ sem is the semantic coverage deviation of the semantic path of the current generated test question to the teaching objective, and β is the weight factor of the semantic coverage deviation;

[0099] S28. Based on the feedback results of semantic coverage deviation, adjust the edge weight set W in the teaching knowledge graph, increase the edge weights of semantic nodes that appear in the path but are not called, optimize the edge strength of the misactivated path, and form a closed loop of semantic graph optimization driven by test question generation:

[0100]

[0101] Among them, W′(u,v) is the edge weight set after feedback adjustment, η is the structure update learning rate, Indicates whether the edge participates in the current round of reasoning path.

[0102] In this embodiment, S3 includes the following steps:

[0103] S31. Receive the teacher's input lesson preparation demand data set D plan , the teacher's input lesson preparation demand data set D plan All text information in the text is semantically encoded and processed, and the teacher's lesson preparation vector representation Ψ is generated through the context understanding module in the education model. plan ,The teacher’s lesson preparation vector represents the information that integrates the teaching goal intention, course context, learning group characteristics and teaching method requirements;

[0104] S32. Calling the teaching knowledge graph G edu The teacher's lesson plan vector is represented by Ψ plan Match each of the teaching semantic path vector sets π in the teaching knowledge graph one by one, calculate their semantic relevance, and obtain the semantic relevance score between the teacher's lesson preparation vector representation and each semantic path;

[0105] S33. Based on the calculated semantic relevance score, filter out the vector representation of teacher lesson preparation Ψ plan The path set that meets the preset semantic relevance threshold is used as the semantic support path set π for generating intelligent lesson preparation content. * ;

[0106] S34. Semantic support path set Π obtained based on screening * And the teacher's lesson plan vector representation Ψ plan , the preliminary intelligent lesson preparation content set C is generated by the education model plan ={c1,c2,...,c k The initial intelligent lesson preparation content set includes the draft syllabus, the first draft of the courseware, classroom handouts, teaching cases, and personalized exercise suggestions. Each content is generated by the semantic generation function of the education model under the dual constraints of path semantics and teacher intention.

[0107] S35. Preliminary intelligent lesson preparation content set C plan Each content in the teaching consistency test and content structure integrity assessment is carried out, and the content adaptation score vector Ω={ω1,ω2,...,ω k}, the rating vector is used to represent the structural adaptation of each content relative to the teacher's input intention and the alignment with the teaching goal;

[0108] S36. The intelligent lesson preparation content with scores lower than the preset tolerance threshold τ in the fitness score vector Ω is returned to the education model for regeneration and correction, and the revision record in the form of structured tags is retained in the regenerated version, and finally a preliminary intelligent lesson preparation content set is generated. The preliminary intelligent lesson preparation content set meets the teaching objectives, contextual logic and individual characteristics requirements.

[0109] In this embodiment, the lesson preparation demand data set includes a teaching objective description, a course theme keyword set, teaching stage information, teaching method preferences, and a student portrait parameter set.

[0110] In this embodiment, S4 includes the following steps:

[0111] S41. Gather the preliminary intelligent lesson preparation content Input into the intelligent interactive presentation module for structured display. The intelligent interactive presentation module presents the teaching outline draft, courseware draft, classroom handouts, teaching cases and personalized exercise suggestions in a variety of ways, using document view, slide view and interactive component view respectively according to the content type.

[0112] S42. Based on the structured display interface, the natural language interaction submodule and the speech recognition interaction submodule are integrated into the interaction interface. The natural language interaction submodule receives the text instruction set Γ input by the teacher. text , the speech recognition interaction submodule receives the teacher speech input set Γ voice , where the text instruction set and the speech input set are uniformly mapped to the teaching feedback semantic space;

[0113] S43. Input natural language into set Γ text and the speech input set Γ voice The common encoding is the teacher interaction feedback representation set Γ feedback ,The teacher interaction feedback representation set is used to express the teacher’s ,editing and modification intentions, reconstruction suggestions, ,content preferences or teaching style adjustment requirements for the ,preliminary intelligent lesson preparation content;

[0114] S44. Representation of the teacher interaction feedback set Γ feedback With intelligent lesson preparation content collection Match each content one by one and identify the content index set I corresponding to the feedback target match , and extract the original semantic path vector set Π of the corresponding content segment match ;

[0115] S45. Perform semantic alignment on each teacher interaction feedback representation and its mapped content segment to extract the offset features and adjustment direction of the teacher's intention in the original path semantics, forming a feedback offset vector set Δ edit ;

[0116] S46. Generate interactive feedback dataset Each feedback data record includes the original semantic path vector π l , feedback offset vector δ l And the feedback target corresponding content index i l .

