Large model-based teacher teaching design generation system

Abstract

The invention discloses a teacher teaching design generation system based on a large model, which belongs to the technical field of teaching management and comprises a student analysis module, a class analysis module and a teaching design module. The student analysis module is used for carrying out real-time classification analysis on students in the whole school and determining learning classifications corresponding to the students; the class analysis module is used for carrying out teaching analysis on students of each class and determining teaching condition data of each subject of the corresponding class; the teaching design module is used for generating teaching design meeting the teaching requirements of teachers and identifying the teaching requirements of the teachers; matching corresponding teaching condition data according to teaching requirements of teachers; identifying teaching subject information corresponding to the teaching demand of the teacher; the teaching subject information and the teaching condition data are analyzed through a preset teaching large model, a plurality of teaching designs to be selected are obtained, the teaching designs to be selected are displayed to a teacher, and the teacher selects the corresponding teaching design to be selected as the teaching design of the application.

Description

Technical Field

[0001] The present invention belongs to the technical field of teaching management, and in particular is a teacher teaching design generation system based on a large model. Background Art

[0002] In the field of education, the quality of teachers' teaching design directly affects students' learning outcomes and teaching quality. Traditional teaching design often relies on teachers' personal experience and intuition, lacking scientificity and systematicness. With the rapid development of technologies such as big data and artificial intelligence, data-driven teaching design has gradually become possible. In particular, the teacher teaching design generation system based on large models can automatically generate teaching designs that meet teaching needs through technologies such as deep learning and natural language processing, thereby greatly improving teaching efficiency and quality.

[0003] However, the existing large-model-based teacher instructional design generation system still has some problems and challenges. The instructional designs generated by the system are often too template-based, lacking in personalization and innovation, and are difficult to meet actual learning needs. In addition, the system does not take into account the teaching style and preferences of teachers, resulting in a deviation between the generated instructional designs and the actual teaching needs of teachers. At the same time, because some systems separate each class for independent analysis, the system needs to occupy a large amount of resources, reducing the attractiveness of schools to use instructional design generation systems.

[0004] Based on this, the present invention provides a teacher teaching design generation system based on a large model. Summary of the invention

[0005] In order to solve the problems existing in the above-mentioned solutions, the present invention provides a teacher teaching design generation system based on a large model.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A teacher teaching design generation system based on a large model, including a student analysis module, a class analysis module, and a teaching design module;

[0008] The student analysis module is used to perform real-time classification analysis on all students in the school and determine the learning categories corresponding to the students.

[0009] Furthermore, the method of real-time classification analysis of all students in the school includes:

[0010] Step SA1: The platform presets corresponding learning feature items according to the school's requirements;

[0011] Step SA2: extracting learning features of the students according to the learning feature items according to each subject, and obtaining the learning features of the students for the corresponding subjects; integrating the learning feature rows of the students into a student feature set of the corresponding subjects;

[0012] Step SA3: classifying the students according to the corresponding subjects according to the student feature set, obtaining the students' learning classifications for the corresponding subjects, and marking the learning classifications with corresponding subject labels; identifying the classification range corresponding to the learning classification, the classification range consisting of the corresponding student feature set;

[0013] Step SA4: identifying the student feature sets of the corresponding students in various subjects in real time, matching the corresponding classification ranges according to the student feature sets, and determining the learning classification of the students in the corresponding subjects according to the classification ranges;

[0014] When no classification range can be matched according to the student feature set, reclassification is performed according to the student feature set to determine the student classification of the student in the corresponding subject, and the classification range corresponding to the learning classification is updated;

[0015] Step SA5: loop step SA4 until a preset stop condition is reached.

[0016] Furthermore, the method of classifying learning according to the student feature set includes:

[0017] The platform sets K benchmark feature sets for each learning classification, K ≥ 1, and K is a positive integer;

[0018] The distance value between the student feature set and the corresponding learning classification is calculated, and the student is assigned to the learning classification with the smallest distance value.

[0019] Furthermore, the method for calculating the distance value between the student feature set and the corresponding learning classification includes:

[0020] Identify K benchmark feature sets corresponding to the learning classification, and calculate the similarity between the learning feature set and the benchmark feature set respectively; mark the obtained similarity as SD i , i represents the corresponding benchmark feature set, i = 1, 2, ..., K;

[0021] Calculate the corresponding distance value according to the distance value formula. The distance value formula is:

[0022]

[0023] Where: PN is the distance value.

[0024] The class analysis module is used to perform teaching analysis on students in each class and determine the teaching condition data for each subject in the corresponding class. The teaching condition data includes the learning feature recommendation interval corresponding to the corresponding learning feature item.

