Intelligent distribution system and method for after-class hosting work

Abstract

The invention discloses an intelligent distribution system and method for after-class hosting homework, and relates to the technical field of intelligent teaching, and the technical key points of the scheme are as follows: classifying and marking each knowledge point according to a classification system and a conceptual model, and converting the marked result into a knowledge point graph; the method comprises the following steps: mapping scores and wrong question conditions of students to corresponding knowledge points in a knowledge graph, obtaining mastery degree scores of the students on each knowledge point, classifying questions according to subjects and question types by utilizing a natural language processing technology and a mode recognition algorithm, evaluating difficulty of single questions, counting average question solving speed of various questions of each student, and obtaining a question solving result. The questions with the corresponding difficulty are matched according to the knowledge point mastering degree score of each student, the total quantity of homework questions is controlled by setting constraint conditions according to the average question solving speed of the students on various question types, the personalized homework list is generated according to the matched questions and the total quantity of the questions, and the students can be helped to use time more effectively.

Description

Technical Field

[0001] The present invention relates to the field of intelligent teaching technology, and in particular to an after-school tutoring homework intelligent distribution system and method. Background Art

[0002] With the in-depth development of educational informatization, the technology of intelligent allocation of after-school tutoring homework is gradually becoming a research hotspot. This technology aims to realize the automation and intelligent allocation of homework through advanced technical means such as data analysis and machine learning, and improve the efficiency and quality of tutoring services. Specifically, the technology of intelligent allocation of after-school tutoring homework will combine students' learning data, ability level, interest preferences and other multi-dimensional information, and perform intelligent matching and recommendation through algorithm models to provide students with homework tasks that are more in line with their personalized needs.

[0003] In a Chinese invention application with application publication number CN111091483A, a personalized student homework assignment system and a homework assignment method thereof are disclosed, which belong to the field of intelligent teaching technology, and include a database module that includes students' stage learning content knowledge points, a test question bank module that includes examination stage learning content knowledge points, a test question sending and feedback module, a test question evaluation module, a personalized homework generation module, and a personalized homework sending module. The mapping relationship between the modules is as follows: the test questions in the test question bank module for the examination stage learning content knowledge points are based on the stage learning content knowledge points in the database module that includes students' stage learning content knowledge points, the test question sending and feedback module sends the test questions in the test question bank module for the examination stage learning content knowledge points to students for testing, and the students then give feedback to the test question evaluation module, and the personalized homework sending module sends the personalized homework manually or electronically to the students corresponding to the mapping relationship.

[0004] In view of the above application, the prior art still has the following deficiencies: In the existing after-school care homework intelligent allocation system, although personalized allocation of student homework has been achieved, there is still a problem of failing to effectively control the time students spend on homework. This may cause some students to spend too much time due to excessive homework or high difficulty, affecting their rest and other learning activities; conversely, too little homework or too low difficulty may not be able to fully challenge students, affecting learning outcomes. Summary of the invention

[0005] 1. Technical issues to be resolved In view of the deficiencies in the prior art, the present invention provides an after-school care homework intelligent distribution system and method, which classifies and marks each knowledge point according to the classification system and conceptual model, converts the marked results into a knowledge point map, maps the students' grades and wrong questions to the corresponding knowledge points in the knowledge map, obtains the students' mastery scores on each knowledge point, uses natural language processing technology and pattern recognition algorithms to classify questions by subject and question type, evaluates the difficulty of single questions, counts the average problem-solving speed of each student in each type of question, matches questions of corresponding difficulty according to each student's knowledge point mastery score, sets constraints to control the total number of homework questions according to the students' average problem-solving speed on each type of question, generates a personalized homework list according to the matched questions and the total number of questions, and solves the problems mentioned in the background technology.

