Practical teaching method and system based on large model, electronic equipment and medium

Abstract

The invention discloses a large model-based practical teaching method and system, electronic equipment and a medium, and relates to the technical field of data processing. The method comprises the following steps: acquiring learning information of a target student; according to historical homework completion data, wrong question data and answering time length data in the learning information, constructing a student ability portrait, and based on the student ability portrait, calculating a knowledge point mastering level index of the target student; inputting the student ability portrait into a pre-constructed test question recommendation model to obtain a recommended test question group of the target student, the recommended test question group comprising a plurality of recommended test questions matched with the student ability portrait; selecting a plurality of target recommended test questions corresponding to the knowledge point mastering level indexes from the recommended test questions, and generating target test paper according to the target recommended test questions; and obtaining answer data of the target student to the target test paper, and generating a practical teaching feedback report of the target student based on the answer data. By implementing the technical scheme provided by the invention, the effect of improving the efficiency of practical teaching is achieved.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and particularly to a training teaching method, system, electronic device and medium based on a large model. Background Art

[0002] With the rapid development of information technology, intelligent teaching systems have gradually become key tools for improving the quality and efficiency of education in today's educational field. Especially in a large-scale educational environment, how to provide personalized learning materials and support according to the specific needs of each student is an important challenge for improving learning effects.

[0003] Currently, most educational institutions use intelligent teaching systems to carry out training teaching. Such systems uniformly assign learning tasks and practice questions to students according to a pre-set teaching syllabus, and evaluate learning effects based on examination scores and homework completion situations. However, in actual applications, due to differences in the knowledge bases and learning abilities of each student, a unified teaching plan is difficult to meet the learning needs of different students, resulting in some students failing to achieve the expected learning effects, thus leading to low efficiency of training teaching. Summary of the Invention

[0004] This application provides a training teaching method, system, electronic device and medium based on a large model, which has the effect of improving the efficiency of training teaching.

[0005] In a first aspect, this application provides a training teaching method based on a large model, including: Obtaining the learning information of a target student; Constructing a student ability portrait of the target student according to the historical homework completion data, wrong question data and answering duration data in the learning information, and calculating a knowledge point mastery level index of the target student based on the student ability portrait; Inputting the student ability portrait into a pre-constructed test question recommendation model to obtain a recommended test question group for the target student, where the recommended test question group includes multiple recommended test questions matching the student ability portrait; Selecting multiple target recommended test questions corresponding to the knowledge point mastery level index from each of the recommended test questions, and generating a target test paper according to each of the target recommended test questions; Obtaining the answering data of the target student for the target test paper, and generating a training teaching feedback report for the target student based on the answering data.

[0006] In a second aspect of this application, there is provided a training teaching system based on a large model, where the system includes: An information acquisition module for obtaining the learning information of a target student; An index determination module, configured to construct a student ability portrait of the target student based on the historical homework completion data, wrong question data, and answering duration data in the learning information, and calculate a knowledge point mastery level index of the target student based on the student ability portrait; A test question matching module, configured to input the student ability portrait into a pre-constructed test question recommendation model to obtain a recommended test question group for the target student, where the recommended test question group includes multiple recommended test questions that match the student ability portrait; A report generation module, configured to select multiple target recommended test questions corresponding to the knowledge point mastery level index from each of the recommended test questions, and generate a target test paper based on each of the target recommended test questions; obtain the answering data of the target student for the target test paper, and generate a practical training teaching feedback report for the target student based on the answering data.

[0007] In a third aspect of the present application, an electronic device is provided, including a memory, a processor, and a program stored on the memory and executable on the processor, and the program can be loaded and executed by the processor to implement a practical training teaching method based on a large model.

[0008] In a fourth aspect of the present application, a computer-readable storage medium is provided, where the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor is caused to implement a practical training teaching method based on a large model.

[0009] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: By adopting the above technical solutions, the student ability portrait constructed based on the historical homework completion data, wrong question data, and answering duration data of the target student can more accurately depict the target student's mastery of different knowledge points, and on this basis, calculate the knowledge point mastery level index of the target student, providing a personalized learning portrait and ability assessment for learners; relying on the recommended test question group obtained by inputting the student ability portrait into the pre-constructed test question recommendation model, it can effectively screen out recommended test questions with a relatively high matching degree to the target student's knowledge structure, and generate a target test paper by selecting target recommended test questions corresponding to the knowledge point mastery level index of the target student from these recommended test questions, so that the difficulty and content of the questions better match the current learning ability and weak links of the target student, thereby improving the answering efficiency and learning effect; after obtaining the answering data of the target student for the target test paper, generate a practical training teaching feedback report, providing a basis for teachers and students to timely master the learning progress and teaching effect, thereby improving the efficiency of practical training teaching. Description of the Drawings

[0010] Figure 1It is a schematic flowchart of a training teaching method based on a large model provided by an embodiment of the present application; Figure 2 It is a schematic structural diagram of a training teaching system based on a large model provided by an embodiment of the present application; Figure 3 It is a schematic structural diagram of an electronic device provided by an embodiment of the present application.

[0011] Explanation of reference numerals: 300, electronic device; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. Detailed implementation manners

[0012] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments.

[0013] In the description of the embodiments of the present application, words such as "for example" or "for illustration" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as "for example" or "for illustration" in the embodiments of the present application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly speaking, the use of words such as "for example" or "for illustration" is intended to present relevant concepts in a specific manner.

[0014] In the description of the embodiments of the present application, the meaning of the term "plurality" refers to two or more. For example, a plurality of systems refers to two or more systems, and a plurality of screen terminals refers to two or more screen terminals. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly indicating the technical features indicated. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. The terms "include", "comprise", "have" and their variants all mean "including but not limited to", unless otherwise particularly emphasized in other ways.

[0015] An embodiment of the present application provides a training teaching method based on a large model. In one embodiment, please refer to Figure 1 , Figure 1 It is a schematic flowchart of a training teaching method based on a large model provided by an embodiment of the present application. This method can be implemented depending on a computer program, which can be integrated in an application or run as an independent tool application. This method can also be implemented depending on a single-chip microcomputer and can also run on a training teaching system based on a large model of the von Neumann architecture. Specifically, this method may include the following steps: Step 101: Obtain learning information of target students.