[0025] Furthermore, the methods for conducting teaching analysis on students in each class include:

[0026] Identify the number of students corresponding to each learning classification in the class according to the subject, calculate the classification proportion corresponding to the learning classification under the subject based on the number of students, integrate the classification proportions corresponding to the subject into classification conditions; determine the teaching condition data of the subject based on the classification conditions.

[0027] Furthermore, a teaching detail map is generated according to the teaching condition data of each class, and the teaching detail map is displayed to corresponding management personnel.

[0028] Furthermore, the method for generating a teaching details map according to the teaching condition data of each class includes:

[0029] Setting a class block template; integrating and setting an initial map according to the class block template;

[0030] Real-time identification of the teaching condition data of the class for the corresponding subject and the corresponding classification conditions, wherein the classification conditions are the classification proportions of each learning classification of the class on the corresponding subject; and real-time addition of the teaching condition data of the class for the corresponding subject to the initial map to form a dynamic display map;

[0031] Generate a subject change curve for the class on the corresponding subject according to the classification condition, wherein the horizontal axis of the subject change curve is time and the vertical axis is the subject value;

[0032] The subject change curves of the class in each subject are added to the dynamic display map, and the current dynamic display map is marked as a teaching details map.

[0033] Furthermore, the method for generating the subject change curve includes:

[0034] Set the initial subject value for each subject of the corresponding class, identify the classification condition corresponding to the initial subject value, and mark it as the benchmark classification condition; mark the initial subject value as KM0; identify the classification condition of the corresponding time, and mark it as the dynamic classification condition;

[0035] Identify the classification weights of the corresponding learning classifications in the baseline classification condition and the dynamic classification condition, and mark the classification weights as FL 1j and FL 2j , j represents the corresponding learning category, j = 1, 2, ..., m, m is the number of learning categories corresponding to the class;

[0036] Calculate the account value at the corresponding time according to the account value formula. The account value formula is:

[0037]

[0038] Where: KM is the subject value at the corresponding time; βj represents the proportional coefficient of the corresponding learning classification, and its value range is (0, 1);

[0039] Generate the account change curve of the corresponding account according to the account value at each time.

[0040] Furthermore, the subject change curve is marked with the teaching time label of the corresponding teacher.

[0041] The teaching design module is used to generate a teaching design that meets the teaching needs of teachers, and identify the teaching needs of teachers, wherein the teaching needs of teachers include teacher information and teaching subject information, and the teacher information includes name and teaching subject;

[0042] Matching corresponding teaching condition data according to the teacher's teaching needs; identifying the teaching subject information corresponding to the teacher's teaching needs;

[0043] The platform establishes a large teaching model; the teaching subject information and teaching condition data are analyzed through the teaching model to obtain several candidate teaching designs, which are displayed to the teachers, who then select the corresponding candidate teaching design as the teaching design to be applied.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The present invention realizes intelligent generation of teaching designs that meet the needs of teachers through the mutual cooperation between the student analysis module, the class analysis module and the teaching design module; this not only avoids the limitation of traditional teaching design that over-relies on the teacher's personal experience and intuition, but also improves the scientificity and systematicness of the teaching design, thereby ensuring the steady improvement of teaching quality and learning effect. Simultaneous analysis of multiple classes is achieved, thereby avoiding the drawback of the existing system that each class needs to be split for independent analysis. This not only improves the processing speed and efficiency of the system, but also reduces the system's occupation of resources, further increasing the school's attractiveness to the use of the teaching design generation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 This is a principle block diagram of the present invention. DETAILED DESCRIPTION

[0048] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. 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.

[0049] like Figure 1 As shown, the teacher teaching design generation system based on the big model includes a student analysis module, a class analysis module and a teaching design module;

[0050] The student analysis module is used to perform real-time classification analysis on all students in the school and determine the learning classification corresponding to each student.

[0051] In one embodiment, each student may be analyzed based on the existing technology, and then each student may be classified to obtain each student classification, and a corresponding learning classification may be set according to the learning situation corresponding to the student classification.

