Intelligent question-setting method for English reading comprehension in junior middle schools based on the educational macro-model

Abstract

The present invention discloses an intelligent proposition method for English reading comprehension in junior middle school based on an educational macro-model, comprising the following steps: S1, identifying proposition ideas and common proposition methods of proposition personnel; S2, generating a proposition strategy knowledge base; S3, obtaining text materials for English reading comprehension in junior middle school and performing preprocessing operations; S4, determining the difficulty distribution of reading materials in different knowledge dimensions; S5, processing the content of the questions in different difficulty dimensions; S6, refining the content details and difficulty progressive levels of the questions through a spatial fractal algorithm; S7, generating a multidimensional evaluation result of the students' knowledge level; S8, the system optimizes and adjusts the next generated question through an educational macro-model combined with fuzzy topology and spatial fractal algorithms according to the multidimensional evaluation results and the students' learning trajectories. While meeting the learning needs of different students, the present invention significantly improves the quality of question generation and the teaching effect.

Description

Technical Field

[0001] The present invention relates to the technical field of English for junior middle schools, and in particular to an intelligent proposition method for English reading comprehension for junior middle schools based on an educational macro model. Background Art

[0002] In the existing technology, the setting of English reading comprehension questions in junior high schools mainly relies on manual design and traditional information processing technology. Teachers or question setters manually compile reading comprehension questions based on the content of the textbooks and the learning objectives of students. These questions usually include multiple-choice questions, fill-in-the-blank questions, and short-answer questions. Although the above-mentioned setting methods can meet the teaching needs to a certain extent, they also have significant limitations. First, the manual setting process is time-consuming and requires high professional ability of the question setters. It is difficult to generate high-quality and personalized questions on a large scale in a short period of time. Secondly, the traditional setting method is difficult to fully consider the learning level and knowledge mastery of students, which easily leads to an imbalance in the difficulty of the questions or incomplete coverage of knowledge points, making it difficult to effectively achieve dynamic optimization of teaching results.

[0003] In addition, in the existing research on the use of information technology for English question setting, many methods are limited to shallow text analysis. For example, question generation technology based on keyword matching or simple rules generates questions by extracting vocabulary or sentence structure from the article. It may be effective when processing simple corpus, but it is still unable to understand complex semantics and generate comprehensive questions. In addition, this type of technology often lacks in-depth analysis of students' learning behavior and answer feedback, resulting in the content and difficulty of the generated questions being out of touch with the students' actual level.

[0004] Existing intelligent question-setting methods lack the ability to make dynamic adjustments. Although some studies have attempted to adjust the difficulty of questions based on student answer data, due to the lack of modeling of multi-dimensional learning characteristics, the adjustment strategies are usually relatively simple and difficult to fully adapt to students' personalized needs. In addition, these technologies cannot effectively utilize students' error patterns and learning progress data for refined analysis, resulting in insufficient targeting and personalization in the teaching process. Summary of the invention

[0005] One purpose of the present invention is to propose an intelligent proposition method for English reading comprehension in junior high school based on an educational macro-model. While meeting the learning needs of different students, the present invention significantly improves the quality of question generation and teaching effect.

[0006] According to an embodiment of the present invention, a method for intelligent question setting for junior high school English reading comprehension based on an educational macro model comprises the following steps:

[0007] S1. Through the educational big model, we conduct in-depth analysis of historical question data, question setters’ question setting records, and students’ answer feedback data, extract typical question setting strategies, and identify question setters’ question setting ideas and common question setting methods;

[0008] S2, based on fuzzy topology algorithm, conduct multi-level classification of proposition strategies, construct feature sets of different question types, and generate proposition strategy knowledge base;

[0009] S3, obtaining text materials for junior high school English reading comprehension and performing preprocessing operations;

[0010] S4. Use the educational big model to perform semantic analysis on the preprocessed text materials, extract the text's theme, paragraph structure, grammatical difficulties and vocabulary difficulty information, compare the semantic analysis results with the question feature set in the proposition strategy knowledge base at multiple levels, annotate the key knowledge points in the text through the fuzzy topology algorithm, and determine the difficulty distribution of the reading materials in different knowledge dimensions;

[0011] S5. Based on the difficulty distribution and proposition strategy knowledge base, the system uses the educational big model to automatically select the appropriate question type, generate a preliminary question framework, and build a hierarchical progressive structure of the question content in combination with the spatial fractal algorithm, and classify the question content according to different difficulty dimensions;

[0012] S6. Based on the determination of the question type and hierarchical structure, the question content and standard answers are automatically generated through the educational model. The question content is matched with the student's learning level based on the semantic structure analysis results of the aforementioned text, and the content details and difficulty progression of the question are refined through the spatial fractal algorithm;

[0013] S7. After students complete the questions, the system collects students' answer feedback data, performs multi-level analysis on the students' answer feedback data through fuzzy topology algorithm, identifies students' weak links in vocabulary mastery, syntactic comprehension and text comprehension, and generates multi-dimensional evaluation results of students' knowledge level;

[0014] S8. Based on the multi-dimensional evaluation results and students' learning trajectories, the system optimizes and adjusts the next generated questions through the educational big model combined with fuzzy topology and spatial fractal algorithms.

