A method and system for assisting the reform of teaching in secondary and higher vocational education

Abstract

The present application provides a method and system for assisting the teaching reform of secondary and higher vocational schools, which obtain first scheduled course data; obtain time series data of the number of students, search index time series data and skill requirement time series data of the first scheduled course data within a set historical time period; obtain estimated data of the number of students, search index estimated data and skill requirement estimated data; generate second scheduled course data; obtain knowledge points from the second scheduled course data based on prior knowledge, and construct a knowledge graph with knowledge points as nodes and connections between knowledge points as edges; adjust the prior model structure; adjust the posterior model structure; input the knowledge nodes in the syllabus into the model according to the needs of teachers, and obtain the corresponding course structure and the recommended order of knowledge nodes, so as to provide more accurate, personalized and evidence-based courses, meet the needs of different teachers, improve students' learning effects and improve the decision-making quality of course reconstruction.

Description

Technical Field

[0001] The present application relates to the technical field of curriculum development, and in particular to a method and system for assisting secondary and higher vocational education reform. Background Art

[0002] In the existing technology, the reform of secondary and higher vocational education often involves the selection of courses and curriculum reconstruction, etc., and these reforms are usually based on subjective evaluation and experience judgment, and generally have the following shortcomings:

[0003] First, subjective evaluation and experience judgment may have subjective errors. Different teachers may have different understandings and evaluations of the same course, which may lead to a lack of consistency and reliability in course reconstruction decisions.

[0004] Secondly, there is a lack of data support and quantitative analysis. Subjective evaluation and empirical judgment often cannot provide sufficient data support and quantitative analysis, which makes the curriculum reconstruction decision lack of sufficient evidence and data analysis;

[0005] Thirdly, it is difficult to meet personalized learning needs. Students have different interests, abilities and learning styles. Subjective evaluation and experience judgment may not meet the specific needs of different types of students and cannot provide each student with the most suitable learning content and teaching methods.

[0006] Finally, it limits teaching innovation and experimentation. Curriculum reconstruction requires teaching innovation and experimentation in order to discover more effective teaching methods and more innovative course designs. Subjective evaluation and empirical judgment are often based on existing knowledge and assumptions, which may hinder teachers' opportunities to try new methods and technologies.

[0007] In addition, curriculum reconstruction often relies on course selection. Due to the complexity and dynamics of course selection, course selection is a challenging prediction task involving multiple factors and uncertainties. Traditional course selection methods often have poor prediction accuracy in the face of the complexity and dynamics of topic selection requirements, and do not consider the importance and relevance of each demand variable, making it difficult for the final prediction results to strike a balance between accuracy and interpretability, which has a certain impact on course selection, thereby reducing the decision-making quality of curriculum reconstruction.

[0008] Therefore, the purpose of the present invention is to provide a secondary and higher vocational teaching reform auxiliary method and system to solve the problems existing in the prior art. Summary of the invention

[0009] The purpose of this application is to provide a method and system to assist in the reform of secondary and higher vocational education, so as to provide more accurate, personalized and evidence-based courses, meet the needs of different teachers, improve student learning outcomes and improve the quality of decision-making in curriculum reconstruction.

[0010] In a first aspect, the present application embodiment provides a method for assisting the reform of secondary and higher vocational education, comprising the following steps:

[0011] Obtain the basic information of the teacher, and create a user information dataset based on the basic information of the teacher;

[0012] Obtain the course information offered by the teacher, and create a course information dataset based on the courses offered;

[0013] Obtain the course type in the course information data set, match the course type with the teacher in the user information data set, and obtain the first scheduled course data;

[0014] Obtain the time series data of the number of students, search index and skill requirements of the first scheduled course data within a set historical time period;

[0015] After preprocessing the student population time series data, search index time series data, and skill demand time series data, the pre-trained prediction model is input to obtain the student population estimate data, search index estimate data, and skill demand estimate data within a future set time period;

[0016] The prediction model is obtained by training the multi-objective neural network based on the gradient similarity supernet with the historical student population time series data samples, the historical search index time series data samples and the historical skill demand time series data samples;

[0017] Based on the estimated data of the number of students, the estimated data of the search index, and the estimated data of the skill demand, the first scheduled course data is adjusted to generate the second scheduled course data;

[0018] According to prior knowledge, knowledge points are obtained from the second set of course data, and a knowledge graph is constructed with the knowledge points as nodes and the connections between the knowledge points as edges;

[0019] The Transformer model’s attention mechanism and loss function are used to learn the importance of each knowledge node, obtain the prior importance of each knowledge node, and adjust the prior model structure according to each prior importance.

[0020] Obtain the interaction data of each knowledge node, input the interaction data into the model to obtain the posterior importance of each knowledge node, and adjust the posterior model structure according to each posterior importance;

[0021] According to the teacher's needs, the knowledge nodes in the syllabus are input into the model to obtain the corresponding course structure and the recommended order of knowledge nodes.