[0006] (II) Technical solution To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for intelligently allocating after-school tutoring homework, comprising the following steps: According to the curriculum standards and textbook content, a preliminary knowledge classification system for each subject is formulated, a conceptual model for each subject is constructed, each knowledge point is classified and labeled according to the classification system and conceptual model, and the labeled results are converted into a knowledge point map using an automatic extraction program; Map students' grades and wrong questions to the corresponding knowledge points in the knowledge graph to form students' personalized learning profiles. Use machine learning algorithms to evaluate the degree of knowledge mastery and obtain students' mastery scores for each knowledge point. Using natural language processing technology and pattern recognition algorithms, we classify questions by subject and question type, and evaluate the difficulty of each question. After classification and difficulty evaluation, we store the questions in a structured question bank, and calculate the average speed of each student in solving each type of question. Match questions of corresponding difficulty based on each student's knowledge mastery score. Set constraints to control the total number of homework questions based on the students' average problem-solving speed on various question types. Generate a personalized homework list based on the matched questions and total number of questions.

[0007] Furthermore, based on the curriculum standards and textbook contents, a preliminary knowledge classification system for each subject is formulated, a conceptual model for each subject is constructed, the text content in the textbooks and teaching aids is annotated, each knowledge point is classified and marked according to the classification system and conceptual model, and the annotated results are converted into a knowledge point map using an automatic extraction program.

[0008] Furthermore, the students' historical performance data are sorted out, each performance score is corresponded to each knowledge point, the students' wrong questions are classified and sorted, the number and type of wrong questions of students on each knowledge point are counted, and the students' performance and wrong question situations are mapped to the corresponding knowledge points in the knowledge graph to form the students' personalized learning files.

[0009] Furthermore, a machine learning algorithm is used to evaluate the degree of knowledge mastery. The existing student grades, wrong question data and mastery scores are used as training sets for training. For the students to be evaluated, the data in the personalized learning files are input into the trained model to obtain the mastery scores of each knowledge point, and the mastery scores are normalized to [0,1].

[0010] Furthermore, the difficulty of a single question is evaluated through the student's historical performance data, and the difficulty coefficient of a single question is calculated. The calculation formula is as follows:

[0011] Among them, DC represents the difficulty coefficient of a single question. represents the score of the ith student for a single question, and O represents the full score for a single question.

[0012] Furthermore, a difficulty threshold is set to classify questions into different difficulty levels, and the questions after classification and difficulty evaluation are stored in a structured question bank, with each question being accompanied by a clear question type and difficulty level; Get the starting and ending time of each student when completing each type of question. For each student, calculate the time taken to complete each question, and count the average problem-solving time for each type of question. Use the average problem-solving time as the average problem-solving speed.

[0013] Furthermore, for the knowledge points that students have learned in the knowledge point map, questions of corresponding difficulty are matched according to the students' mastery scores on each knowledge point, and the question difficulty coefficient is within the range of plus or minus 0.5 of the mastery score; For the knowledge points that students have just learned, questions with a difficulty coefficient less than 0.4 are matched as after-school homework.

[0014] Furthermore, the total time for students to complete the homework is set. Based on the average speed of students in solving various types of questions, constraints are set to control the total number of questions in the homework. The constraints are:

[0015] in, Indicates the total number of homework questions. represents the proportion of questions of the kth question type, represents the number of questions of the k-th question type, , K is a positive integer, Indicates the total duration, represents the average speed of solving the k-th question type and the d-th difficulty level of the current student, represents the number of questions of the kth question type and the dth difficulty level, , D is a positive integer.

[0016] Furthermore, a personalized homework list is generated based on the matching topics and the total amount, and pushed to the corresponding students. After the homework is graded, the student's personalized learning profile is updated based on the grading results.

[0017] An after-school homework intelligent distribution system includes: a knowledge graph module, a student learning analysis module, a question bank management and evaluation module, and a homework generation and distribution module; wherein, The knowledge graph module initially formulates the knowledge classification system of each subject according to the curriculum standards and textbook content, builds the concept model of each subject, classifies and marks each knowledge point according to the classification system and concept model, and uses the automatic extraction program to convert the marked results into a knowledge point graph; The student learning analysis module maps students' grades and wrong questions to the corresponding knowledge points in the knowledge graph to form students' personalized learning profiles. It uses machine learning algorithms to evaluate the degree of knowledge mastery and obtain the students' mastery scores on each knowledge point. The question bank management and evaluation module uses natural language processing technology and pattern recognition algorithms to classify questions by subject and question type, and evaluate the difficulty of each question. The classified and evaluated questions are stored in a structured question bank, and the average speed of each student in solving each type of question is counted. The homework generation and assignment module matches questions of corresponding difficulty according to each student's knowledge mastery score, sets constraints to control the total number of homework questions based on the students' average problem-solving speed on various question types, and generates a personalized homework list based on the matched questions and total number of questions.