[0015] Optionally, the S1 specifically includes the following steps:

[0016] S11. Based on the education big model, preliminary data cleaning and feature extraction are performed on the history question data set. During the cleaning process, incomplete, repeated or irrelevant question data are removed, and the core feature set F related to the reading comprehension questions is extracted. h , including question types, difficulty levels, vocabulary, and grammatical structures;

[0017] S12. Based on the education big model, semantic analysis is performed on the question setter's question record set to extract the question intention feature set F p, including topic selection direction, knowledge point coverage and question difficulty distribution;

[0018] S13. Perform statistical analysis on the students’ answer feedback data set, calculate the students’ answer accuracy and error distribution on each question, and extract the learning level feature set F s , including depth of comprehension, vocabulary mastery and grammatical accuracy;

[0019] S14. Use the fuzzy logic reasoning model to perform multi-dimensional mapping and clustering on the historical proposition feature set, proposition intention feature set and student learning level feature set, and establish a feature mapping function:

[0020] M(F h ,F p ,F s )={m 1 ,m 2 ,…,m v};

[0021] Among them, m v Represents the typical proposition strategy after feature clustering, including the feature distribution of different question types, difficulty level division and knowledge point coverage relationship;

[0022] S15. Based on the feature mapping function, the clustering results are semantically optimized through the education model to identify the proposition ideas and common proposition methods of the proposition personnel and generate a proposition strategy set:

[0023] P m ={p m1 ,p m2 ,…,p mw};

[0024] Among them, p mw It represents the key content of a proposition strategy, including topic, type combination and knowledge point coverage distribution.

[0025] Optionally, the S2 specifically includes the following steps:

[0026] S21. Using fuzzy topology algorithm to solve the proposition strategy set P m Perform multi-level classification and construct a topological structure model of proposition strategy T = (X, τ), X = P m represents the feature space of propositional strategies, τ represents the family of open sets in the fuzzy topological structure;

[0027] S22. Calculate the fuzzy membership vector of each proposition strategy according to the topological structure model T. Extract feature sets of different question types:

[0028] F t ={f t1,f t2 ,…,f tn};

[0029] in, The feature vector representing the j-th question type, including the question difficulty D tj =μ 1 (p mj ), Vocabulary Selection V tj =μ 2 (p mj ), grammatical structure G tj =μ 3 (p mj ) and text content relevance C tj =μ 4 (p mj );

[0030] S23. Feature set F for different question types t Each eigenvector f in tj Perform quantitative analysis and establish a feature quantization function to obtain feature data Q(f tj ):

[0031] Q(f tj )=w D ·D tj +w V ·V tj +w G ·G tj +w C ·C tj ;

[0032] Among them, w D ,w V ,w G ,w C is the weight coefficient of each feature dimension, D tj ,V tj ,G tj ,C tj are the fuzzy membership of the j-th question type in terms of question difficulty, vocabulary selection, grammatical structure, and text content relevance;

[0033] S24, the quantized feature data Q(f tj ) Input the fuzzy topological classification model, use the improved fuzzy C-means clustering algorithm to perform multi-dimensional fuzzy clustering, and solve the clustering objective function:

[0034]

[0035] Among them, u jk is the feature vector f tjFor the fuzzy membership of the kth cluster center, m>1 is the fuzzy index, which controls the fuzziness of the membership, v k is the kth cluster center, representing the characteristic center of the question type, and c is the preset number of clusters;

[0036] By iteratively optimizing the fuzzy membership u jk and cluster center v k Form a multi-level classification structure of question types:

[0037] H={h 1 ,h 2 ,…,h c};

[0038] Among them, h c Indicates the classification result of the topic at the cth level;

[0039] S25. Generate a proposition strategy knowledge base based on the multi-level classification structure H and quantitative features:

[0040] K p ={(h k ,v k )|k=1,2,…,c};

[0041] Among them, v k is the feature center of the k-th question type, including statistical information on question difficulty, vocabulary selection, grammatical structure, and text content relevance.

[0042] Optionally, the S4 specifically includes the following steps:

[0043] S41. Obtaining text materials for junior high school English reading comprehension D t , and perform preprocessing operations on the text material D t Perform standardization, remove redundant symbols, perform sentence and segmentation, and obtain the preprocessed text material D′ t ;

[0044] S42. Use the educational model to process the preprocessed text material D′ t Perform semantic analysis to extract the topic of the text s , paragraph structure p 、Grammar Difficulties G d and vocabulary difficulty information V d ;

[0045] S43, the semantic analysis results are compared with the question feature set F in the question strategy knowledge base t Conduct multi-level comparisons;

[0046] S44. Using fuzzy topology algorithm to analyze the text D′ tThe key knowledge points in the text are annotated and the fuzzy topological space T′=(X′,τ′) is constructed, where X′ represents the knowledge point set of the preprocessed text.

[0047] S45. Calculate each knowledge point x′ j ∈X′ fuzzy membership in different knowledge dimensions Forming the difficulty distribution matrix:

[0048]

[0049] in, Represents knowledge point x′ j The membership degree on the kth knowledge dimension, n is the number of knowledge points, and m is the number of knowledge dimensions;

[0050] S46. According to the difficulty distribution matrix M d , determine the difficulty distribution of reading materials in different knowledge dimensions D d :

[0051]

[0052] Among them, d k Represents the average difficulty value of the kth knowledge dimension.

[0053] Optionally, the S5 specifically includes the following steps:

[0054] S51. According to the difficulty distribution of reading materials in different knowledge dimensions D d and proposition strategy knowledge base K p Using the Educational Model to Select and Difficulty Distribution D d Matching topic type set T q , each question type t i Satisfy t i ∈h k , and v k With D d The similarity S(v k ,D d )maximum:

[0055]

[0056] Among them, w i represents the weight of the i-th knowledge dimension, v k,i and d i are the characteristic value and difficulty distribution value on the i-th dimension respectively;

[0057] S52. Preliminary screening of the topic type set T q Generate a set of question frameworks using the educational model q , each topic frame fi ∈F q Including topic theme, content framework and preset difficulty distribution;

[0058] S53, combine the spatial fractal algorithm to build a hierarchical structure of the topic content, and for each topic framework f i Define the progressive hierarchy function L(f i ):

[0059]

[0060] Among them, l j (f i ) is the content of the topic at level j, r j It represents the progressive scale factor of this level, and n is the number of fractal progressive levels;

[0061] S54. Classify the content of the questions according to the depth of understanding of the knowledge points and define the classification difficulty matrix M g :

[0062] M g =[m ij ] n×l ;

[0063] Among them, m ij Indicates the difficulty value of the jth question at the i-th level;

[0064] S55. Screen the generated question content and hierarchical structure, and finally form a question set that meets the needs of students at different learning levels. t .