[0016] Among them, the target students refer to individual students who are currently in school and need to receive practical training. The students have established learning files in the teaching management platform and participated in daily teaching activities on the platform.

[0017] Learning information refers to the set of learning process data of the target student recorded in the teaching management platform, which specifically includes three dimensions: one is the historical homework completion data, which records the submission records and completion status of all homework of the target student within the preset time period; the second is the wrong question data, which includes the error records of the target students in various test questions and their corresponding knowledge points and question type information; the third is the answer time data, which records in detail the time taken by the target students to answer different types of test questions.

[0018] Specifically, it is necessary to first obtain the learning information of the target students, which is the basis and premise for building a personalized teaching plan. The system collects the learning process data of the target students in a preset time period (such as the most recent semester or a school year) through the teaching management platform, mainly including historical homework completion data, wrong question data, and answer time data. Among them, the historical homework completion data records the submission and completion quality of each assignment by the target students; the wrong question data contains the error records of the target students in various test questions, covering different knowledge points and question types; the answer time data records in detail the time required for the target students to answer different types of test questions. These raw data are recorded in real time through the data collection module of the teaching management platform, and after data cleaning and preprocessing, outliers and invalid data are eliminated to form a standardized learning information data set. To ensure the timeliness and reliability of the data, the system will regularly update the learning information, usually weekly or monthly as a cycle for data synchronization and update. Through comprehensive and systematic learning information collection, sufficient data support can be provided for the subsequent construction of an accurate student ability portrait, thereby realizing the accurate recommendation of teaching resources and the planning of personalized learning paths. This method of collecting learning information based on detailed data can objectively reflect the actual learning status and ability level of the target students, laying a solid foundation for realizing intelligent and personalized practical training teaching.

[0019] Step 102: Construct a student ability profile of the target student based on the historical homework completion data, wrong question data, and answer time data in the learning information, and calculate the target student's knowledge point mastery level index based on the student ability profile.

[0020] Among them, the student ability portrait refers to the digital representation of the learning characteristics of the target student through a neural network model, which includes three-dimensional quantitative indicators: one is the homework completion rate extracted from historical homework completion data, used to characterize the persistence and initiative of learning; the second is the wrong-question frequency of each question type obtained through question type mapping, used to characterize the mastery level of different types of questions; the third is the answering efficiency of each question type obtained by comparing with the preset reference answering duration, used to characterize the problem-solving speed and thinking efficiency. These quantitative indicators are processed by the neural network model and finally form a multi-dimensional vector, comprehensively describing the ability characteristics of the target student.

[0021] The knowledge point mastery level index is a comprehensive evaluation index obtained through knowledge graph analysis based on the student ability portrait, used to quantitatively represent the overall knowledge mastery level of the target student.

[0022] Specifically, after obtaining the learning information of the target student, it is necessary to construct a student ability portrait that can accurately reflect the individual characteristics of the student, which is the core link to achieve personalized teaching. In specific implementation, first, extract the homework completion rate of the target student within the preset time period from the historical homework completion data, and this completion rate reflects the learning initiative and persistence of the student; second, perform question type mapping on the wrong-question data to obtain the wrong-question distribution characteristics, and calculate the wrong-question frequency corresponding to different question types. These data reflect the mastery level and weak links of the student in different types of questions; at the same time, count the average answering duration of each question type in the answering duration data, compare it with the preset reference answering duration, and obtain the answering efficiency corresponding to each question type. This index reflects the problem-solving speed and thinking efficiency of the student. The system inputs the characteristic data such as the homework completion rate, question type answering efficiency, and wrong-question frequency into the pre-trained neural network model, and generates the student ability portrait of the target student through model operation.

[0023] On this basis, further calculate the knowledge point mastery level index. First, construct a knowledge graph containing the mapping relationships between multiple knowledge points according to the preset knowledge point database, and this graph reflects the logical associations between different knowledge points; then, based on the student ability portrait, mark the states of the knowledge point nodes in the knowledge graph, divided into two types: mastered and unmastered; then obtain the association strength between the knowledge point nodes, and combine the node state information. Through the matrix operations of the knowledge point state matrix and the knowledge point association matrix, obtain the knowledge point propagation probability, perform iterative calculation until the loss value converges, and finally obtain the mastery probability of the target student for each knowledge point; finally, perform weighted integration on the mastery probabilities of each knowledge point to obtain the overall knowledge point mastery level index.

[0024] This method based on multi-dimensional data analysis and deep learning can comprehensively and accurately depict students' learning characteristics and knowledge mastery, providing a scientific basis for subsequent question recommendation and personalized teaching. By introducing a knowledge graph, not only the mastery of individual knowledge points is considered, but also the relevance between knowledge points is fully taken into account, making the evaluation results more objective and comprehensive.

[0025] Based on the above embodiments, as an optional embodiment, in step 102: According to the historical homework completion data, wrong-question data, and answering duration data in the learning information, construct a student ability portrait of the target student. This step may further include the following steps: Step 201: Extract the homework completion rate of the target student within a preset time period from the historical homework completion data; perform question type mapping on the wrong-question data to obtain the wrong-question distribution characteristics of the target student, and based on the wrong-question distribution characteristics, calculate the wrong-question frequencies corresponding to multiple question types.

[0026] Specifically, first count the total number of homework assignments and the number of actually completed homework of the target student within a preset time period (such as the most recent semester), and calculate the homework completion rate through the ratio of the two. This completion rate reflects the student's learning attitude and persistence. At the same time, analyze the wrong-question data, map each wrong question to the corresponding question type category according to the preset question type classification rules, for example, map the wrong questions to specific question types such as multiple-choice questions, fill-in-the-blank questions, and calculation questions. According to the mapping results, statistically obtain the wrong-question distribution characteristics, that is, the distribution of the number of wrong questions under different question types, and divide the number of wrong questions of each question type by the total number of questions of that question type to obtain the wrong-question frequencies corresponding to each question type. This processing method can accurately reflect the weak links of students in different question types.