[0018] (III) Beneficial effects The present invention provides an after-school care homework intelligent allocation system and method, which has the following beneficial effects: (1) Knowledge points are classified and labeled according to the classification system and conceptual model, and converted into a knowledge point map, which provides basic data support for the subsequent intelligent assignment of homework. This enables the intelligent assignment of after-school homework to more accurately understand the students' knowledge mastery and make more reasonable homework assignment decisions.

[0019] (2) By scoring students’ mastery of each knowledge point, we can more accurately understand their learning needs. Based on this, we can recommend homework that suits their current learning level, ensure the relevance and effectiveness of the homework, and avoid homework that is too difficult or too easy, thereby improving learning outcomes.

[0020] (3) Counting each student’s average problem-solving speed for each type of question can help us understand students’ learning situations and ability differences. Based on these data, we can develop a more personalized homework assignment plan for each student to meet their different learning needs and development potential.

[0021] (4) By accurately matching the questions to the students' current level of knowledge, we can avoid situations where the homework is too difficult, causing students to feel frustrated, or too easy, causing a lack of learning challenges. By setting constraints to control the total number of questions in the homework, we can ensure that the homework is neither too heavy for students to complete nor too easy to be challenging, thus helping students use their time more effectively. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A schematic diagram of the steps of the after-school tutoring homework intelligent allocation method of the present invention; Figure 2 It is a schematic diagram of the structure of the after-school tutoring homework intelligent allocation system of the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] See also Figure 1 The present invention provides an after-school tutoring homework intelligent allocation method, comprising the following steps: Step 1: According to the curriculum standards and textbook content, a preliminary knowledge classification system for each subject is formulated, a conceptual model for each subject is constructed, each knowledge point is classified and labeled according to the classification system and conceptual model, and the labeled results are converted into a knowledge point map using an automatic extraction program; The step 1 includes the following contents: Step 101: Collect and organize the curriculum standards for each grade and subject of primary or junior high school, textbooks for each subject, and main supplementary teaching materials to ensure the comprehensiveness and accuracy of the information; Step 102: Based on the curriculum standards and textbook content, a preliminary knowledge classification system for each subject is formulated, including a hierarchical structure of chapters, themes, and sub-themes. For example, junior high school Chinese is divided into multiple themes such as "reading", "writing", and "ancient poetry appreciation", and further refined into sub-themes; Step 103: Based on the classification system, further construct a concept model for each subject. The concept model includes all important concepts and principles of the subject, as well as the logical relationships between them. For example, construct a concept model for junior high school Chinese, including concepts such as literary common sense, rhetoric, and article structure, as well as their logical relationships. It should be noted that logical relationships include but are not limited to hierarchical relationships, parallel relationships, and causal relationships; among them, Hierarchical relationship: indicates the hierarchical or inclusive relationship between different concepts. For example, "literary common sense" may be a broader concept, which includes sub-concepts such as "ancient literature", "modern literature", and "foreign literature", and these sub-concepts can be further subdivided into more specific works, authors, genres, etc.; Parallel relationship: refers to the parallel relationship between several concepts at the same level or category. There is no direct inclusion or inclusion relationship between them, but they together constitute the core content of the level or category. In rhetoric, "metaphor", "personification", "exaggeration" and so on are parallel relationships, and they are all specific manifestations of rhetoric; Causal relationship: There may be causal relationships between some concepts, that is, one concept is the cause or result of the generation or existence of another concept. Although there may be fewer direct causal relationships in the conceptual model of junior high school Chinese, causal relationships are still an important dimension of thinking when understanding certain literary phenomena or works; Step 104: Detailed annotation of the text content in the textbooks and supplementary materials, categorizing and marking each knowledge point according to the classification system and conceptual model. During the annotation process, it should be ensured that the granularity of the knowledge points is moderate, neither too trivial nor too general; Step 105: Use the automatic extraction program to convert the annotated results into a preliminary form of a knowledge point map, including concept extraction, attribute extraction, and instance extraction. On the basis of automatic extraction, fine-grained mining and supplementation of knowledge points are performed. If the results of automatic extraction are not comprehensive or accurate enough, they need to be expanded and corrected using external knowledge. For example, for the ancient poetry appreciation part, it can be further refined to the knowledge points such as the author, background, and artistic conception of each poem; An automatic extraction program is an automated data processing tool that can automatically identify and extract the information required by users from data in various formats such as text, images, and videos through preset rules or algorithms. This information includes entity names, relationships, events, emotional tendencies, etc., depending on the design goals and application scenarios of the program; Step 106: Evaluate the knowledge point map through knowledge questions and answers, simulation tests, etc., and revise and improve it according to the evaluation results. Evaluate the constructed knowledge point map, mainly considering the two dimensions of accuracy and coverage, and verify the practicality and effectiveness of the knowledge point map through actual applications (such as knowledge questions and answers, simulation tests, etc.).