[0065] Optionally, the S6 specifically includes the following steps:

[0066] S61. Generate a topic content set C based on the topic type set and the progressive hierarchy using the educational model t , each topic content c k Includes question stem, option set and preliminarily generated standard answers;

[0067] S62. Combine the semantic structure analysis results with the students’ learning level and use the educational model to generate the content of the questions. k Perform matching and establish an adaptation scoring function:

[0068]

[0069] Among them, w i is the weight of the i-th feature dimension, R s,i is the value of the semantic analysis result in the i-th dimension, L s,i is the value of the student’s learning level in the i-th dimension, c k,iFor the title content c k The eigenvalue in the i-th dimension;

[0070] S63, select the adaptation scoring function S(c k ,R s ,L s ) Large topic content c k As the basic content of the current difficulty level, the content details of the question are refined by combining the spatial fractal algorithm to generate the final question content set C f , each final question c′ k The detailed rules are:

[0071] c′ k (x) = c k (x)+r j d(x);

[0072] Among them, c k (x) is the preliminary topic content c k The eigenvalue at position x, r j is the scaling factor of the progressive level, and d(x) is the feature shift function of semantic analysis, which is defined as:

[0073]

[0074] S64, the generated topic content C f Automatically generate standard answer A f The standard answer is based on the large model of the text content D′ t Deep semantic analysis, combined with the question stem and option set O k Solved by the optimal path selection method, defined as:

[0075]

[0076] Among them, P(o|Q k ,D′ t ) indicates that option o is in the question Q k and text content D′ t The semantic matching probability under ;

[0077] S65, the content of the refined question C f And the generated standard answer A f Verify the content and answers of the questions to match the learning levels of students at different levels, and output a complete set of questions that meet the requirements:

[0078] F c ={(c′ 1 ,a 1 ),(c′ 2 ,a 2),…,(c′ p ,a p )}.

[0079] Optionally, the S7 specifically includes the following steps:

[0080] S71. After the student completes the question, the system collects the student's answer feedback data, including the correct answer rate, answering time and answer error type of each question;

[0081] S72. Extracting core features related to student performance from the answer feedback data, the core features include the student's correct understanding of knowledge points, adaptability to question difficulty, and distribution of error types;

[0082] S73. Use fuzzy topology algorithm to conduct multi-level analysis on students’ answer feedback characteristics. According to the three main dimensions of vocabulary mastery, syntactic comprehension and text comprehension, map students’ performance characteristics to different levels of knowledge points and construct a fuzzy affiliation model that reflects students’ knowledge level.

[0083] S74. Analyze students’ performance data in different dimensions, and use the fuzzy affiliation model to comprehensively evaluate the characteristic data of each dimension to generate students’ ability evaluation results in each knowledge dimension, including the proficiency of vocabulary, the application level of syntactic rules, and the depth of understanding of the main idea of ​​the text;

[0084] S75. Analyze students’ performance on specific knowledge points based on their knowledge dimension assessment results, identify students’ weak links and knowledge blind spots, and generate a set of key features that reflect students’ knowledge weaknesses, covering the core areas of students’ enhanced learning;

[0085] S76. Generate a knowledge level report for students based on the assessment results. The knowledge level report describes the student's performance in vocabulary, syntax, and text comprehension, and matches the student's weak links with specific knowledge points.

[0086] The beneficial effects of the present invention are:

[0087] (1) The present invention can conduct in-depth semantic analysis of English reading materials by combining the educational big model with the fuzzy topological algorithm, extract the multi-dimensional features of the text's theme, paragraph structure, grammatical difficulties and vocabulary difficulty, and use the proposition strategy knowledge base to realize the intelligent selection of question types and the automatic generation of question content. Compared with the traditional manual proposition method, the present invention constructs a hierarchical progressive structure of question content based on the spatial fractal algorithm to ensure the dynamic adaptability of question difficulty and knowledge point coverage, thereby realizing personalized question generation. The present invention can adjust the question content and difficulty in real time according to the students' learning level and answer feedback, significantly improving the adaptability of the question to the students' actual needs. Through semantic analysis and intelligent generation technology, the proposition efficiency and scope of application are significantly improved.

[0088] (2) The present invention collects and analyzes students' answer feedback data and uses fuzzy topological algorithm to construct a multidimensional evaluation model of students' knowledge level, so as to accurately identify students' weak links in vocabulary mastery, syntactic comprehension and text comprehension. The multidimensional evaluation model can deeply model students' learning behavior and generate dynamic personalized knowledge weakness analysis results, further guiding question generation and teaching optimization. Compared with the existing intelligent proposition system based on static evaluation, the present invention introduces a real-time adjustment mechanism based on feedback, which enables the system to dynamically optimize the question content to better support students' learning process and improve teaching efficiency.

[0089] (3) The present invention introduces a spatial fractal algorithm in the process of question generation to construct a hierarchical progressive structure of question content and a multi-dimensional difficulty distribution model, so that the question content can be refined at multiple progressive levels, gradually improving from basic understanding to high-level analysis, ensuring that the questions can fully cover the students' cognitive level and learning objectives. Traditional methods usually adopt static question design and cannot refine the hierarchy of question content. The present invention realizes precise control of question content and difficulty distribution through progressive modeling of fractal algorithm, which significantly improves the quality of question generation and teaching effect while meeting the learning needs of different students. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] 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:

[0091] Figure 1 The present invention is a flowchart of an intelligent proposition method for English reading comprehension in junior high school based on an educational macro-model. DETAILED DESCRIPTION

[0092] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0093] refer to Figure 1 , an intelligent proposition method for English reading comprehension in junior middle schools based on an educational macro model, comprising the following steps:

[0094] S1. Through the educational big model, we conduct in-depth analysis of historical question data, question setters’ question setting records, and students’ answer feedback data, extract typical question setting strategies, and identify question setters’ question setting ideas and common question setting methods;