[0027] Step 202: Statistically calculate the average answering duration of each question type in the answering duration data, and compare the average answering duration with the preset reference answering duration to obtain the answering efficiency corresponding to each question type.

[0028] Specifically, process the answering duration data. First, sum up all the answering duration data of the target student under each question type and divide by the number of questions of the corresponding question type to obtain the average answering duration of each question type. The reference answering duration for each question type is preset, and these reference durations are standard answering times statistically obtained based on a large number of student sample data. By calculating the ratio of the actual average answering duration to the reference answering duration, the answering efficiency value corresponding to each question type is obtained. When the answering efficiency value is greater than 1, it means that the student's problem-solving speed is faster than the reference level; when the answering efficiency value is less than 1, it means that the problem-solving speed is slower. This calculation method can objectively reflect the student's problem-solving efficiency and thinking speed.

[0029] Step 203: Input the homework completion rate, the answering efficiency corresponding to each question type, and the wrong-question frequency corresponding to each question type into a pre-trained neural network model to generate a student ability portrait of the target student.

[0030] Specifically, a pre-trained neural network model is used to generate the student ability portrait. The input layer of this neural network model contains multiple nodes, which respectively receive the homework completion rate, the answering efficiency values of each question type, and the wrong-question frequency as input features. These feature data are normalized before input to eliminate the dimensional differences between different features. The neural network model transforms the input features into a feature vector with a fixed dimension, that is, the student ability portrait, through non-linear transformations of multiple hidden layers. The model uses a large number of labeled student data samples in the training stage and optimizes the model parameters through the backpropagation algorithm, ensuring the generalization ability of the model. This deep learning-based method can automatically extract the complex relationships between features, generate a more accurate and comprehensive description of the student's ability, and provide a reliable basis for subsequent personalized teaching.

[0031] Based on the above embodiments, as an optional embodiment, in step 102: Based on the student ability portrait, calculating the knowledge mastery level index of the target student, this step may further include the following steps: Step 204: According to a preset knowledge point database, construct a knowledge graph containing the mapping relationships between multiple knowledge points; based on the student ability portrait, mark the knowledge point nodes in the knowledge graph, and the knowledge point nodes include the knowledge point nodes with the mastered node state and the knowledge point nodes with the unmastered node state.

[0032] Specifically, first construct a knowledge graph according to the preset knowledge point database, which contains the basic information of each knowledge point and the logical association relationships between knowledge points. By taking the knowledge points as the nodes in the graph and the association relationships between knowledge points as the edges, a directed graph structure is established. For example, in the mathematics subject, there is an association between the two knowledge points of "quadratic equation of one variable" and "factorization", and their dependency relationship can be represented by a directed edge. Then, based on the generated student ability portrait, analyze the performance characteristics of the student on each knowledge point, and mark the state of each knowledge point node in the knowledge graph accordingly. Specifically, when the student performs well in the questions related to a certain knowledge point, mark the knowledge point node as the mastered state; otherwise, mark it as the unmastered state. This knowledge graph-based representation method can intuitively show the association relationships between knowledge points and the student's knowledge mastery situation.

[0033] Step 205: Obtain the association strength between each knowledge point node, and combine the association strength between each knowledge point node and the node state of each knowledge point node to calculate the mastery probability of each knowledge point by the target student.

[0034] Specifically, the association strength is calculated for each pair of connected knowledge point nodes in the knowledge graph. The association strength can be comprehensively determined by multiple factors such as the question association degree between two knowledge points and the course content association degree corresponding to the knowledge points. Then, a knowledge point state matrix and a knowledge point association matrix are constructed. The state matrix represents the current mastery state of each knowledge point node, and the association matrix represents the association strength between knowledge points. The knowledge point propagation probability is obtained through the product operation of these two matrices, and an iterative calculation method is used to continuously update the propagation probability until the loss value converges below a preset threshold, and finally the mastery probability of each knowledge point for the target student is obtained. This method based on matrix operation considers the associated influence between knowledge points and can more accurately evaluate the student's knowledge mastery degree.

[0035] Based on the above embodiments, as an optional embodiment, in step 205: Combining the association strength between each knowledge point node and the node state of each knowledge point node to calculate the mastery probability of each knowledge point for the target student. This step may further include the following steps: Step 215: Construct a knowledge point state matrix according to the node state of each knowledge point node; Generate a knowledge point association matrix based on the association strength between each knowledge point node.

[0036] Specifically, first, an n×n-dimensional knowledge point state matrix S is constructed according to the marked knowledge point node states, where n is the total number of knowledge points, and each element sij of the matrix represents the influence state of the i-th knowledge point on the j-th knowledge point. When the state of the i-th knowledge point is mastered, all elements sij in the corresponding row take the value of 1, otherwise they take the value of 0. For example, in a mathematical knowledge system containing 5 knowledge points, if the second knowledge point "solving quadratic equations with one unknown" is marked as mastered, then all elements in the second row of the state matrix S are 1. At the same time, an n×n-dimensional knowledge point association matrix R is constructed, and the element rij in the matrix represents the association strength of the i-th knowledge point on the j-th knowledge point, with a value range of [0, 1]. The association strength is calculated by multiple factors such as the course content overlap degree, question association degree, and knowledge application association degree between knowledge points. For example, the association strength between "factorization" and "solving quadratic equations with one unknown" may be 0.8, indicating that they are closely related. This matrix representation method can comprehensively describe the state relationship and association relationship between knowledge points.

[0037] Step 225: Perform matrix operation on the knowledge point state matrix and the knowledge point association matrix to obtain the knowledge point propagation probability.

[0038] Specifically, perform matrix multiplication on the knowledge point status matrix S and the knowledge point association matrix R, P = S × R, to obtain the initial knowledge point propagation probability matrix P. Specifically, for each element pij in matrix P, its calculation formula is pij = Σ(sik × rkj), where k ranges from 1 to n. This means that the propagation probability of the i-th knowledge point to the j-th knowledge point is obtained by accumulating the influences of the i-th knowledge point on the j-th knowledge point via all intermediate knowledge points k. For example, if "factorization" (knowledge point i) affects "root formula" (knowledge point j) through "solving quadratic equations with one unknown" (knowledge point k), then the product of the status value of "factorization" (sik) and the association strength (rkj) between "solving quadratic equations with one unknown" and "root formula" needs to be considered. This matrix operation method takes into account both direct and indirect influences between knowledge points.