[0117] In this embodiment, S5 includes the following steps:

[0118] S51. Receive and parse the interactive feedback dataset F interact , extracting the content modification information, teaching focus adjustment requirements and personalized teaching preferences involved in the feedback;

[0119] S52. Offset vector δ based on teacher feedback l The semantic features of the corresponding original semantic path vector π l Perform semantic reconstruction to generate the adjusted path semantic vector π′ l ,The semantic reconstruction process keeps the original structural logic of the path unchanged, and only performs minimal ,disturbance optimization on the node content, edge weight distribution, or knowledge point organization;

[0120] S53. Set all the reconstructed path semantic vectors Π′={π′1,π′2,...,π′ s} is input as a semantic driving factor into the content regeneration module of the education model, linking the original teacher preparation vector representation Ψ plan Jointly generate an adjusted set of intelligent lesson preparation content

[0121] S54. In the interactive feedback dataset F interact Based on the teacher feedback offset vector delta l With the original semantic path vector pi l As input, the intelligent lesson preparation content collection Conduct semantic-guided reconstruction to generate a collection of intelligent lesson preparation content after fine-tuning the content The adjusted intelligent lesson preparation content collection retains the original semantic path logical structure and integrates the content modifications, teaching focus adjustments, and personalized needs expressed in teacher feedback. Specifically, it includes the draft teaching outline, preliminary courseware, class handouts, teaching cases, and personalized exercise suggestions.

[0122] S55. Final output of the adjusted intelligent lesson preparation content set The adjusted intelligent lesson preparation content collection is used to support teachers' teaching implementation in real time and dynamically support personalized teaching design.

[0123] In this implementation, the teaching outline draft: based on the original semantic path vector π l For the covered teaching objective nodes, if the teacher feedback offset vector contains information on teaching objective decomposition and knowledge point reordering, the priority of the teaching objective nodes in the path will be reordered, and the order of items and the level of objective expression in the draft teaching outline will be updated accordingly;

[0124] Courseware Draft: Based on the Original Semantic Path Vector π l If the teacher's feedback offset vector contains slide layout preferences, diagram suggestions, or content simplification requirements, the weights of nodes related to visual expression in the path are adjusted, and the courseware page structure and information distribution are reconstructed.

[0125] Lecture Notes: Based on the original semantic path vector π l For the knowledge point coverage section in the teacher feedback, if the offset vector contains the need to change the example questions, expand the lecture structure, or adjust the image-text ratio, the corresponding knowledge point segment is partially updated and the exemplary content node in the path is replaced;

[0126] Teaching case: Based on the original semantic path vector π l If the teacher feedback offset vector contains the teaching context reconstruction, background replacement or style correction requirements, the dialogue framework and context setting of the path segment are regenerated, and the starting and ending points of the path remain unchanged.

[0127] Personalized exercise suggestions: based on the original semantic path vector π l The associated knowledge point nodes and their graph edge weight information. If the teacher's feedback offset vector contains ability adaptation, difficulty control or test point reinforcement requirements, the node weights and edge structure in the path segment will be fine-tuned to reconstruct the matching degree of question type and difficulty while maintaining consistency with the original path logic.

[0128] An intelligent-assisted teacher preparation system based on an educational macro model is applied to an intelligent-assisted teacher preparation method based on an educational macro model, including the following modules:

[0129] The data processing module is used to collect primary and secondary school textbooks, teaching syllabuses, past examination questions and exercise banks, and to perform text cleaning, format standardization and key knowledge point annotation on the collected education data to generate structured education data sets;

[0130] The educational model building module is used to pre-train and fine-tune structured educational datasets to build educational models for primary and secondary school scenarios;

[0131] The teaching knowledge graph construction module is used to construct the teaching knowledge graph based on the teaching semantic blocks, teaching objectives and knowledge point relationships, forming a semantic path reasoning mechanism;