[0095] S2, based on fuzzy topology algorithm, conduct multi-level classification of proposition strategies, construct feature sets of different question types, and generate proposition strategy knowledge base;

[0096] S3, obtaining text materials for junior high school English reading comprehension and performing preprocessing operations;

[0097] S4. Use the educational big model to perform semantic analysis on the preprocessed text materials, extract the text's theme, paragraph structure, grammatical difficulties and vocabulary difficulty information, compare the semantic analysis results with the question feature set in the proposition strategy knowledge base at multiple levels, annotate the key knowledge points in the text through the fuzzy topology algorithm, and determine the difficulty distribution of the reading materials in different knowledge dimensions;

[0098] S5. Based on the difficulty distribution and proposition strategy knowledge base, the system uses the educational big model to automatically select the appropriate question type, generate a preliminary question framework, and build a hierarchical progressive structure of the question content in combination with the spatial fractal algorithm, and classify the question content according to different difficulty dimensions;

[0099] S6. Based on the determination of the question type and hierarchical structure, the question content and standard answers are automatically generated through the educational model. The question content is matched with the student's learning level based on the semantic structure analysis results of the aforementioned text, and the content details and difficulty progression of the question are refined through the spatial fractal algorithm;

[0100] S7. After students complete the questions, the system collects students' answer feedback data, performs multi-level analysis on the students' answer feedback data through fuzzy topology algorithm, identifies students' weak links in vocabulary mastery, syntactic comprehension and text comprehension, and generates multi-dimensional evaluation results of students' knowledge level;

[0101] S8. Based on the multi-dimensional evaluation results and students' learning trajectories, the system optimizes and adjusts the next generated questions through the educational big model combined with fuzzy topology and spatial fractal algorithms.

[0102] In this implementation, S1 specifically includes the following steps:

[0103] S11. Based on the education big model, preliminary data cleaning and feature extraction are performed on the history question data set. During the cleaning process, incomplete, repeated or irrelevant question data are removed, and the core feature set F related to the reading comprehension questions is extracted. h , including question types, difficulty levels, vocabulary, and grammatical structures;

[0104] S12. Based on the education big model, semantic analysis is performed on the question setter's question record set to extract the question intention feature set F p , including topic selection direction, knowledge point coverage and question difficulty distribution;

[0105] S13. Perform statistical analysis on the students’ answer feedback data set, calculate the students’ answer accuracy and error distribution on each question, and extract the learning level feature set F s , including depth of comprehension, vocabulary mastery and grammatical accuracy;

[0106] S14. Use the fuzzy logic reasoning model to perform multi-dimensional mapping and clustering on the historical proposition feature set, proposition intention feature set and student learning level feature set, and establish a feature mapping function:

[0107] M(F h ,F p ,F s )={m 1 ,m 2 ,…,m v};

[0108] Among them, m v Represents the typical proposition strategy after feature clustering, including the feature distribution of different question types, difficulty level division and knowledge point coverage relationship;

[0109] S15. Based on the feature mapping function, the clustering results are semantically optimized through the education model to identify the proposition ideas and common proposition methods of the proposition personnel and generate a proposition strategy set:

[0110] P m ={p m1 ,p m2 ,…,p mw};

[0111] Among them, p mw It represents the key content of a proposition strategy, including topic, type combination and knowledge point coverage distribution.

[0112] In this implementation, S2 specifically includes the following steps:

[0113] S21. Using fuzzy topology algorithm to solve the proposition strategy set P m Perform multi-level classification and construct a topological structure model of proposition strategy T = (X, τ), X = P m represents the feature space of propositional strategies, τ represents the family of open sets in the fuzzy topological structure;

[0114] S22. Calculate the fuzzy membership vector of each proposition strategy according to the topological structure model T. Extract feature sets of different question types:

[0115] F t ={f t1 ,f t2 ,…,f tn};

[0116] in, The feature vector representing the j-th question type, including the question difficulty D tj =μ 1 (p mj ), Vocabulary Selection V tj =μ 2 (p mj ), grammatical structure G tj =μ 3 (p mj ) and text content relevance C tj =μ 4 (p mj );

[0117] S23. Feature set F for different question types t Each eigenvector f in tj Perform quantitative analysis and establish a feature quantization function to obtain feature data Q(f tj ):

[0118] Q(f tj )=w D ·D tj +w V ·V tj +w G ·G tj +w C ·C tj ;

[0119] Among them, w D ,w V ,w G ,w C is the weight coefficient of each feature dimension, D tj ,V tj ,G tj ,C tjare the fuzzy membership of the j-th question type in terms of question difficulty, vocabulary selection, grammatical structure, and text content relevance;

[0120] S24, the quantized feature data Q(f tj ) Input the fuzzy topological classification model, use the improved fuzzy C-means clustering algorithm to perform multi-dimensional fuzzy clustering, and solve the clustering objective function:

[0121]

[0122] Among them, u jk is the feature vector f tj For the fuzzy membership of the kth cluster center, m>1 is the fuzzy index, which controls the fuzziness of the membership, v k is the kth cluster center, representing the characteristic center of the question type, and c is the preset number of clusters;

[0123] By iteratively optimizing the fuzzy membership u jk and cluster center v k Form a multi-level classification structure of question types:

[0124] H={h 1 ,h 2 ,…,h c};

[0125] Among them, h c Indicates the classification result of the topic at the cth level;

[0126] S25. Generate a proposition strategy knowledge base based on the multi-level classification structure H and quantitative features:

[0127] K p ={(h k ,v k )|k=1,2,…,c};

[0128] Among them, v k is the feature center of the k-th question type, including statistical information on question difficulty, vocabulary selection, grammatical structure, and text content relevance.