[0039] Step 235: Iteratively calculate the knowledge point propagation probability until the loss value corresponding to the knowledge point propagation probability converges, and obtain the mastery probability of each knowledge point for the target student.

[0040] Specifically, use the iterative calculation method to update the knowledge point propagation probability. The iterative formula is Pt+1 = Pt × R, where t represents the number of iterations. After each iteration, calculate the Euclidean distance between the current propagation probability matrix Pt+1 and the result of the previous iteration Pt as the loss value L, that is, L = ||Pt+1 - Pt||. When the loss value L is less than the preset threshold ε (for example, ε = 0.001), it is considered that the propagation probability converges. At this time, the element pij in matrix P is the final mastery probability of the target student for the j-th knowledge point. This iterative calculation takes into account the cumulative effect of knowledge propagation. For example, the mastery status of "linear equation with one unknown" may, through multiple iterations, affect the mastery probability of "root formula" via intermediate knowledge points such as "quadratic equation with one unknown" and "factorization". By setting appropriate iterative termination conditions, the stability and reliability of the calculation results are ensured.

[0041] Step 206: Weightedly integrate the mastery probabilities of each knowledge point to obtain the knowledge point mastery level index.

[0042] Specifically, it is necessary to perform weighted integration on the obtained mastery probabilities of each knowledge point to obtain the final knowledge point mastery level index. First, a weight coefficient is assigned to each knowledge point, and the assignment of weights takes into account factors such as the importance of the knowledge point and its relevance to other knowledge points. Then, the mastery probability of each knowledge point is multiplied by the corresponding weight coefficient and summed to obtain a comprehensive index value between 0 and 1, and this index value is the knowledge point mastery level index. A higher index value indicates that the student has a better mastery of the knowledge system, while a lower index value indicates that the student has obvious deficiencies in knowledge mastery. This way of weighted integration not only considers the independent mastery of each knowledge point but also reflects the importance of different knowledge points through weights, making the evaluation result more practically meaningful.

[0043] Step 103: Input the student ability portrait into the pre-constructed test question recommendation model to obtain a recommended test question set for the target student. The recommended test question set includes multiple recommended test questions that match the student ability portrait.

[0044] Among them, the pre-constructed test question recommendation model refers to a machine learning model constructed based on the collaborative filtering algorithm. The input of this model is the student ability portrait, and the output is a set of test questions that match this ability portrait. A large amount of historical answering data is used for training during the construction stage of this model to optimize the accuracy of similarity calculation.

[0045] The recommended test question set refers to a set of test questions screened by the test question recommendation model. Each test question in this test question set has a high degree of match with the ability portrait of the target student. The recommended test question set generated in this way can accurately match the learning needs of the student and achieve personalized test question recommendation.

[0046] Specifically, after obtaining the student ability portrait of the target student, it is necessary to input the ability portrait into a pre-constructed test question recommendation model to obtain a recommended test question set that matches the current ability level of the student. Specifically, first, based on the historical answering data in the test question bank, a collaborative filtering algorithm is used to construct a test question recommendation model. This model performs test question matching by calculating the similarity between the student ability portrait vector and the test question feature vector. The test question feature vector includes multiple feature dimensions such as the test question difficulty coefficient, knowledge point coverage, and question type category. During the recommendation process, for the input student ability portrait, the cosine similarity is used to calculate its matching degree with the test question feature vector, and the test questions with higher similarity are selected as candidate recommended test questions. At the same time, in order to ensure the diversity and pertinence of the recommended test questions, test questions with different difficulty gradients and different knowledge point coverages are further selected from the candidate test questions to form the final recommended test question set. This test question recommendation method based on the ability portrait can provide personalized test questions with moderate difficulty and reasonable knowledge point distribution for the target student, avoiding both the frustration caused by overly difficult questions and the low learning efficiency caused by overly easy questions, thus achieving the teaching goal of teaching students in accordance with their aptitude.

[0047] Based on the above embodiments, as an optional embodiment, in step 103: Before the step of inputting the student ability portrait into the pre-constructed test question recommendation model, the following steps may further be included: Step 301: Obtain a preset number of historical student data, and each historical student data includes historical student ability portrait data and historical test question answering data corresponding to the historical student ability portrait data.

[0048] Specifically, first, obtain a preset number (for example, 10,000) of historical student data from the historical database of the teaching management platform. Each piece of historical student data includes two parts: one is the historical student ability portrait data, which contains features such as the homework completion rate, the answering efficiency of each question type, and the frequency of wrong questions; the other is the historical test question answering data corresponding to this ability portrait, which records the answering situation of this student on different test questions, including information such as whether the answer is correct and the answering time. The acquisition of these historical data needs to ensure the integrity and representativeness of the data. Therefore, when selecting, student samples with different learning levels, different grades, and different subject backgrounds will be considered to improve the generalization ability of subsequent model training.

[0049] Step 302: Use the historical student ability portrait data as input features and the corresponding historical test question answering data as training labels to construct a training sample set.

[0050] Specifically, preprocess and organize the obtained historical data to construct a training sample set for model training. Specifically, convert the ability portrait data of each historical student into a feature vector with a fixed dimension as the input feature, and each dimension in this feature vector is normalized to eliminate the influence of dimension. At the same time, convert the corresponding historical test question answering data into a standardized label vector as the training label, and this label vector records the performance of students on various types of test questions. For example, the test question answering results can be encoded in binary form, where 1 represents a correct answer and 0 represents a wrong answer, thus forming a complete training sample pair. In this way, a training set containing tens of thousands of training samples is finally constructed to provide sufficient data support for subsequent model training.

[0051] Step 303: Train a preset neural network model with the training sample set until the output result of the preset neural network model converges to obtain a pre-constructed test question recommendation model.