[0025] To use, combine steps 101 to 106: The knowledge points are classified and marked according to the classification system and conceptual model, and converted into a knowledge point map, which provides basic data support for the subsequent intelligent assignment of homework, so that the intelligent assignment of after-school care homework can more accurately understand the students' knowledge mastery and make more reasonable homework assignment decisions.

[0026] Step 2: Map the students’ scores and wrong questions to the corresponding knowledge points in the knowledge graph to form the students’ personalized learning profiles. Use machine learning algorithms to evaluate the students’ mastery of knowledge and obtain the students’ mastery scores for each knowledge point. The step 2 includes the following contents: Step 201: Collect students' basic information (such as name, grade, subject preference, study habits, etc.) and learning history data (such as previous test scores, homework completion status, wrong question records, etc.); Step 202: sorting the student's historical performance data (such as test scores, homework scores, etc.), ensuring that each performance can correspond to each knowledge point, classifying and sorting the student's wrong questions, analyzing the knowledge point tested by each wrong question, and counting the number and type of wrong questions of the student on each knowledge point, mapping the student's performance and wrong question situation to the corresponding knowledge point in the knowledge graph, and forming the student's personalized learning profile; Step 203: Use a machine learning algorithm (such as decision tree, random forest, neural network, etc.) to evaluate the degree of knowledge mastery. Use the existing student scores and wrong question data and mastery scores as training sets for training. By adjusting the algorithm parameters and optimizing the model structure, the accuracy and stability of the evaluation are improved. For the students to be evaluated, input the data in their learning files into the trained model to obtain the mastery scores of each knowledge point, and normalize the mastery scores to [0,1]. It should be noted that the calculation method for the mastery level score of knowledge points adopts existing calculation methods, including but not limited to algorithms based on test evaluation, algorithms based on statistical models, and algorithms based on complexity and weight; among them, Algorithms based on test evaluation: By setting specific test questions or items, the students' mastery of each knowledge point is quantitatively evaluated. This method can provide specific data results, which facilitates accurate comparison and analysis. For example, the students' mastery of each knowledge point can be directly calculated based on their scores on the questions related to each knowledge point. Algorithms based on statistical models: students' scores are considered to be normally distributed, and their mastery is evaluated by calculating the deviation of their scores from the overall mean (multiple standard deviation). For example, the student's score can be subtracted from the overall mean and then divided by the standard deviation. The resulting standardized score can be mapped to a normal distribution table to determine whether the student's mastery is excellent, good, passing or failing. Algorithm based on complexity and weight: Considering that the importance and difficulty of different knowledge points may be different, different weights can be set for each knowledge point. The student's score on each knowledge point will be adjusted according to its weight to more accurately reflect his or her overall mastery level. After weighted calculation, the mastery level score on that knowledge point is calculated. This method requires experts to conduct a weighted assessment of each knowledge point.