[0132] The intelligent lesson preparation content generation module is used to receive the lesson preparation demand data input by teachers, combine the education big model with the teaching knowledge map, and generate preliminary intelligent lesson preparation content such as the teaching outline draft, courseware draft, classroom handouts, teaching cases and personalized exercise suggestions;

[0133] The interactive feedback processing module is used to collect teachers' edits, style preferences, and teaching adjustment requirements for intelligent lesson preparation content through the natural language and speech recognition interactive interface, and generate semantic offset representation and interactive feedback datasets;

[0134] The dynamic optimization module is used to reconstruct the original semantic path and update the graph path based on teacher feedback, generate dynamically adjusted intelligent lesson preparation content, and drive the subsequent test question generation process.

[0135] Example 1:

[0136] On October 12, 2024, the third-year mathematics group of A Experimental Middle School introduced this invention for the first time in the teaching topic of "Graphs and Properties of Quadratic Functions" to conduct an intelligent proposition-assisted experiment.

[0137] At 8:16 AM, Mathematics teacher Li logged into the intelligent teaching assistance system through the school teaching platform and entered the following lesson preparation requirements:

[0138] Grade: Grade 9; Subject: Mathematics; Teaching Objectives: Master the relationship between the vertex form and the general form of quadratic functions; Be able to judge the properties of functions based on graphs; Current Progress: Lesson 3 of the teaching unit, has covered "y=ax 2 +bx+c"; Teacher's personalized preferences: tend to favor structured graph questions, multiple-question questions, and life-context guided questions; Teacher input text instructions: Please generate a set of exercises containing moderately difficult quadratic function graph questions, including multiple-choice, fill-in-the-blank, and solutions, covering three core knowledge points. It is recommended that the total number of questions be controlled within 10.

[0139] After receiving the input, the system starts the semantic parsing module at 08:17:41 to convert the teacher input into a semantic embedding representation Ψ plan , calling the structured education dataset for semantic path matching, and completing the construction of the knowledge graph subgraph related to the teaching objectives at 08:18:07, which contains 24 nodes and a total of 51 semantic path candidate sets.

[0140] At 08:18:09, the system performed context-aware path reasoning, selecting a path set that covered the "axial symmetry of the quadratic function image," "the relationship between the function opening direction and the positive and negative coefficients," and "the change in vertex coordinates," generating path vectors π1, π2, and π3, with path structure depths of 5, 7, and 6, respectively.

[0141] At 08:18:14, the education model generated 13 preliminary candidate test questions based on the path semantic vector, including:

[0142] Multiple-choice questions: 5; fill-in-the-blank questions: 3; solution questions: 5; scenario questions involve three types of life contexts: "traffic light change cycle", "parabolic basketball shooting", and "water fountain design".

[0143] At 08:18:16, the system activated the mutant fish school algorithm to optimize the global fitness of 13 questions, setting the evaluation indicators including:

[0144] Knowledge point coverage C k : Must cover 3 input target knowledge points; Difficulty distribution D s : Controlled at medium to high difficulty (system-defined score between 4.0–4.5 / 5); language standardization score L f :Requires fluent sentences and complete questions; the top 9 questions with the best fitness function scores are selected. The optimization took 0.91 seconds to generate a preliminary set of questions.

[0145] At 08:19, teacher Li began to preview the test questions and provided the following modification instructions through voice interaction:

[0146] "The coordinates of the image in question 4 are unclear. I suggest adding a picture." "Question 6 is too broad. Can it be split into two questions?" "Question 8 can be framed in a school context, such as the design of a parabolic school gate."

[0147] The system recognizes the speech content in real time and generates feedback offset vectors δ4, δ6, and δ8, which act on the relevant nodes in the path vectors π2 and π3 to adjust the path hierarchy.

[0148] At 08:21:08, the system automatically reconstructed the test question generation path and updated the test question content as follows:

[0149] Question 4 adds dynamic image auxiliary information; Question 6 is split into two questions, with the second question focusing on "the influence of the sign of coefficient a on the opening direction"; Question 8 switches the scene to "the arch design of the campus main gate", and the image uses a parabola to express the gate structure.

[0150] At 08:22, the teacher clicked "Save and Generate PDF", and the system automatically generated a structured exercise paper containing questions, answers, test point annotations and preset scoring standards, and simultaneously uploaded it to the school platform and class homework system.