[0052] In one embodiment, the method for performing real-time classification analysis on all students in a school includes:

[0053] Step SA1: Preset various learning feature items, such as learning ability, learning attention, etc. The learning ability can be characterized by homework completion, test scores, etc.; the learning attention can be characterized by learning attention to different interesting teaching during class, or students can choose and fill in it by themselves. The specific learning feature items are set according to the school's teaching requirements. For example, in some cases, there may be only one learning feature item corresponding to the learning ability;

[0054] Step SA2: extract learning features of each student according to each subject based on the learning feature items to obtain learning features corresponding to the corresponding learning feature items; for example, the platform party presets the learning features of each learning feature item according to the actual situation of the school, that is, predefines the range of student situations corresponding to the learning features, and then directly identifies and matches; learning features can also be identified based on other existing methods; the learning features of the student are integrated into the student feature set of the student, that is, the student feature set is composed of the learning features corresponding to the student;

[0055] Step SA3: Classify each student according to the student feature set of each student, obtain the learning classification of each student for the corresponding subject, and label the learning classification with the corresponding subject label; identify the classification range corresponding to the learning classification, and the classification range is combined by the student feature set of each student corresponding to the learning classification;

[0056] Step SA4: identifying the student feature set of each student in each subject in real time, matching the corresponding classification range according to the student feature set, and determining the student's learning classification according to the matched classification range;

[0057] When the classification range cannot be matched according to the student feature set, the student feature set is classified according to the classification method in step SA2 to determine the student classification corresponding to the student, and the classification range corresponding to the learning classification is updated;

[0058] Step SA5: Loop step SA4 to achieve real-time update of the learning classification corresponding to each student in the school, and new students can also be quickly classified; the loop can be ended manually, or other conditions for stopping the loop can be preset, such as stopping the analysis when there is no new student feature set, and re-analyzing when there is a new student feature set.

[0059] In one embodiment, the method for performing learning classification according to a student feature set includes:

[0060] The platform party presets the number of learning categories, that is, the platform party sets the corresponding number of learning categories according to the classification requirements of the school, and sets one or more benchmark feature sets for each learning category, that is, the student feature set of representative students belonging to the learning category is used as the benchmark feature set, and one or more can be selected; the number of benchmark feature sets corresponding to the learning category is marked as K, K ≥ 1, and K is a positive integer;

[0061] Calculate the distance between the student feature set and the corresponding learning classification, and assign the corresponding student to the learning classification with the smallest distance value.

[0062] Distance value calculation:

[0063] Identify the K benchmark feature sets corresponding to the learning classification, calculate the similarity between the learning feature set and each benchmark feature set, and use the existing similarity algorithm to calculate; mark the obtained similarity as SD i , i represents the corresponding benchmark feature set, i = 1, 2, ..., K;

[0064] Calculate the corresponding distance value according to the distance value formula. The distance value formula is:

[0065]

[0066] Where: PN is the distance value.

[0067] In one embodiment, students are classified according to the student feature set, and the classification can be performed according to an existing classification method.

[0068] The class analysis module is used to conduct teaching analysis on students in each class and determine the teaching condition data for each subject of the corresponding class. The teaching condition data include learning feature recommendation intervals corresponding to corresponding learning feature items such as teaching difficulty recommendation intervals and teaching interest recommendation intervals; a teaching detail graph is generated based on the teaching condition data of each class, and the teaching detail graph is displayed to corresponding management personnel, so that school management personnel can intuitively understand the actual learning situation of students in each grade and class for each subject.

[0069] In one embodiment, the method for conducting teaching analysis on students in each class includes:

[0070] Identify the learning categories corresponding to each student in the corresponding class, count the proportion of students in each learning category in the class, mark them as classification proportions, and integrate the proportions of each learning category into the classification conditions of the class;

[0071] Determine the teaching condition data based on the classification conditions; that is, after clarifying the classification proportion of each learning classification of the class, the existing analysis method can be used to determine the recommended intervals of teaching difficulty, teaching interest and other related learning feature recommendation intervals.

[0072] Exemplarily, a teaching analysis model is established based on a neural network such as a CNN network or a DNN network, and a corresponding training set is established manually for training. The training set includes input data and output data. The input data is the classification condition, and the output data is the teaching condition data. Specifically, the platform can hire corresponding teaching experts to set up the training set or summarize the existing teaching professional conclusions and other data to form a training set; the teaching analysis model after successful training is used for analysis to obtain the teaching condition data.

[0073] In one embodiment, the method for generating a teaching details map according to the teaching condition data of each class includes:

[0074] Set up corresponding class tile templates for each class in the school. The class tile templates include the class name, recommended intervals for learning feature items, and other additional class information that needs to be supplemented; set up the initial map based on each class tile template, that is, integrate each class tile template into a unified map according to the classification requirements such as class;

[0075] Real-time identification of the teaching condition data of the corresponding subjects of each class and the corresponding classification conditions. The classification conditions are the proportions of each learning classification corresponding to the class, and are calculated based on the proportion of the number of students corresponding to the learning classification to the total number of students in the class. The teaching condition data of the corresponding subjects of the corresponding class are added to the class block template in the initial map in real time to form a dynamic display map, that is, it changes dynamically with the changes in the teaching condition data of each class.