[0027] When used, combine the contents of step 201 to step 203: By scoring students' mastery of each knowledge point, we can more accurately understand their learning needs. Based on this, we can recommend homework that suits their current learning level, ensure the homework is targeted and effective, and avoid homework that is too high or too low in difficulty, thereby improving learning outcomes.

[0028] Step 3: Use natural language processing technology and pattern recognition algorithms to classify questions by subject and question type, and evaluate the difficulty of each question. Store the classified and evaluated questions in a structured question bank, and calculate the average speed of each student in solving each type of question. The step three includes the following contents: Step 301: Collect a large number of homework questions from multiple channels (such as textbooks, supplementary teaching materials, online question banks, etc.), and use natural language processing technology (NLP) and pattern recognition algorithms to automatically classify the questions by subject and question type, such as mathematics can be divided into algebra, geometry, etc., and Chinese can be divided into reading comprehension, writing, etc.; Natural language processing refers to the technology of using the natural language used by humans to communicate with machines. Through artificial processing of natural language, computers can read and understand the meaning of natural language texts, and express given intentions, ideas, etc. in natural language texts. The basic principles of pattern recognition usually include steps such as data collection, preprocessing, feature extraction, model building and classification judgment. Among them, the quality of data collection and preprocessing directly affects the accuracy of subsequent steps; feature extraction is the process of converting raw data into mathematical feature vectors; model building is to build a classifier based on training data; classification judgment is to use the classifier to classify new data; Step 302: Obtain the student's historical performance data (such as test scores, homework scores, etc.), evaluate the difficulty of a single question based on the student's historical performance data, and calculate the difficulty coefficient of the single question. The calculation formula is as follows:

[0029] Among them, DC represents the difficulty coefficient of a single question. represents the score of the i-th student on a single question, and O represents the full score of a single question; Step 303: Based on the difficulty coefficient, a difficulty threshold is set to classify the questions into different difficulty levels. For example, the difficulty is classified into one star to five stars. When the difficulty coefficient of the question is less than 0.2, the question difficulty is one star. When the difficulty coefficient of the question is greater than or equal to 0.2 and less than 0.4, the question difficulty is two stars. When the difficulty coefficient of the question is greater than or equal to 0.4 and less than 0.6, the question difficulty is three stars. When the difficulty coefficient of the question is greater than or equal to 0.6 and less than 0.8, the question difficulty is four stars. When the difficulty coefficient of the question is greater than or equal to 0.8, the question difficulty is five stars. Step 304: storing the questions after classification and difficulty evaluation in a structured question bank, with each question having a clear question type and difficulty level; Step 304: Obtain the start time and end time of each student when completing each question, calculate the time it takes for each student to complete each question, and calculate the average time it takes to solve each question (of different question types and different difficulty levels), and use the average time as the average speed; When used, combine the contents of step 301 to step 304: Counting each student’s average problem-solving speed for each type of question can help us understand students’ learning situations and ability differences. Based on these data, a more personalized homework assignment plan can be developed for each student to meet their different learning needs and development potential.

[0030] Step 4: Match questions of corresponding difficulty based on each student's knowledge mastery score. Set constraints to control the total number of homework questions based on the students' average problem-solving speed on various question types. Generate a personalized homework list based on the matched questions and total number of questions.

[0031] The step 4 includes the following contents: Step 401: extracting each student's knowledge point mastery score from the personalized learning profile, and matching the knowledge points that the student has learned in the knowledge point map with questions of corresponding difficulty according to their mastery scores, with the question difficulty coefficient being within the range of plus or minus 0.5 of the mastery score. For example, if the student's mastery score for the knowledge point "fraction addition and subtraction" is 0.3 points (out of 1 point), then assigning fraction addition and subtraction questions with a difficulty coefficient between 2.5 and 3.5 as homework; Step 402: Compare the current time point with the course schedule to find the most recently completed knowledge point. The course schedule clearly lists the teaching plan for each semester, month, week, and day. The teaching plan includes specific teaching dates, class time allocation, and corresponding teaching content and knowledge points. Step 403: For knowledge points that students have just learned, such as the grammar of "present progressive tense" (taking English as an example), questions with a difficulty coefficient less than 0.4 are assigned as after-school tutoring homework to help students consolidate basic knowledge. These questions are usually of low difficulty and easy for students to understand and complete; Step 404: According to the school or class regulations, the total time for students to complete the homework is set. Based on the average speed of students in solving various types of questions, constraints are set to control the total number of homework questions. The constraints are:

[0032] in, Indicates the total number of homework questions. represents the proportion of questions of the kth question type, represents the number of questions of the k-th question type, , K is a positive integer, Indicates the total duration, represents the average speed of solving the k-th question type and the d-th difficulty level of the current student, represents the number of questions of the kth question type and the dth difficulty level, , D is a positive integer; Step 405: Generate a personalized homework list based on the matching topics and total amount, and push it to the students. After the students complete the homework, push the homework to the corresponding teacher for correction. After the homework is corrected, the student's personalized learning profile is updated according to the correction results to provide more accurate data support for the next homework assignment.

[0033] When used, combine the contents of step 401 to step 405: By accurately matching questions to students' current level of knowledge mastery, we can avoid situations where the homework is too difficult, causing students to feel frustrated, or too easy, causing a lack of learning challenges. By setting constraints to control the total number of homework questions, we can ensure that the homework load is neither too heavy, causing students to be unable to complete it, nor too easy, causing it to lose its challenge. This can help students use their time more effectively.

[0034] See also Figure 2 The present invention also provides an after-school tutoring homework intelligent distribution system, including: a knowledge graph module, a student learning analysis module, a question bank management and evaluation module, and a homework generation and distribution module; wherein, The knowledge graph module initially formulates the knowledge classification system of each subject according to the curriculum standards and textbook content, builds the concept model of each subject, classifies and marks each knowledge point according to the classification system and concept model, and uses the automatic extraction program to convert the marked results into a knowledge point graph; The student learning analysis module maps students' grades and wrong questions to the corresponding knowledge points in the knowledge graph to form students' personalized learning profiles. It uses machine learning algorithms to evaluate the degree of knowledge mastery and obtain the students' mastery scores on each knowledge point. The question bank management and evaluation module uses natural language processing technology and pattern recognition algorithms to classify questions by subject and question type, and evaluate the difficulty of each question. The classified and evaluated questions are stored in a structured question bank, and the average speed of each student in solving each type of question is counted. The homework generation and assignment module matches questions of corresponding difficulty according to each student's knowledge mastery score, sets constraints to control the total number of homework questions based on the students' average problem-solving speed on various question types, and generates a personalized homework list based on the matched questions and total number of questions.

[0035] In the application, the several formulas involved are all calculated by removing dimensions and taking their numerical values. The formula is a formula of the most recent real situation obtained by collecting a large amount of data and performing software simulation. The coefficients in the formula are set by technicians in this field according to actual conditions.

[0036] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. A person of ordinary skill in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented in combination with electronic hardware, computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0037] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0038] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for intelligently allocating after-school tutoring homework, characterized in that: The following steps are involved: According to the curriculum standards and textbook content, a preliminary knowledge classification system for each subject is formulated, a conceptual model for each subject is constructed, each knowledge point is classified and labeled according to the classification system and conceptual model, and the labeled results are converted into a knowledge point map using an automatic extraction program; Map students' grades and wrong questions to the corresponding knowledge points in the knowledge graph to form students' personalized learning profiles. Use machine learning algorithms to evaluate the degree of knowledge mastery and obtain students' mastery scores for each knowledge point. Using natural language processing technology and pattern recognition algorithms, we classify questions by subject and question type, and evaluate the difficulty of each question. After classification and difficulty evaluation, we store the questions in a structured question bank, and count the average speed of each student in solving each type of question. Match questions of corresponding difficulty based on each student's knowledge mastery score. Set constraints to control the total number of homework questions based on the students' average problem-solving speed on various question types. Generate a personalized homework list based on the matched questions and total number of questions.