[0151] The teacher feedback collection system's statistics for the day are as follows:

[0152] The system takes 4 minutes and 26 seconds to generate test questions; the teacher makes 3 manual revisions (all via voice interaction); the student completion rate is 98.1% (47 out of 48 students in the class completed the test); the average completion time is 21 minutes and 43 seconds; the accuracy of knowledge points covered on the test paper (AI statistics): 95.8%; the teacher satisfaction score is 4.82 / 5.

[0153] In contrast, Li spent 47 minutes on the exercise paper of the previous unit "Pythagorean Theorem and Triangle" using the traditional method. The teacher manually revised it 7 times, and the student completion rate was only 92.4%. The knowledge point coverage accuracy was 82.7%, and the satisfaction score was 3.94 / 5.

[0154] In terms of test question training data, the function question corpus used in this round of training totals:

[0155] High-quality quadratic function questions in history: 4130; structural annotation data matching the semantic nodes of the textbook: 3120; teacher manual correction feedback records: 984 (used for fine-tuning the path semantic offset); total path nodes called during the system operation process: 1724.

[0156] It can be seen from Example 1 that the present invention can not only automatically generate a set of test questions that meet the needs of teachers, but also dynamically adapt to teacher feedback through knowledge graph path reasoning, semantic embedding and graph structure optimization, and achieve test question output with semantic consistency, reasonable difficulty and high consistency with teaching objectives, which greatly improves the efficiency of teachers' question setting, the consistency of students' answers and the overall teaching evaluation accuracy, and solves the problems of low efficiency in test question generation, difficulty in controlling quality and lack of personalized interactive feedback in existing methods.

[0157] The present invention constructs a multi-level graph structure including teaching target nodes, knowledge point nodes and teaching semantic block nodes, introduces semantic activation function and path redundancy penalty function, uses path inference algorithm to obtain the optimal teaching path vector, and dynamically adjusts the graph edge weight based on the semantic coverage deviation between the generated test questions and the target knowledge points after the test questions are generated, thereby realizing a generation-driven self-learning update mechanism of the graph structure, breaking through the limitations of the static use of traditional knowledge graphs, making the semantic path generated by the test questions highly consistent with the teaching structure, and significantly improving the path generation accuracy and semantic coverage.

[0158] The present invention uniformly maps the teacher's natural language input and voice instructions into interactive feedback representations, generates a feedback offset vector through semantic alignment analysis, and acts on the original path semantic vector for minimum disturbance reconstruction, thereby realizing structured modification of multiple types of content such as teaching outlines, handouts, cases, and test question suggestions. On this basis, feedback data synchronously updates the teaching knowledge graph and generation model, forming a closed-loop content optimization mechanism, which significantly improves the responsiveness of intelligently generated content to teachers' teaching intentions, supports dynamic coordination of teaching content and test question style with the teaching rhythm and classroom feedback, and provides an intelligent decision-making support foundation for large-scale personalized teaching.