[0052] Specifically, use a preset neural network model structure for model training. This neural network model includes an input layer, multiple hidden layers, and an output layer, where the number of nodes in the input layer is the same as the dimension of the ability portrait features, and the number of nodes in the output layer corresponds to the number of test question categories. During the training process, first randomly divide the training sample set into a training set and a validation set at a ratio of 8:2. Then use the backpropagation algorithm to optimize the model parameters. In each training round, calculate the cross-entropy loss between the model output and the true label, and update the model parameters through the gradient descent method. At the same time, use the validation set to evaluate the model performance. When the change in the validation set loss value for consecutive multiple rounds (such as 10 rounds) is less than a preset threshold (such as 0.001), it is considered that the model converges, and at this time, the final test question recommendation model is obtained. This training method based on deep learning can automatically learn the complex relationship between the ability portrait and test question answering, thus realizing accurate test question recommendation.

[0053] Step 104: Select multiple target recommended test questions corresponding to the knowledge point mastery level index from the recommended test questions, and generate a target test paper based on each target recommended test question.

[0054] Among them, the target recommended test question refers to the final test question selected from the recommended test question group according to the knowledge point mastery level index.

[0055] The target test paper refers to a complete test paper organized and arranged in a standardized format by the selected target recommended test questions.

[0056] Specifically, it is first necessary to select the target recommended questions that are most suitable for the current knowledge level of the target students from the recommended question group. During the specific implementation, the question screening rules are set according to the calculated knowledge point mastery level index, and the questions in the recommended question group are scored and sorted. The scoring rules comprehensively consider multiple factors such as the matching degree between the question difficulty and the knowledge point mastery level index, the knowledge point coverage of the questions, and the question type distribution. For example, when the knowledge point mastery level index is between 0.3 and 0.5, basic questions and highly targeted special training questions are preferentially selected; when the index is between 0.7 and 0.9, more comprehensive improvement questions tend to be selected. On this basis, by setting the quantity ratio of various question types (such as multiple-choice questions accounting for 30%, fill-in-the-blank questions accounting for 20%, and answer questions accounting for 50%), the corresponding number of target recommended questions is selected from the questions with higher scores according to the ratio requirements. Finally, the selected target recommended questions are sorted and arranged according to the principle of from easy to difficult and progressive knowledge points, and a target test paper containing complete information such as question numbers, score distributions, and answering requirements is generated according to the requirements of the standard test paper format. This test paper generation method based on the knowledge point mastery level index can ensure that the test paper difficulty matches the student's ability level, not only ensuring the pertinence and adaptability of the test paper, but also meeting the standard requirements of teaching evaluation.

[0057] Based on the above embodiments, as an alternative embodiment, in step 104: selecting multiple target recommended questions corresponding to the knowledge point mastery level index from each recommended question, this step may further include the following steps: Step 401: Divide the knowledge point mastery level index into multiple level intervals and set the corresponding question difficulty ranges for each level interval.

[0058] Specifically, the knowledge point mastery level index (with a value range of 0 to 1) is divided into multiple continuous level intervals. For example, the range from 0 to 1 is divided into four intervals: [0, 0.3), [0.3, 0.5), [0.5, 0.7), [0.7, 1]. For each level interval, the corresponding question difficulty range is set, and the difficulty coefficient also uses a standardized value from 0 to 1. Specifically, the question difficulty range corresponding to the [0, 0.3) interval is [0.1, 0.3], mainly basic questions; the difficulty range corresponding to the [0.3, 0.5) interval is [0.2, 0.5], which are basic reinforcement questions; the difficulty range corresponding to the [0.5, 0.7) interval is [0.4, 0.7], which are medium-difficulty questions; the difficulty range corresponding to the [0.7, 1] interval is [0.6, 0.9], which are relatively difficult question types. This interval division and difficulty matching method ensure the adaptability of the question difficulty to the student's ability level.

[0059] Step 402: Obtain the difficulty coefficients of each recommended question; group each recommended question according to the corresponding difficulty coefficient to obtain multiple subsets of questions with different difficulty levels.

[0060] Specifically, first obtain the difficulty coefficient of each question in the recommended question group. The difficulty coefficient is a standardized value calculated comprehensively based on factors such as the knowledge point coverage range of the question, the complexity of the problem-solving steps, and the historical correct rate. Then, divide the questions into different subsets of questions according to their difficulty coefficients. For example, questions with a difficulty coefficient in the range of [0.1, 0.3] are classified into the basic question subset, questions with a difficulty coefficient in the range of [0.2, 0.5] are classified into the basic reinforcement question subset, questions with a difficulty coefficient in the range of [0.4, 0.7] are classified into the medium-difficulty question subset, and questions with a difficulty coefficient in the range of [0.6, 0.9] are classified into the more difficult question subset. Through this grouping method, a question bank structure with distinct difficulty levels is formed.

[0061] Step 403: According to the level interval to which the knowledge point mastery level index belongs, randomly select multiple recommended questions as target recommended questions from the subset of questions corresponding to the corresponding difficulty level according to the preset question type distribution ratio.

[0062] Specifically, determine the level interval to which it belongs according to the knowledge point mastery level index of the target student, and select the target recommended questions from the subset of questions corresponding to this interval. The selection process follows the preset question type distribution ratio. For example, the quantity ratio of multiple-choice questions, fill-in-the-blank questions, and solution questions is 3:2:5. On the premise of meeting the question type ratio requirements, randomly select questions from the subset of questions corresponding to the corresponding difficulty level. For example, when the knowledge point mastery level index of the student is 0.45, randomly select questions that meet the question type ratio requirements from the basic reinforcement question subset as the target recommended questions. This method of selecting questions based on the level interval and difficulty level not only ensures the adaptability of the question difficulty but also ensures the diversity of question selection through random selection, and can provide personalized question combinations for students with different ability levels.

[0063] Step 105: Obtain the answering data of the target student for the target test paper, and generate a practical training teaching feedback report for the target student based on the answering data.