[0076] A subject change curve of the subject in the class is generated according to the classification conditions obtained in real time, with the horizontal axis being time and the vertical axis being the subject value; the subject change curves of each subject in the class are added to the dynamic display graph, and the current dynamic display graph is marked as a teaching details graph.

[0077] Among them, the method for generating the subject change curve includes:

[0078] Set the initial subject value of each subject for the corresponding class. All of them can be 1, or they can be adjusted based on 1 according to the difference in the initial classification conditions of the subject in the class. For example, taking a certain classification condition as the standard, its corresponding initial subject value is 1, and other classification conditions are compared with it, and 1 is increased or decreased according to the degree of excellence; identify the classification condition corresponding to the initial subject value, mark it as the benchmark classification condition; mark the initial subject value as KM0;

[0079] Identify the classification conditions corresponding to the corresponding time and mark them as dynamic classification conditions;

[0080] Identify the classification weights of the corresponding learning classifications in the baseline classification condition and the dynamic classification condition respectively, and mark the obtained classification weights as FL 1j and FL 2j , respectively represent the classification proportions of the corresponding learning categories in the baseline classification condition and the dynamic classification condition, j represents the corresponding learning category, j = 1, 2, ..., m, m is the number of corresponding learning categories in the class;

[0081] Calculate the account value at the corresponding time according to the account value formula. The account value formula is:

[0082]

[0083] Where: KM is the subject value at the corresponding time; βj represents the proportional coefficient of the corresponding learning classification, with a value range of (0, 1) and is set by the platform according to the needs of the school;

[0084] Generate a corresponding subject change curve based on the subject value at each time. The teaching period of the corresponding teacher can be marked in the subject change curve.

[0085] The teaching design module is used to generate teaching designs that meet the teaching needs of teachers, identify the teaching needs of teachers, and the teaching needs of teachers include teacher information, teaching subject information, etc. Teacher information includes name, teaching subject, etc. The teaching subject information is used to indicate what subject the teacher wants to generate teaching design, such as teaching objectives, teaching content, teaching methods, etc.; match the corresponding teaching condition data according to the teaching needs of teachers; identify the teaching subject information corresponding to the teaching needs of teachers;

[0086] The platform establishes a large teaching model to generate several optional teaching designs based on teaching subject information and teaching condition data, which are marked as candidate teaching designs;

[0087] The teaching subject information and teaching condition data are analyzed through the preset teaching model to obtain several candidate teaching designs, which are displayed to teachers, who then select the corresponding candidate teaching design as the applied teaching design.

[0088] In one embodiment, the establishment of the teaching large model can be based on the existing large model establishment method.

[0089] For example, collect a large amount of history teaching subject information, history teaching condition data, and history teaching design cases. Clean and preprocess the data to remove irrelevant information and retain key features. Label and classify the data to provide supervision information for model training.

[0090] Choose the appropriate deep learning model architecture, such as Transformer, BERT, GPT, etc. These models have excellent performance in natural language processing and text generation. According to the task requirements, adjust and optimize the model architecture appropriately.

[0091] Use the preprocessed data to train the model and adjust the model parameters to minimize the loss function. Use a variety of optimization strategies, such as learning rate adjustment, gradient clipping, regularization, etc., to improve the training efficiency and generalization ability of the model. Regularly verify and evaluate the model to ensure that the model performs well on the test set.

[0092] Extract key features from the learning feature recommendation interval, such as teaching difficulty, teaching interest, etc. Combine these features with the teacher's course requirements to form a unified feature representation. Use the feature representation to guide the model to generate a teaching design that meets the requirements.

[0093] Some innovative methods, such as generative adversarial networks (GANs), can also be introduced to generate more diverse and innovative teaching designs.

[0094] The above formulas are all calculated by removing dimensions and taking numerical values. The formula is a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained by simulating a large amount of data.

[0095] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A teacher teaching design generation system based on a large model, characterized by: It includes student analysis module, class analysis module and teaching design module; The student analysis module is used to perform real-time classification analysis on all students in the school to determine the learning categories corresponding to the students; The class analysis module is used to perform teaching analysis on students in each class and determine the teaching condition data of the corresponding class for each subject, wherein the teaching condition data includes the learning feature recommendation interval corresponding to the corresponding learning feature item; The teaching design module is used to generate a teaching design that meets the teaching needs of teachers, and identify the teaching needs of teachers, wherein the teaching needs of teachers include teacher information and teaching subject information, and the teacher information includes name and teaching subject; Matching corresponding teaching condition data according to the teacher's teaching needs; identifying the teaching subject information corresponding to the teacher's teaching needs; The platform establishes a large teaching model; the teaching subject information and teaching condition data are analyzed through the teaching model to obtain several candidate teaching designs, which are displayed to the teachers, who then select the corresponding candidate teaching design as the teaching design to be applied.