[0159] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. An intelligent teacher-assisted lesson preparation method based on an educational model, characterized in that: The steps include: S1. Collect and integrate multi-source teacher preparation data related to teacher preparation, and pre-process the teacher preparation data to generate a standardized teacher preparation dataset; S2. Use semantic parsing of standardized teacher lesson preparation datasets to construct a teaching knowledge graph; The S2 comprises the following steps: S21. Based on structured education dataset D struct For each educational text sample x i Divide the teaching semantic units and construct the teaching semantic block set B i ={b i1 ,b i2 ,...,b im }, the teaching semantic block represents the smallest information unit with consistent teaching logic, which is used to capture the semantic dependency structure of teaching objectives and knowledge points; S22. Teaching semantic block set B i Each teaching semantic block b ij Apply the context embedding representation function to generate the semantic embedding vector E ij , and map the semantic embedding vector to the semantic compression space S sem , retaining semantic factors related to test question generation; S23. Introducing the semantic activation function A(k t ,E ij )=cosine(φ k (k t ),E ij ) Calculate the teaching semantic block b ij For knowledge point k t The semantic support strength of k (k t ) represents the embedding representation of the knowledge point, which is used to guide the correlation between the knowledge point and the context semantics; S24. Construct the teaching knowledge graph G based on the set of all knowledge point labels K, the set of teaching objectives G, and the set of teaching semantic blocks B. edu =(V,E,W), where the node set V includes the knowledge point node v k , teaching target node v g and teaching semantic block node v b , the edge set E includes semantic support edges, target dependency edges and chapter structure edges, and the weight set W is composed of the semantic activation function value and the context embedding representation function; S25. Based on teaching knowledge graph G edu For a given teaching target node v g and knowledge point node v k Execute context-aware path reasoning algorithm to calculate the test question generation teaching path P inf : Among them, P inf is the teaching path that is ultimately used to generate test questions. For all teaching target nodes v g To knowledge point node v k Redundancy(p) is the path redundancy penalty function used to avoid repeated references to the same semantic block, λ div To control the balance coefficient between diversity rewards and structural overlap, the obtained teaching path is used to guide the educational macro-model to generate goal-oriented test questions; S26. Set the teaching path P inf Encoded as a semantic path vector π, it serves as the input semantic control factor of the educational model generation module, guiding the model to generate a set of test questions with rigorous logic, consistent semantics, and clear knowledge points. S27. Generate a set of test questions Perform knowledge point call analysis to extract the set of knowledge points called in actual generation and teaching objectives set And compare it with the target set K, G in the path reasoning phase to calculate the semantic coverage deviation: Among them, Δsem is the semantic coverage deviation of the current generated test question semantic path to the teaching objective, and β is the weight factor of the semantic coverage deviation; S28. Based on the feedback results of semantic coverage deviation, adjust the edge weight set W in the teaching knowledge graph, increase the edge weights of semantic nodes that appear in the path but are not called, optimize the edge strength of the misactivated path, and form a closed loop of semantic graph optimization driven by test question generation: Among them, W′(u,v) is the edge weight set after feedback adjustment, η is the structure update learning rate, To indicate whether the edge is involved in the current round of reasoning path; S3. Perform semantic understanding and contextual analysis on the lesson preparation demand data input by teachers based on the educational macro model, match the lesson preparation demand data with the teaching knowledge graph, and generate preliminary intelligent lesson preparation content; S4. Present the initially generated intelligent lesson preparation content and provide an interactive interface that combines natural language and voice recognition, allowing teachers to preview, edit, and provide feedback on the intelligent lesson preparation content in real time, generating interactive feedback data; S5. Analyze the interactive feedback data, extract the content modifications, key adjustments and personalized needs involved in the feedback, and dynamically optimize the initially generated intelligent lesson preparation content based on the feedback information to generate adjusted intelligent lesson preparation content; S6. Based on the teacher's historical lesson preparation data and the interactive feedback data, the intelligent lesson preparation content is adaptively learned and optimized, the weights of key knowledge points in the teaching knowledge graph are adjusted, and the content generation strategy of the education model is updated to generate the final intelligent lesson preparation plan.

2. The intelligent teaching preparation method based on the educational model according to claim 1 is characterized in that: Said S1 comprises the following steps: S11. Obtain the original education data set D raw The original education data set includes the primary and secondary school textbook content set T m , syllabus content collection T s , Collection of past exam questions T e And the collection of primary and secondary school learning question bank content T q ; S12. For the original education data set D raw Perform text cleaning to remove and standardize the garbled information, redundant symbols, non-teaching content, and format errors contained in the original education data set, and obtain the cleaned education text set D clean ; S13. Cleaned educational text set D clean Perform format standardization processing, standardize the paragraph structure, question type identification, knowledge point description and teaching terms, and obtain the standardized education data set D norm ; S14. Educational data set D based on standardized format norm Mark key knowledge points and associate each educational data sample with its corresponding teaching knowledge point set K = {k1, k2, ..., k n } to match and label a knowledge point in the dataset to form the labeled education data set D tag , where k i Represents the i-th knowledge point, and the annotation results are embedded into each sample data in a structured label manner; S15. For the labeled educational data set D tag Re-index and group according to the subject, grade, course content and knowledge point category, organize each sample data into a structured form, and construct a structured education dataset D struct , where each piece of structured education data includes: education text sample x i , corresponding knowledge point label k i and its classification information i ,Classification information includes subject attributes, grade dimensions and question type categories.