[0064] Among them, the answering data refers to all the data information generated by the target student during the process of completing the target test paper recorded by the system. Specifically, it can include the score records of each test question, that is, the actual score obtained in the total score; the answering duration data, which records the time used for each test question from the start to the submission; the problem-solving process data, including key information such as the solution steps and calculation process of subjective questions; the error type annotation, which classifies the wrong answers, such as conceptual understanding errors, calculation errors, unclear examination of questions, etc.; and the answering efficiency of question types, which measures the problem-solving efficiency by comparing the actual time used with the expected completion time. These answering data are stored in a structured form for subsequent analysis and processing.

[0065] The training teaching feedback report refers to the teaching evaluation document automatically generated by the system based on the answering data. This structured feedback report can objectively reflect the learning status of students, provide a basis for teachers to formulate personalized teaching plans, and at the same time help students clarify their own learning directions.

[0066] Specifically, when obtaining the answering data of the target test paper, the system will record multi-dimensional information of the target student during the answering process, including the score situation, answering duration, problem-solving steps and error types of each test question. For objective questions (such as multiple-choice questions and fill-in-the-blank questions), the system directly records whether the answer is correct or not; for subjective questions (such as answer questions), the answering process is scored in detail through a preset scoring rule, and at the same time, the score distribution of key steps is recorded. In addition, the system will also count the answering duration of each test question and compare it with the expected completion time to evaluate the problem-solving efficiency of students. Based on these detailed answering data, the system generates a comprehensive training teaching feedback report. The report first shows the overall performance, including the total score of the test paper, the scoring rates of each question type and the score distribution of knowledge points; secondly, it analyzes specific weak links, such as identifying frequently wrong knowledge points through error analysis of wrong questions, or finding question types with low problem-solving efficiency through answering duration analysis; finally, it gives targeted improvement suggestions, such as recommending review materials for specific knowledge points, or suggesting adjusting the practice intensity of a certain type of question. This feedback report based on multi-dimensional answering data can not only accurately reflect the learning status of students, but also provide data support for subsequent personalized teaching and effectively improve the teaching effect.

[0067] Based on the above embodiments, as an optional embodiment, in step 105: generating a training teaching feedback report for the target student based on the answering data, this step may further include the following steps: Step 501: Based on the answering data, calculate the first correct rate of the overall target test paper and the second correct rate corresponding to each knowledge point in the target test paper; compare the first correct rate with the preset reference correct rate to obtain the evaluation result of the learning effect of the target student.

[0068] Specifically, the system first calculates two types of accuracy indicators based on the answer data: the first accuracy refers to the overall accuracy of the target test paper, which is calculated by dividing the total score of the target student in the entire test paper by the full score of the test paper; the second accuracy refers to the specific accuracy of each knowledge point, which is calculated by counting the scores of all test questions involving the knowledge point. For example, when the full score of the test paper is 100 points and the student scores 85 points, the first accuracy is 0.85; when the total score of the test questions corresponding to a certain knowledge point is 30 points and the score of this part is 24 points, the second accuracy of the knowledge point is 0.8. The system compares the calculated first accuracy with the preset reference accuracy (for example, 0.75). When the first accuracy is higher than the reference accuracy, the evaluation result is "good"; when it is lower than the reference accuracy, the evaluation result is "need to be improved". This accuracy-based evaluation method can objectively reflect the students' overall learning effect and the mastery of specific knowledge points.

[0069] Step 502: Acquire incorrectly answered questions in the answer data, and identify the error types of the incorrectly answered questions based on a preset error type library to obtain error type identification results.

[0070] Specifically, the system analyzes and processes the incorrectly answered questions in the answer data. First, all test questions with scores below full marks are extracted from the answer data, and then the answers to these test questions are classified and identified based on the preset error type library. The error type library contains a variety of common error types, such as misunderstanding of concepts, calculation errors, incorrect problem-solving ideas, unclear questions, etc. The system classifies each incorrectly answered question into the corresponding error type by matching error features. For example, when a key concept is used incorrectly in a question, the system identifies it as a conceptual misunderstanding error; when there is an obvious error in the calculation process, it is identified as a calculation error. This error type identification can help accurately locate students' learning problems and lay the foundation for providing targeted improvement suggestions in the future.

[0071] Step 503: According to the learning effect evaluation results and the error type identification results, the corresponding learning suggestions are matched in the preset learning suggestion library; the first accuracy rate, each second accuracy rate, and the learning suggestions are integrated into a practical teaching feedback report for the target students.

[0072] Specifically, the system generates a complete practical teaching feedback report based on the foregoing analysis results. First, according to the learning effect evaluation results and error type identification results, the system matches the corresponding recommended content in the pre-set learning recommendation library. The learning recommendation library contains recommended solutions for different learning effects and error types, such as knowledge point review recommendations for conceptual understanding errors, practice strengthening recommendations for calculation errors, etc. Then, the system integrates the first correct rate, the second correct rate of each knowledge point, the error type analysis results, and the matched learning recommendations into a final practical teaching feedback report according to a standardized report template. This report not only contains quantitative evaluation data but also provides specific improvement directions, which can provide clear learning guidance for students and also provide a reference basis for teachers to adjust teaching strategies. This structured feedback report generation method effectively improves the accuracy and practicality of teaching feedback.

[0073] Referring to Figure 2 , a practical teaching system based on a large model provided by an embodiment of the present application, the system includes: an information acquisition module, an index determination module, a question matching module, and a report generation module, where: The information acquisition module is used to acquire the learning information of the target student; The index determination module is used to construct a student ability portrait of the target student according to the historical homework completion data, wrong question data, and answering duration data in the learning information, and calculate the knowledge point mastery level index of the target student based on the student ability portrait; The question matching module is used to input the student ability portrait into a pre-constructed question recommendation model to obtain a recommended question group for the target student, and the recommended question group includes multiple recommended questions that match the student ability portrait; The report generation module is used to select multiple target recommended questions corresponding to the knowledge point mastery level index from each recommended question, and generate a target test paper according to each target recommended question; obtain the answering data of the target student for the target test paper, and generate a practical teaching feedback report for the target student based on the answering data.