[0129] In this implementation, S4 specifically includes the following steps:

[0130] S41. Obtaining text materials for junior high school English reading comprehension D t , and perform preprocessing operations on the text material D t Perform standardization, remove redundant symbols, perform sentence and segmentation, and obtain the preprocessed text material D′ t ;

[0131] S42. Use the educational model to process the preprocessed text material D′t Perform semantic analysis to extract the topic of the text s , paragraph structure p 、Grammar Difficulties G d and vocabulary difficulty information V d ;

[0132] S43, the semantic analysis results are compared with the question feature set F in the question strategy knowledge base t Conduct multi-level comparisons;

[0133] S44. Using fuzzy topology algorithm to analyze the text D′ t The key knowledge points in the text are annotated and the fuzzy topological space T′=(X′,τ′) is constructed, where X′ represents the knowledge point set of the preprocessed text.

[0134] S45. Calculate each knowledge point x′ j ∈X′ fuzzy membership in different knowledge dimensions Forming the difficulty distribution matrix:

[0135]

[0136] in, Represents knowledge point x′ j The membership degree on the kth knowledge dimension, n is the number of knowledge points, and m is the number of knowledge dimensions;

[0137] S46. According to the difficulty distribution matrix M d , determine the difficulty distribution of reading materials in different knowledge dimensions D d :

[0138]

[0139] Among them, d k Represents the average difficulty value of the kth knowledge dimension.

[0140] In this implementation, S5 specifically includes the following steps:

[0141] S51. According to the difficulty distribution of reading materials in different knowledge dimensions D d and proposition strategy knowledge base K p Using the Educational Model to Select and Difficulty Distribution D d Matching topic type set T q , each question type t i Satisfy t i ∈h k , and v k With D d The similarity S(v k ,D d )maximum:

[0142]

[0143] Among them, w i represents the weight of the i-th knowledge dimension, v k,i and d i are the characteristic value and difficulty distribution value on the i-th dimension respectively;

[0144] S52. Preliminary screening of the topic type set T q Generate a set of question frameworks using the educational model q , each topic frame f i ∈F q Including topic theme, content framework and preset difficulty distribution;

[0145] S53, combine the spatial fractal algorithm to build a hierarchical structure of the topic content, and for each topic framework f i Define the progressive hierarchy function L(f i ):

[0146]

[0147] Among them, l j (f i ) is the content of the topic at level j, r j It represents the progressive scale factor of this level, and n is the number of fractal progressive levels;

[0148] S54. Classify the content of the questions according to the depth of understanding of the knowledge points and define the classification difficulty matrix M g :

[0149] M g =[m ij ] n×l ;

[0150] Among them, m ij Indicates the difficulty value of the jth question at the i-th level;

[0151] S55. Screen the generated question content and hierarchical structure, and finally form a question set that meets the needs of students at different learning levels. t .

[0152] In this implementation, S6 specifically includes the following steps:

[0153] S61. Generate a topic content set C based on the topic type set and the progressive hierarchy using the educational model t , each topic content c k Includes question stem, option set and preliminarily generated standard answers;

[0154] S62, combine the semantic structure analysis results and the students' learning level to use the educational model to generate the content of the questions k Perform matching and establish an adaptation scoring function:

[0155]

[0156] Among them, w i is the weight of the i-th feature dimension, R s,i is the value of the semantic analysis result in the i-th dimension, L s,i is the value of the student’s learning level in the i-th dimension, c k,i For the title content c k The eigenvalue in the i-th dimension;

[0157] S63, select the adaptation scoring function S(c k ,R s ,L s ) Large topic content c k As the basic content of the current difficulty level, the content details of the question are refined by combining the spatial fractal algorithm to generate the final question content set C f , each final question c′ k The refinement rules are:

[0158] c′ k (x) = c k (x)+r j d(x);

[0159] Among them, c k (x) is the preliminary topic content c k The eigenvalue at position x, r j is the scaling factor of the progressive level, and d(x) is the feature shift function of semantic analysis, which is defined as:

[0160]

[0161] S64, the generated topic content C f Automatically generate standard answer A f The standard answer is based on the large model of the text content D′ t Deep semantic analysis, combined with the question stem and option set O k Solved by the optimal path selection method, defined as:

[0162]

[0163] Among them, P(o|Q k ,D′ t ) indicates that option o is in the question Q k and text content D′ tThe semantic matching probability under ;

[0164] S65, the content of the refined question C f And the generated standard answer A f Verify the content and answers of the questions to match the learning levels of students at different levels, and output a complete set of questions that meet the requirements:

[0165] F c ={(c′ 1 ,a 1 ),(c′ 2 ,a 2 ),…,(c′ p ,a p )}.

[0166] In this implementation, S7 specifically includes the following steps:

[0167] S71. After the student completes the question, the system collects the student's answer feedback data, including the correct answer rate, answering time and answer error type of each question;

[0168] S72. Extracting core features related to student performance from the answer feedback data, the core features include the student's correct understanding of knowledge points, adaptability to question difficulty, and distribution of error types;

[0169] S73. Use fuzzy topology algorithm to conduct multi-level analysis on students’ answer feedback characteristics. According to the three main dimensions of vocabulary mastery, syntactic comprehension and text comprehension, map students’ performance characteristics to different levels of knowledge points and construct a fuzzy affiliation model that reflects students’ knowledge level.

[0170] S74. Analyze students’ performance data in different dimensions, and use the fuzzy affiliation model to comprehensively evaluate the characteristic data of each dimension to generate students’ ability evaluation results in each knowledge dimension, including the proficiency of vocabulary, the application level of syntactic rules, and the depth of understanding of the main idea of ​​the text;

[0171] S75. Analyze students’ performance on specific knowledge points based on their knowledge dimension assessment results, identify students’ weak links and knowledge blind spots, and generate a set of key features that reflect students’ knowledge weaknesses, covering the core areas of students’ enhanced learning;

[0172] S76. Generate a knowledge level report for students based on the assessment results. The knowledge level report describes the student's performance in vocabulary, syntax, and text comprehension, and matches the student's weak links with specific knowledge points.