3. The intelligent teaching preparation method based on the educational model according to claim 2 is characterized in that: The S3 includes the following steps: S31. Receive the teacher's input lesson preparation demand data set D plan , the teacher's input lesson preparation demand data set D plan All text information in the text is semantically encoded and processed, and the teacher's lesson preparation vector representation Ψ is generated through the context understanding module in the education model. plan ,The teacher’s lesson preparation vector represents the information that integrates the teaching goal intention, course context, learning group characteristics and teaching method requirements; S32. Calling the teaching knowledge graph G edu The teacher's lesson plan vector is represented by Ψ plan Match each of the teaching semantic path vector sets π in the teaching knowledge graph one by one, calculate their semantic relevance, and obtain the semantic relevance score between the teacher's lesson preparation vector representation and each semantic path; S33. Based on the calculated semantic relevance score, filter out the vector representation of teacher lesson preparation Ψ plan The path set that meets the preset semantic relevance threshold is used as the semantic support path set π for generating intelligent lesson preparation content. * ; S34. Semantic support path set Π obtained based on screening * And the teacher's lesson plan vector representation Ψ plan , the preliminary intelligent lesson preparation content set C is generated by the education model plan ={c1,c2,...,c k The initial intelligent lesson preparation content set includes the draft syllabus, the first draft of the courseware, classroom handouts, teaching cases, and personalized exercise suggestions. Each content is generated by the semantic generation function of the education model under the dual constraints of path semantics and teacher intention. S35. Preliminary intelligent lesson preparation content set C plan Each content in the teaching consistency test and content structure integrity assessment is carried out, and the content adaptation score vector Ω={ω1,ω2,...,ω k }, the rating vector is used to represent the structural adaptation of each content relative to the teacher's input intention and the alignment with the teaching goal; S36. The intelligent lesson preparation content with scores lower than the preset tolerance threshold τ in the fitness score vector Ω is returned to the education model for regeneration and correction, and the revision record in the form of structured tags is retained in the regenerated version, and finally a preliminary intelligent lesson preparation content set is generated. The preliminary intelligent lesson preparation content set meets the teaching objectives, contextual logic and individual characteristics requirements.

4. The intelligent teaching preparation method based on the educational model according to claim 3 is characterized in that: The lesson preparation demand data set includes teaching objective description, course theme keyword set, teaching stage information, teaching method preference and student portrait parameter set.

5. The intelligent teaching preparation method based on the educational model according to claim 4 is characterized in that: The S4 comprises the following steps: S41. Gather the preliminary intelligent lesson preparation content Input into the intelligent interactive presentation module for structured display. The intelligent interactive presentation module presents the teaching outline draft, courseware draft, classroom handouts, teaching cases and personalized exercise suggestions in a variety of ways, using document view, slide view and interactive component view according to the content type. S42. Based on the structured display interface, the natural language interaction submodule and the speech recognition interaction submodule are integrated into the interaction interface. The natural language interaction submodule receives the text instruction set Γ input by the teacher. text , the speech recognition interaction submodule receives the teacher speech input set Γ voice , where the text instruction set and the speech input set are uniformly mapped to the teaching feedback semantic space; S43. Input natural language into set Γ text and the speech input set Γ voice The common encoding is the teacher interaction feedback representation set Γ feedback ,The teacher interaction feedback representation set is used to express the teacher’s ,editing and modification intentions, reconstruction suggestions, ,content preferences or teaching style adjustment requirements for the ,preliminary intelligent lesson preparation content; S44. Representation of the teacher interaction feedback set Γ feedback With intelligent lesson preparation content collection Match each content one by one and identify the content index set I corresponding to the feedback target match , and extract the original semantic path vector set Π of the corresponding content segment match ; S45. Perform semantic alignment on each teacher interaction feedback representation and its mapped content segment to extract the offset features and adjustment direction of the teacher's intention in the original path semantics, forming a feedback offset vector set Δ edit ; S46. Generate interactive feedback dataset Each feedback data record includes the original semantic path vector π l , feedback offset vector δ l And the feedback target corresponding content index i l .