2. The teacher teaching design generation system based on the big model according to claim 1 is characterized in that: Methods for real-time classification analysis of all students in the school include: Step SA1: The platform presets corresponding learning feature items according to the school's requirements; Step SA2: extracting learning features of the students according to the learning feature items according to each subject, and obtaining the learning features of the students for the corresponding subjects; integrating the learning feature rows of the students into a student feature set of the corresponding subjects; Step SA3: classifying the students according to the corresponding subjects according to the student feature set, obtaining the students' learning classifications for the corresponding subjects, and marking the learning classifications with corresponding subject labels; identifying the classification range corresponding to the learning classification, the classification range consisting of the corresponding student feature set; Step SA4: identifying the student feature sets of the corresponding students in various subjects in real time, matching the corresponding classification ranges according to the student feature sets, and determining the learning classification of the students in the corresponding subjects according to the classification ranges; When no classification range can be matched according to the student feature set, reclassification is performed according to the student feature set to determine the student classification of the student in the corresponding subject, and the classification range corresponding to the learning classification is updated; Step SA5: loop step SA4 until a preset stop condition is reached.

3. The teacher teaching design generation system based on the big model according to claim 2 is characterized in that: Methods for classifying learning based on student characteristic sets include: The platform sets K benchmark feature sets for each learning classification, K ≥ 1, and K is a positive integer; The distance value between the student feature set and the corresponding learning classification is calculated, and the student is assigned to the learning classification with the smallest distance value.

4. The teacher teaching design generation system based on the big model according to claim 3 is characterized in that: Methods for calculating the distance value between the student feature set and the corresponding learning classification include: Identify K benchmark feature sets corresponding to the learning classification, and calculate the similarity between the learning feature set and the benchmark feature set respectively; mark the obtained similarity as SD i , i represents the corresponding benchmark feature set, i = 1, 2, ..., K; Calculate the corresponding distance value according to the distance value formula. The distance value formula is: Where: PN is the distance value.

5. The teacher teaching design generation system based on the big model according to claim 1 is characterized in that: Methods for conducting teaching analysis on students in each class include: Identify the number of students corresponding to each learning classification in the class according to the subject, calculate the classification proportion corresponding to the learning classification under the subject based on the number of students, integrate the classification proportions corresponding to the subject into classification conditions; determine the teaching condition data of the subject based on the classification conditions.

6. The teacher teaching design generation system based on the big model according to claim 1 is characterized in that: A teaching detail map is generated according to the teaching condition data of each class, and the teaching detail map is displayed to corresponding management personnel.

7. The teacher teaching design generation system based on the big model according to claim 6 is characterized in that: The method of generating a teaching detail map according to the teaching condition data of each class includes: Setting a class block template; integrating and setting an initial map according to the class block template; Real-time identification of the teaching condition data of the class for the corresponding subject and the corresponding classification conditions, wherein the classification conditions are the classification proportions of each learning classification of the class on the corresponding subject; and real-time addition of the teaching condition data of the class for the corresponding subject to the initial map to form a dynamic display map; Generate a subject change curve for the class on the corresponding subject according to the classification condition, wherein the horizontal axis of the subject change curve is time and the vertical axis is the subject value; The subject change curves of the class in each subject are added to the dynamic display map, and the current dynamic display map is marked as a teaching details map.

8. The teacher teaching design generation system based on the big model according to claim 7 is characterized in that: The methods for generating the subject change curve include: Set the initial subject value for each subject of the corresponding class, identify the classification condition corresponding to the initial subject value, and mark it as the benchmark classification condition; mark the initial subject value as KM0; identify the classification condition of the corresponding time, and mark it as the dynamic classification condition; Identify the classification weights of the corresponding learning classifications in the baseline classification condition and the dynamic classification condition, and mark the classification weights as FL 1j and FL 2j , j represents the corresponding learning category, j = 1, 2, ..., m, m is the number of learning categories corresponding to the class; Calculate the account value at the corresponding time according to the account value formula. The account value formula is: Where: KM is the subject value at the corresponding time; βj represents the proportional coefficient of the corresponding learning classification, and its value range is (0, 1); Generate the account change curve of the corresponding account according to the account value at each time.

9. The teacher teaching design generation system based on the big model according to claim 7 is characterized in that: The subject change curve is marked with the corresponding teacher’s teaching time labels.

Images