[0173] Embodiment 1:

[0174] In order to verify the feasibility and effectiveness of the intelligent proposition method for junior high school English reading comprehension based on the educational big model of the present invention, the following Example 1 is described in detail in combination with the actual teaching scenario. This Example 1 describes the application process in English teaching in a key junior high school in a certain city, and demonstrates the advantages of the present invention over the traditional proposition method through data comparison.

[0175] Eighth grade students in a key junior high school in a certain city are preparing for an English reading comprehension test. The school hopes to generate a set of reading comprehension questions for students through an intelligent question-setting system to improve teaching efficiency and accuracy. There are 4 classes in the grade, with a total of 200 students. There are certain differences in English proficiency, ranging from students with weak vocabulary foundation to those who can master complex syntax. The traditional question-setting method requires teachers to write questions based on textbooks or reference materials, which usually takes a long time and it is difficult to achieve an accurate match between the difficulty of the questions and the students' levels, especially when personalized questions are required for different students.

[0176] First, the teacher imports the reading comprehension materials through the intelligent proposition system. The reading comprehension materials are carefully selected by the teaching and research group and include 5 articles with clear themes and paragraph structures, covering environmental protection, cultural exchange and content suitable for the cognitive level of junior high school students. The system pre-processes these materials, removes redundant characters and divides them into sentences, and then uses the educational big model for semantic analysis to extract the theme, paragraph structure, grammatical difficulty and vocabulary difficulty information of each article.

[0177] The system compares the analysis results with the proposition strategy knowledge base to determine the appropriate question types, including multiple-choice questions, fill-in-the-blank questions, and short-answer questions. Subsequently, the system combines the spatial fractal algorithm to construct a hierarchical progressive structure of the question content. In Example 1, for an article about "protecting the earth's ecological environment", the system generates the following hierarchical questions:

[0178] 1. Basic level: Ask the article theme: "What is the main topic of the article?" and set four options to cover different dimensions of environmental protection;

[0179] 2. Level of detail: Ask specific facts: "According to the passage, which of the following is not a reason to protect forests?"

[0180] 3. Higher-level questions: Questions that involve reasoning and critical thinking: "If you were the environmental advocate mentioned in the article, how would you promote awareness of protecting forests?"

[0181] After each question is generated, the system automatically generates a standard answer and marks the difficulty of the question. At the same time, the system assigns questions to different students and dynamically adjusts the question content according to the students' learning level. In Example 1, for students with weaker foundations, the system generates questions with simple vocabulary and clear options, while for students with stronger English proficiency, the system adds long and difficult sentences and contextual reasoning questions.

[0182] After the students completed the questions, the system collected each student's answer data, including the accuracy rate, answering time and error types. Through the fuzzy topological algorithm, the system conducted a multi-dimensional analysis of the students' performance and identified that some students had weak links in vocabulary comprehension and syntactic application. Subsequently, the system generated personalized exercises for these weak links for subsequent learning reinforcement.

[0183] In order to evaluate the actual effect of the present invention, a comparative test was conducted between the present invention and the traditional proposition method, and the data are as follows in Table 1:

[0184] Table 1 Comparison test data between the present invention and the traditional proposition method

[0185]

[0186] In Table 1 above, the system generated a personalized question set of 600 questions for 200 students, covering different levels of difficulty and knowledge points. The feedback from students after answering the questions showed that the correct rate of students with weak foundations increased by 15%, and the time taken by students with strong abilities to answer advanced questions was reduced by 20%. In addition, teachers reported that the questions generated by the system not only reduced the workload of setting questions, but also significantly improved the efficiency of classroom teaching.

[0187] In the test, 50 students were selected for intensive training with personalized exercises. It was found that after practicing with personalized questions automatically generated by the system, the students' accuracy rate in tests targeting weak links increased from an average of 70% to 90%. In contrast, the accuracy rate of another group of students who practiced using traditional methods only increased by 10%.

[0188] This Example 1 verifies the significant advantages of the present invention in intelligent question setting, dynamic adjustment of question content and weak link analysis through actual teaching scenarios. Compared with traditional methods, the present invention excels in efficiency, accuracy and personalization.

[0189] The present invention can conduct in-depth semantic analysis of English reading materials by combining the educational big model with the fuzzy topological algorithm, extract the multi-dimensional features of the text's theme, paragraph structure, grammatical difficulties and vocabulary difficulty, and use the proposition strategy knowledge base to realize the intelligent selection of question types and the automatic generation of question content. Compared with the traditional manual proposition method, the present invention constructs a hierarchical progressive structure of question content based on the spatial fractal algorithm to ensure the dynamic adaptability of question difficulty and knowledge point coverage, thereby realizing personalized question generation. The present invention can adjust the question content and difficulty in real time according to the students' learning level and answer feedback, significantly improving the adaptability of the question to the students' actual needs. Through semantic analysis and intelligent generation technology, the proposition efficiency and scope of application are significantly improved.

[0190] The present invention constructs a multi-dimensional evaluation model of students' knowledge level by collecting and analyzing students' answer feedback data using the fuzzy topology algorithm, so as to accurately identify the weak links of students in vocabulary mastery, syntactic understanding, and text understanding. The multi-dimensional evaluation model can deeply model students' learning behaviors and generate dynamic personalized analysis results of knowledge weak points, further guiding question generation and teaching optimization. Compared with the existing intelligent question proposition system based on static evaluation, the present invention introduces a real-time adjustment mechanism based on feedback, enabling the system to dynamically optimize the question content to better support students' learning processes and improve teaching efficiency.

[0191] The present invention introduces a spatial fractal algorithm in the process of question generation to construct a hierarchical progressive structure of question content and a multi-dimensional difficulty distribution model, enabling the question content to be refined at multiple progressive levels, gradually ascending from basic understanding to higher-order analysis, ensuring that the questions can comprehensively cover students' cognitive levels and learning objectives. Traditional methods usually adopt static question designs and cannot refine the hierarchy of question content. However, the present invention realizes precise control of question content and difficulty distribution through progressive modeling of the fractal algorithm, significantly improving the quality of question generation and teaching effects while meeting the learning needs of different students.