6. The method for preparing lessons for intelligently-assisted teachers based on an educational macro model according to claim 1, characterized in that: The S5 comprises the following steps: S51. Receive and parse the interactive feedback dataset F interact , extracting the content modification information, teaching focus adjustment requirements and personalized teaching preferences involved in the feedback; S52. Offset vector δ based on teacher feedback l The semantic features of the corresponding original semantic path vector π l Perform semantic reconstruction to generate the adjusted path semantic vector π′ l ,The semantic reconstruction process keeps the original structural logic of the path unchanged, and only performs minimal ,disturbance optimization on the node content, edge weight distribution, or knowledge point organization; S53. Set all reconstructed path semantic vectors Π′={π′1,π′2,...,π′ s } is input as a semantic driving factor into the content regeneration module of the education model, linking the original teacher preparation vector representation Ψ plan Jointly generate an adjusted set of intelligent lesson preparation content S54. In the interactive feedback dataset F interact Based on the teacher feedback offset vector delta l With the original semantic path vector pi l As input, the intelligent lesson preparation content collection Conduct semantic-guided reconstruction to generate a collection of intelligent lesson preparation content after fine-tuning the content The adjusted intelligent lesson preparation content collection retains the original semantic path logical structure and integrates the content modifications, teaching focus adjustments, and personalized needs expressed in teacher feedback. Specifically, it includes the draft teaching outline, preliminary courseware, class handouts, teaching cases, and personalized exercise suggestions. S55. Final output of the adjusted intelligent lesson preparation content set The adjusted intelligent lesson preparation content collection is used to support teachers' teaching implementation in real time and dynamically support personalized teaching design.

7. The method for preparing lessons for intelligently-assisted teachers based on an educational model according to claim 6, characterized in that: The draft syllabus: Based on the original semantic path vector π l For the covered teaching objective nodes, if the teacher feedback offset vector contains information on teaching objective decomposition and knowledge point reordering, the priority of the teaching objective nodes in the path will be reordered, and the order of items and the level of objective expression in the draft teaching outline will be updated accordingly; The first draft of the courseware: Based on the original semantic path vector π l If the teacher's feedback offset vector contains slide layout preferences, diagram suggestions, or content simplification requirements, the weights of nodes related to visual expression in the path are adjusted, and the courseware page structure and information distribution are reconstructed. The lecture notes: Based on the original semantic path vector π l For the knowledge point coverage section in the teacher feedback, if the offset vector contains the need to change the example questions, expand the lecture structure, or adjust the image-text ratio, the corresponding knowledge point segment is partially updated and the exemplary content node in the path is replaced; The teaching case: Based on the original semantic path vector π l If the teacher feedback offset vector contains the teaching context reconstruction, background replacement or style correction requirements, the dialogue framework and context setting of the path segment are regenerated, and the starting and ending points of the path remain unchanged. The personalized exercise suggestion is based on the original semantic path vector π l The associated knowledge point nodes and their graph edge weight information. If the teacher's feedback offset vector contains ability adaptation, difficulty control or test point reinforcement requirements, the node weights and edge structure in the path segment will be fine-tuned to reconstruct the matching degree of question type and difficulty while maintaining consistency with the original path logic.

8. An intelligent-assisted teacher lesson preparation system based on an educational macro model, applied to an intelligent-assisted teacher lesson preparation method based on an educational macro model according to any one of claims 1 to 7, characterized in that: Includes the following modules: The data processing module is used to collect primary and secondary school textbooks, teaching syllabuses, past examination questions and exercise banks, and to perform text cleaning, format standardization and key knowledge point annotation on the collected education data to generate structured education data sets; The educational model building module is used to pre-train and fine-tune structured educational datasets to build educational models for primary and secondary school scenarios; The teaching knowledge graph construction module is used to construct the teaching knowledge graph based on the teaching semantic blocks, teaching objectives and knowledge point relationships, forming a semantic path reasoning mechanism; The intelligent lesson preparation content generation module is used to receive the lesson preparation demand data input by teachers, combine the education big model with the teaching knowledge map, and generate preliminary intelligent lesson preparation content such as the teaching outline draft, courseware draft, classroom handouts, teaching cases and personalized exercise suggestions; The interactive feedback processing module is used to collect teachers' edits, style preferences, and teaching adjustment requirements for intelligent lesson preparation content through the natural language and speech recognition interactive interface, and generate semantic offset representation and interactive feedback datasets; The dynamic optimization module is used to reconstruct the original semantic path and update the graph path based on teacher feedback, generate dynamically adjusted intelligent lesson preparation content, and drive the subsequent test question generation process.

Images