[0074] On the basis of the above embodiment, the index determination module is further used to extract the homework completion rate of the target student within a preset time period from the historical homework completion data; perform question type mapping on the wrong question data to obtain the wrong question distribution characteristics of the target student, and calculate the wrong question frequency corresponding to multiple question types based on the wrong question distribution characteristics; count the average answering duration of each question type in the answering duration data, and compare the average answering duration with the preset reference answering duration to obtain the answering efficiency corresponding to each question type; input the homework completion rate, the answering efficiency corresponding to each question type, and the wrong question frequency corresponding to each question type into a pre-trained neural network model to generate a student ability portrait of the target student.

[0075] Based on the above embodiments, the index determination module is further configured to construct a knowledge graph including mapping relationships between multiple knowledge points according to a preset knowledge point database; mark knowledge point nodes in the knowledge graph based on the student ability portrait, where the knowledge point nodes include knowledge point nodes with a mastered node status and knowledge point nodes with an unmastered node status; obtain the association strength between each knowledge point node, and calculate the mastery probability of each knowledge point for the target student by combining the association strength between each knowledge point node and the node status of each knowledge point node; perform weighted integration on the mastery probabilities of each knowledge point to obtain the knowledge point mastery level index.

[0076] Based on the above embodiments, the index determination module is further configured to construct a knowledge point status matrix according to the node status of each knowledge point node; generate a knowledge point association matrix based on the association strength between each knowledge point node; perform matrix operations on the knowledge point status matrix and the knowledge point association matrix to obtain the knowledge point propagation probability; perform iterative calculations on the knowledge point propagation probability until the loss value corresponding to the knowledge point propagation probability converges, to obtain the mastery probability of each knowledge point for the target student.

[0077] Based on the above embodiments, the question matching module is further configured to obtain a preset number of historical student data, where each historical student data includes historical student ability portrait data and corresponding historical question answering data of the historical student ability portrait data; use the historical student ability portrait data as input features and the corresponding historical question answering data as training labels to construct a training sample set; train a preset neural network model through the training sample set until the output result of the preset neural network model converges, to obtain a pre-constructed question recommendation model.

[0078] Based on the above embodiments, the question matching module is further configured to divide the knowledge point mastery level index into multiple level intervals, and set corresponding question difficulty ranges for each level interval; obtain the difficulty coefficients of each recommended question; group each recommended question according to the corresponding difficulty coefficient to obtain multiple question subsets with different difficulty levels; randomly select multiple recommended questions as target recommended questions from the question subsets of the corresponding difficulty level according to the level interval to which the knowledge point mastery level index belongs, according to a preset question type distribution ratio.

[0079] Based on the above embodiments, the report generation module is further configured to calculate the first correct rate of the overall target test paper and the second correct rate corresponding to each knowledge point in the target test paper based on the answer data; compare the first correct rate with a preset reference correct rate to obtain the learning effect evaluation result of the target student; obtain the wrongly answered questions in the answer data, and identify the error types of the wrongly answered questions based on a preset error type library to obtain the error type identification result; match the corresponding learning suggestions in a preset learning suggestion library according to the learning effect evaluation result and the error type identification result; integrate the first correct rate, each second correct rate, and the learning suggestions into a practical training teaching feedback report for the target student.

[0080] It should be noted that when the device provided in the above embodiment realizes its functions, only the division of the above function modules is used for illustration. In actual application, the above functions can be allocated to different function modules according to needs, that is, the internal structure of the device is divided into different function modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be seen in the method embodiment, which will not be repeated here.

[0081] This application also discloses an electronic device. Refer to Figure 3 , Figure 3 which is a schematic structural diagram of an electronic device disclosed in an embodiment of this application. The electronic device 300 may include: at least one processor 301, at least one network interface 304, a user interface 303, a memory 305, and at least one communication bus 302.

[0082] Among them, the communication bus 302 is used to realize the connection and communication between these components.

[0083] Among them, the user interface 303 may include a display (Display) interface and a camera (Camera) interface. Optionally, the user interface 303 may further include a standard wired interface and a wireless interface.

[0084] Among them, the network interface 304 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface).

[0085] Among them, the processor 301 may include one or more processing cores. The processor 301 connects various parts within the entire server through various interfaces and lines. By running or executing instructions, programs, code sets, or instruction sets stored in the memory 305, and by calling the data stored in the memory 305, it executes various functions of the server and processes data. Optionally, the processor 301 may be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), or programmable logic array (PLA). The processor 301 may integrate a combination of one or more of a central processing unit (CPU), a graphics processing unit (GPU), and a modem, etc. Among them, the CPU mainly processes the operating system, user interface graphics, and application programs, etc.; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communication. It can be understood that the above-mentioned modem may not be integrated into the processor 301 and may be implemented separately through a single chip.

[0086] Among them, the memory 305 may include random access memory (RAM) and may also include read-only memory. Optionally, the memory 305 includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area. Among them, the program storage area may store instructions for implementing the operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store the data involved in the above-mentioned various method embodiments. Optionally, the memory 305 may also be at least one storage device located far from the aforementioned processor 301. Refer to Figure 3 , in the memory 305 as a computer storage medium, there may be included an operating system, a network communication module, a user interface module, and an application program of a training and teaching method based on a large model.

[0087] In Figure 3In the electronic device 300 shown, the user interface 303 is mainly used to provide an interface for the user to input and obtain the data input by the user; and the processor 301 can be used to call an application program stored in the memory 305 for a training and teaching method based on a large model. When executed by one or more processors 301, the electronic device 300 is caused to execute the method of one or more of the above embodiments. It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0088] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0089] In several implementation manners provided by this application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some service interfaces. The indirect couplings or communication connections of the devices or units can be in electrical or other forms.

[0090] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0091] In addition, in each embodiment of this application, the functional units can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0092] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present application. The aforementioned memory includes: various media such as USB flash drives, mobile hard disks, magnetic disks, or optical discs that can store program codes.

[0093] The above are only exemplary embodiments of the present disclosure, and the scope of the present disclosure cannot be limited thereby. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure still fall within the scope covered by the present disclosure. Those skilled in the art will readily think of other implementation manners of the present disclosure after considering the specification and the practice of the disclosure.