[0192] As described above, the above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent replacements or changes, and should be covered by the protection scope of the present invention.

Claims

1. An intelligent proposition method for English reading comprehension in junior middle schools based on an educational model, characterized in that: The steps include: S1. Through the educational big model, we conduct in-depth analysis of historical question data, question setters’ question setting records, and students’ answer feedback data, extract typical question setting strategies, and identify question setters’ question setting ideas and common question setting methods; S2, based on fuzzy topology algorithm, conduct multi-level classification of proposition strategies, construct feature sets of different question types, and generate proposition strategy knowledge base; S3, obtaining text materials for junior high school English reading comprehension and performing preprocessing operations; S4. Use the educational big model to perform semantic structure analysis on the preprocessed text materials, extract the text's theme, paragraph structure, grammatical difficulty and vocabulary difficulty information, compare the semantic structure analysis results with the question feature set in the proposition strategy knowledge base at multiple levels, annotate the key knowledge points in the text through the fuzzy topology algorithm, and determine the difficulty distribution of the reading materials in different knowledge dimensions; S5. Based on the difficulty distribution and proposition strategy knowledge base, the system uses the educational big model to automatically select the appropriate question type, generate a preliminary question framework, and build a hierarchical progressive structure of the question content in combination with the spatial fractal algorithm, and classify the question content according to different difficulty dimensions; S6. Based on the determination of the question type and hierarchical structure, the question content and standard answers are automatically generated through the educational model. The question content is matched with the student's learning level based on the semantic structure analysis results of the aforementioned text, and the content details and difficulty progression of the question are refined through the spatial fractal algorithm; S7. After students complete the questions, the system collects students' answer feedback data, performs multi-level analysis on the students' answer feedback data through fuzzy topology algorithm, identifies students' weak links in vocabulary mastery, syntactic comprehension and text comprehension, and generates multi-dimensional evaluation results of students' knowledge level; S8. Based on the multi-dimensional evaluation results and students' learning trajectories, the system optimizes and adjusts the next generated questions through the educational big model combined with fuzzy topology and spatial fractal algorithms.

2. According to claim 1, a method for intelligent question setting for English reading comprehension in junior middle schools based on an educational model is characterized in that: The S1 specifically includes the following steps: S11. Based on the education big model, preliminary data cleaning and feature extraction are performed on the history question data set. During the cleaning process, incomplete, repeated or irrelevant question data are removed, and the core feature set F related to the reading comprehension questions is extracted. h , including question types, difficulty levels, vocabulary, and grammatical structures; S12. Based on the education big model, semantic analysis is performed on the question setter's question record set to extract the question intention feature set F p , including topic selection direction, knowledge point coverage and question difficulty distribution; S13. Perform statistical analysis on the students’ answer feedback data set, calculate the students’ answer accuracy and error distribution on each question, and extract the learning level feature set F s , including depth of comprehension, vocabulary mastery and grammatical accuracy; S14. Use the fuzzy logic reasoning model to perform multi-dimensional mapping and clustering on the historical proposition feature set, proposition intention feature set and student learning level feature set, and establish a feature mapping function: M(F h ,F p ,F s )={m1,m2,…,m v }; Among them, m v represents the vth typical proposition strategy after feature clustering, including the feature distribution of different question types, difficulty level division and knowledge point coverage relationship; S15. Based on the feature mapping function, the clustering results are semantically optimized through the education model to identify the proposition ideas and common proposition methods of the proposition personnel and generate a proposition strategy set: in, Represents the key content of the vth proposition strategy, including the topic, type combination and knowledge point coverage distribution.

3. The intelligent proposition method for English reading comprehension in junior middle schools based on the educational model according to claim 1 is characterized in that: The S2 specifically includes the following steps: S21. Using fuzzy topology algorithm to solve the proposition strategy set P m Perform multi-level classification and construct a topological structure model of proposition strategy T = (X, τ), X = P m represents the feature space of propositional strategies, τ represents the family of open sets in the fuzzy topological structure; S22. Calculate the fuzzy membership vector of each proposition strategy according to the topological structure model T. Extract feature sets of different question types: F t ={f t1 ,f t2 ,…,f tn }; in, The feature vector representing the j-th question type, including the question difficulty D tj =μ1(p mj ), Vocabulary Selection V tj =μ2(p mj ), grammatical structure G tj =μ3(p mj ) and text content relevance C tj =μ4(p mj ); S23. Feature set F for different question types t Each eigenvector f in tj Perform quantitative analysis and establish a feature quantization function to obtain feature data Q(f tj ): Q(f tj )=w D ·D tj +w V ·V tj +w G ·G tj +w C ·C tj ; Among them, w D ,w V ,w G ,w C is the weight coefficient of each feature dimension, D tj ,V tj ,G tj ,C tj are the fuzzy membership of the j-th question type in terms of question difficulty, vocabulary selection, grammatical structure, and text content relevance; S24, the quantized feature data Q(f tj ) Input the fuzzy topological classification model, use the improved fuzzy C-means clustering algorithm to perform multi-dimensional fuzzy clustering, and solve the clustering objective function: Among them, u jk is the feature vector f tj For the fuzzy membership of the kth cluster center, m>1 is the fuzzy index, which controls the fuzziness of the membership, v k is the kth cluster center, representing the characteristic center of the question type, and c is the preset number of clusters; By iteratively optimizing the fuzzy membership u jk and cluster center v k Form a multi-level classification structure of question types: H={h1,h2,…,h c }; Among them, h c Indicates the classification result of the topic at the cth level; S25. Generate a proposition strategy knowledge base based on the multi-level classification structure H and quantitative features: K p ={(h k ,v k )∣k=1,2,…,c}; Among them, v k is the feature center of the k-th question type, including statistical information on question difficulty, vocabulary selection, grammatical structure, and text content relevance.