2. The after-school care homework intelligent allocation method according to claim 1 is characterized by: Based on the curriculum standards and textbook content, a preliminary knowledge classification system for each subject is formulated, a conceptual model for each subject is constructed, the text content in the textbooks and teaching aids is annotated, and each knowledge point is classified and marked according to the classification system and conceptual model. Using an automatic extraction program, the annotated results are converted into a knowledge point map.

3. The after-school care homework intelligent allocation method according to claim 1 is characterized by: Organize students' historical performance data, correspond each performance to each knowledge point, classify and organize students' wrong questions, count the number and type of wrong questions students have on each knowledge point, map students' performance and wrong questions to the corresponding knowledge points in the knowledge graph, and form students' personalized learning files.

4. The after-school care homework intelligent allocation method according to claim 3 is characterized by: A machine learning algorithm is used to evaluate the degree of knowledge mastery. Existing student scores, wrong question data, and mastery scores are used as training sets for training. For students to be evaluated, the data in the personalized learning archives are input into the trained model to obtain the mastery scores of each knowledge point, and the mastery scores are normalized to [0,1].

5. The after-school care homework intelligent allocation method according to claim 1 is characterized by: Through the students' historical performance data, the difficulty of a single question is evaluated and the difficulty coefficient of a single question is calculated. The calculation formula is as follows: ； Among them, DC represents the difficulty coefficient of a single question. represents the score of the ith student for a single question, and O represents the full score for a single question.

6. The after-school care homework intelligent allocation method according to claim 5 is characterized by: Set difficulty thresholds, classify questions into different difficulty levels, and store the questions after classification and difficulty assessment in a structured question bank. Each question is accompanied by question type and difficulty level. Get the starting and ending time of each student when completing each type of question. For each student, calculate the time taken to complete each question, and count the average problem-solving time for each type of question. Use the average problem-solving time as the average problem-solving speed.

7. The after-school care homework intelligent allocation method according to claim 1 is characterized by: For the knowledge points that students have learned in the knowledge point map, questions of corresponding difficulty are matched according to the students' mastery scores on each knowledge point. The difficulty coefficient of the questions is within the range of plus or minus 0.5 of the mastery score. For the knowledge points that students have just learned, questions with a difficulty coefficient less than 0.4 will be used as after-school homework.

8. The after-school care homework intelligent allocation method according to claim 7 is characterized by: Set the total time for students to complete the homework. According to the average speed of students in solving various types of questions, set constraints to control the total number of homework questions. Constraints: ； in, Indicates the total number of homework questions. represents the proportion of questions of the kth question type, represents the number of questions of the k-th question type, , K is a positive integer, Indicates the total duration, represents the average speed of solving the k-th question type and the d-th difficulty level of the current student, represents the number of questions of the kth question type and the dth difficulty level, , D is a positive integer.

9. The after-school care homework intelligent allocation method according to claim 8, characterized in that: Based on the matching questions and the total number of questions, a personalized homework list is generated and pushed to the corresponding students. After the homework is graded, the student's personalized learning profile is updated based on the grading results.

10. An after-school tutoring homework intelligent distribution system, used to implement the method described in any one of claims 1 to 9, characterized in that: include: The knowledge graph module initially formulates the knowledge classification system of each subject according to the curriculum standards and textbook content, builds the concept model of each subject, classifies and marks each knowledge point according to the classification system and concept model, and uses the automatic extraction program to convert the marked results into a knowledge point graph; The student learning analysis module maps students' grades and wrong questions to the corresponding knowledge points in the knowledge graph to form students' personalized learning profiles. It uses machine learning algorithms to evaluate the degree of knowledge mastery and obtain the students' mastery scores on each knowledge point. The question bank management and evaluation module uses natural language processing technology and pattern recognition algorithms to classify questions by subject and question type, and evaluate the difficulty of each question. The classified and evaluated questions are stored in a structured question bank, and the average speed of each student in solving each type of question is counted. The homework generation and assignment module matches questions of corresponding difficulty according to each student's knowledge mastery score, sets constraints to control the total number of homework questions based on the students' average problem-solving speed on various question types, and generates a personalized homework list based on the matched questions and total number of questions.

Images