[0094] The present application aims to cover any variations, uses, or adaptive changes of the present disclosure, which follow the general principles of the present disclosure and include common general knowledge or conventional technical means in the technical field not recorded in the present disclosure. The specification and the embodiments are only regarded as exemplary.

Claims

1. A practical teaching method based on a large model, characterized in that: include: Obtain learning information of target students; Constructing a student ability profile of the target student according to the historical homework completion data, wrong question data, and answering time data in the learning information, and calculating the knowledge point mastery level index of the target student based on the student ability profile; Inputting the student ability profile into a pre-built test question recommendation model to obtain a recommended test question group for the target student, wherein the recommended test question group includes a plurality of recommended test questions that match the student ability profile; Selecting a plurality of target recommended test questions corresponding to the knowledge point mastering level index from each of the recommended test questions, and generating a target test paper based on each of the target recommended test questions; The target student's answer data on the target test paper is obtained, and based on the answer data, a practical training teaching feedback report for the target student is generated.

2. The large model-based practical teaching method according to claim 1 is characterized in that: The step of constructing a student ability profile of the target student based on the historical homework completion data, wrong question data, and answer time data in the learning information includes: Extracting the homework completion rate of the target student within a preset time period from the historical homework completion data; Mapping the wrong question data to question types to obtain the wrong question distribution characteristics of the target student, and calculating wrong question frequencies corresponding to multiple question types based on the wrong question distribution characteristics; Counting the average answering time of each question type in the answering time data, and comparing the average answering time with a preset reference answering time to obtain the answering efficiency corresponding to each question type; The homework completion rate, the answer efficiency corresponding to each question type and the frequency of wrong questions corresponding to each question type are input into a pre-trained neural network model to generate a student ability profile of the target student.

3. The practical teaching method based on a large model according to claim 1 is characterized in that: The step of calculating the target student's knowledge point mastery level index based on the student ability portrait includes: According to the preset knowledge point database, a knowledge graph containing mapping relationships between multiple knowledge points is constructed; Based on the student ability portrait, mark knowledge point nodes in the knowledge graph, wherein the knowledge point nodes include knowledge point nodes with a node status of mastered and knowledge point nodes with a node status of unmastered; Obtaining the association strength between each of the knowledge point nodes, and combining the association strength between each of the knowledge point nodes and the node status of each of the knowledge point nodes, calculating the probability of the target student mastering each of the knowledge points; The mastery probability of each knowledge point is weighted and integrated to obtain a knowledge point mastery level index.

4. The large model-based practical teaching method according to claim 3 is characterized in that: The calculating the probability of the target student mastering each of the knowledge points by combining the association strength between the knowledge point nodes and the node status of each of the knowledge point nodes includes: Constructing a knowledge point state matrix according to the node state of each of the knowledge point nodes; Based on the association strength between each of the knowledge point nodes, a knowledge point association matrix is ​​generated; Performing matrix operations on the knowledge point state matrix and the knowledge point association matrix to obtain knowledge point propagation probability; The knowledge point propagation probability is iteratively calculated until the loss value corresponding to the knowledge point propagation probability converges, thereby obtaining the probability of the target student mastering each of the knowledge points.

5. The large model-based practical teaching method according to claim 1 is characterized in that: Before inputting the student ability profile into the pre-built test question recommendation model, the method further includes: Acquire a preset number of historical student data, each of which includes historical student ability portrait data and historical test answer data corresponding to the historical student ability portrait data; The history student ability portrait data is used as input features, and the corresponding history test answer data is used as training labels to construct a training sample set; The preset neural network model is trained by using the training sample set until the output result of the preset neural network model converges, thereby obtaining a pre-constructed test question recommendation model.

6. The practical teaching method based on a large model according to claim 1 is characterized in that: The step of selecting a plurality of target recommended test questions corresponding to the knowledge point mastering level index from each of the recommended test questions comprises: Dividing the knowledge point mastering level index into multiple level intervals, and setting a corresponding test question difficulty range for each level interval; Obtaining the difficulty coefficient of each of the recommended test questions; Grouping the recommended test questions according to the corresponding difficulty coefficients to obtain multiple test question subsets of different difficulty levels; According to the level interval to which the knowledge point mastering level index belongs, a plurality of recommended test questions are randomly selected from a test question subset of the corresponding difficulty level according to a preset question type distribution ratio as target recommended test questions.

7. The large model-based practical teaching method according to claim 1 is characterized in that: Generating a practical teaching feedback report for the target student based on the answer data includes: Based on the answer data, calculating a first accuracy rate of the target test paper as a whole and a second accuracy rate corresponding to each knowledge point in the target test paper; Comparing the first accuracy rate with a preset reference accuracy rate to obtain a learning effect evaluation result of the target student; Acquire incorrectly answered questions in the answer data, and identify the error types of the incorrectly answered questions based on a preset error type library to obtain error type identification results; According to the learning effect evaluation result and the error type identification result, matching corresponding learning suggestions in a preset learning suggestion library; The first accuracy rate, each of the second accuracy rates, and the learning suggestions are integrated into a practical teaching feedback report for the target student.

8. A practical training teaching system based on a large model, characterized in that: The system comprises: Information acquisition module, used to obtain learning information of target students; An index determination module, for constructing a student ability profile of the target student according to the historical homework completion data, wrong question data, and answering time data in the learning information, and calculating the knowledge point mastery level index of the target student based on the student ability profile; A test question matching module, used for inputting the student ability profile into a pre-built test question recommendation model to obtain a recommended test question group for the target student, wherein the recommended test question group includes a plurality of recommended test questions that match the student ability profile; The report generation module is used to select multiple target recommended test questions corresponding to the knowledge point mastery level index from each of the recommended test questions, and generate a target test paper based on each of the target recommended test questions; obtain the target student's answer data on the target test paper, and generate a practical training teaching feedback report for the target student based on the answer data.

9. An electronic device, characterized in that: It includes a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes the large model-based practical training teaching method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed, the large-model-based practical training teaching method as described in any one of claims 1-7 is executed.

Images