4. The intelligent proposition method for English reading comprehension in junior middle schools based on the educational macro model according to claim 1 is characterized in that: The S4 specifically comprises the following steps: S41. Obtaining text materials for junior high school English reading comprehension D t , and perform preprocessing operations on the text material D t Perform standardization, remove redundant symbols, perform sentence and segmentation, and obtain the preprocessed text material D′ t ; S42. Use the educational model to process the pre-processed text material D′ t Perform semantic structure analysis to extract the topic of the text s , paragraph structure p 、Grammar Difficulties G d and vocabulary difficulty information V d ; S43, combine the semantic structure analysis results with the question feature set F in the question strategy knowledge base t Conduct multi-level comparisons; S44. Using fuzzy topology algorithm to analyze the text D′ t The key knowledge points in the text are annotated and the fuzzy topological space T′=(X′,τ′) is constructed, where X′ represents the knowledge point set of the preprocessed text. S45. Calculate each knowledge point x′ j ∈X′ fuzzy membership in different knowledge dimensions Forming the difficulty distribution matrix: in, Represents knowledge point x′ j The membership degree on the kth knowledge dimension, n 1 is the number of knowledge points, m 1 is the number of knowledge dimensions; S46. According to the difficulty distribution matrix M d , determine the difficulty distribution of reading materials in different knowledge dimensions D d : Among them, d k Represents the average difficulty value of the kth knowledge dimension.

5. The intelligent proposition method for English reading comprehension in junior middle schools based on the educational model according to claim 3 is characterized in that: The S5 specifically includes the following steps: S51. According to the difficulty distribution of reading materials in different knowledge dimensions D d and proposition strategy knowledge base K p Using the Educational Model to Select and Difficulty Distribution D d Matching topic type set T q , each question type t i Satisfy t i ∈h k , and v k With D d The similarity S(v k ,D d )maximum: Among them, w i represents the weight of the i-th knowledge dimension, v k,i and d i are the characteristic value and difficulty distribution value on the i-th dimension respectively; S52. Preliminary screening of the topic type set T q Generate a set of question frameworks using the educational model q , each topic frame f i ∈F q Including topic theme, content framework and preset difficulty distribution; S53, combine the spatial fractal algorithm to build a hierarchical structure of the topic content, and for each topic framework f i Define the progressive hierarchy function L(f i ): Among them, l j (f i ) is the content of the topic at level j, r j Indicates the progressive scaling factor of this level, n 2 is the number of layers of the fractal progression; S54. Classify the content of the questions according to the depth of understanding of the knowledge points and define the classification difficulty matrix M g : Among them, m 2 ij Indicates the difficulty value of the jth question at the i-th level; S55. Screen the generated question content and hierarchical structure, and finally form a question set that meets the needs of students at different learning levels. t .

6. The intelligent proposition method for English reading comprehension in junior middle schools based on the educational macro model according to claim 1 is characterized in that: The S6 specifically comprises the following steps: S61. Generate a topic content set C based on the topic type set and the progressive hierarchy using the educational model t , each topic content c k Includes question stem, option set and preliminarily generated standard answers; S62. Combine the semantic structure analysis results with the students’ learning level and use the educational model to generate the content of the questions. k Perform matching and establish an adaptation scoring function: Among them, w i is the weight of the i-th feature dimension, R s,i is the value of the semantic structure analysis result in the i-th dimension, L s,i is the value of the student’s learning level in the i-th dimension, c k,i For the title content c k The eigenvalue in the i-th dimension; S63, select the adaptation scoring function S(c k ,R s ,L s ) Large topic content c k As the basic content of the current difficulty level, the content details of the question are refined by combining the spatial fractal algorithm to generate the final question content set C f , each final question c′ k The refinement rules are: c′ k (x)=c k (x)+r j ·d(x); Among them, c k (x) is the preliminary topic content c k The eigenvalue at position x, r j is the scaling factor of the progressive level, and d(x) is the feature shift function of semantic structure analysis, which is defined as: S64, the generated topic content C f Automatically generate standard answer A f The standard answer is based on the large model of the text content D′ t A deep semantic structure analysis based on the question stem and option set O k Solved by the optimal path selection method, defined as: Among them, P(o|Q k ,D′ t ) indicates that option o is in the question Q k and text content D′ t The semantic matching probability under ; S65, the content of the refined question C f And the generated standard answer A f Verify the content and answers of the questions to match the learning levels of students at different levels, and output a complete set of questions that meet the requirements: F c ={(c′1,a1),(c′2,a2),…,(c′ p ,a p )}。 7. The intelligent proposition method for English reading comprehension in junior middle schools based on the educational macro model according to claim 1 is characterized in that: The S7 specifically comprises the following steps: S71. After the student completes the question, the system collects the student's answer feedback data, including the correct answer rate, answering time and answer error type of each question; S72. Extracting core features related to student performance from the answer feedback data, the core features include the student's correct understanding of knowledge points, adaptability to question difficulty, and distribution of error types; S73. Use fuzzy topology algorithm to conduct multi-level analysis on students’ answer feedback characteristics. According to the three main dimensions of vocabulary mastery, syntactic comprehension and text comprehension, map students’ performance characteristics to different levels of knowledge points, and construct a fuzzy affiliation model that reflects students’ knowledge level. S74. Analyze students’ performance data in different dimensions, and use the fuzzy affiliation model to comprehensively evaluate the characteristic data of each dimension to generate students’ ability evaluation results in each knowledge dimension, including the proficiency of vocabulary, the application level of syntactic rules, and the depth of understanding of the main idea of ​​the text; S75. Analyze students’ performance on specific knowledge points based on their knowledge dimension assessment results, identify students’ weak links and knowledge blind spots, and generate a set of key features that reflect students’ knowledge weaknesses, covering the core areas of students’ enhanced learning; S76. Generate a student knowledge level report based on the assessment results. The knowledge level report describes the student's performance in vocabulary, syntax, and text comprehension, and matches the student's weak links with specific knowledge points.

Images