[0022] Optionally, the method acquires knowledge points from the second course data based on prior knowledge, and constructs an edge set E={e1, e2,…, e m}, where e is the abstract representation of the connection or relationship between knowledge points, and m represents the number of connections between knowledge points;

[0023] Represent each knowledge point as a d-dimensional vector v i , the dimension d is determined by the characteristics of the node;

[0024] According to the connection between knowledge points, the weight w of the corresponding connection is set ij , and w ij =weight(v i ,v j ), w ij Represents node v i and v j The weight of the edge between them;

[0025] Construct edge weight set W = {w ij};

[0026] According to the knowledge graph, the knowledge graph structure G = (V, E, W) is constructed, where V represents the node set, E represents the edge set, and W represents the edge weight set.

[0027] Optionally, the step of inputting the interaction data into the model to obtain the a posteriori importance of each knowledge node, and adjusting the a posteriori model structure according to each a posteriori importance, includes:

[0028] Obtain the interaction data of each knowledge node and pre-process the interaction data;

[0029] The preprocessed interaction data are input into the prior model;

[0030] Use the softmax function in the Transformer model to output the posterior importance of each knowledge node.

[0031] Optionally, the knowledge nodes in the syllabus are input into the model according to the teacher's needs to obtain the corresponding course structure and the recommended order of the knowledge nodes, including:

[0032] Create a list List = [P1, P2, ..., P l ], where P is the knowledge node in the syllabus and l is the number of knowledge nodes;

[0033] Extract knowledge nodes from the teacher's needs and record the indexes corresponding to the knowledge nodes in a list, forming a knowledge node list Node=[N1,N2,…,N g ], where g≤l, Node represents the index of the knowledge node, N g When the value is 1, the knowledge node is a required subject. g When the value is 0, the knowledge node is a learnable subject;

[0034] Select the node set V'' corresponding to the knowledge node in the knowledge node list = {v1'', v2'', ..., v g ''}, and select the dimension vector corresponding to each knowledge node in the set to form a node vector P = [w1'', w2'', ..., w g ''] T , and input the node vector P into the encoder of the Transformer model to obtain the corresponding node embedding P''=[w1",w2",…,w g ”];

[0035] Input the node embedding and knowledge node list into the decoder of the Transformer model and get the decoding result S = [s1", s2", ..., s l ”];

[0036] Input the decoding result into the softmax model to obtain the probability of each knowledge node p = [p1", p2", ..., p l ”], where p l " represents the probability of this knowledge node entering the course structure, and the knowledge node corresponding to the knowledge node index with the largest probability is selected as the prior knowledge node.

[0037] Optionally, adjusting the prior model structure according to each prior importance includes:

[0038] Setting the node height of each knowledge node according to the prior importance, and drawing the hierarchy of each knowledge node according to the node height and the connection between the nodes;

[0039] Set the knowledge node border of each knowledge node;

[0040] If the knowledge node is a required subject, the knowledge node border of the knowledge node is set to be a solid line and bold; if the knowledge node is a learnable subject, the knowledge node border of the knowledge node is set to be a dotted line.

[0041] Optionally, the method further comprises:

[0042] The first scheduled course data is adjusted based on the estimated data of the number of students, the estimated data of the search index and the estimated data of the skill demand to obtain an adjustment factor;

[0043] Adjust the prior model structure according to each prior importance to generate a first adjustment feature;

[0044] Adjust the posterior model structure according to each posterior importance to generate a second adjustment feature;

[0045] Processing the adjustment factor based on the first adjustment feature and the second adjustment feature;

[0046] Determine whether to generate the second fixed course data based on the processing result.

[0047] Optionally, the Transformer model includes an encoder and a decoder, the structure of the encoder is composed of n hierarchical layers, each hierarchical layer includes a self-attention module SA, a first residual sublayer and a first feedforward network sublayer; the structure of the decoder is composed of n hierarchical layers, each hierarchical layer includes a masked multi-head attention layer MHSA, a second residual sublayer and a second feedforward network sublayer.

[0048] Optionally, the multi-objective neural network based on gradient similarity supernet includes an input layer, a shared feature extraction layer, a supernet layer, a task-specific layer, an attention mechanism layer, a fusion layer and an output layer.

[0049] Optionally, the method described above pre-trains the prediction model in the following manner:

[0050] Obtain the historical student number time series data samples, historical search index time series data samples and historical skill demand time series data samples of the first scheduled course data;

[0051] After preprocessing the historical student population time series data samples, the historical search index time series data samples, and the historical skill demand time series data samples, the historical student population time series data samples, the historical search index time series data samples, and the historical skill demand time series data samples are converted into supervised data of student population, supervised data of search index, and supervised data of skill demand, respectively;

[0052] The supervised data of student numbers, search index and skill requirements are input into a multi-objective neural network based on gradient similarity hypernet for training, and the prediction model is obtained after reaching the preset training cutoff condition.

[0053] In a second aspect, the embodiment of the present application provides a secondary and higher vocational teaching reform auxiliary system, including:

[0054] The first acquisition module is used to acquire the basic information of the teacher and create a user information data set based on the basic information of the teacher;

[0055] The second acquisition module is used to obtain the course information offered by the teacher and create a course information data set according to the offered courses;

[0056] The third acquisition module is used to obtain the course type in the course information data set, match the course type with the teacher in the user information data set, and obtain the first scheduled course data;

[0057] A fourth acquisition module is used to obtain the time series data of the number of students, the time series data of the search index and the time series data of the skill requirements of the first scheduled course data within a set historical time period;

[0058] The prediction module is used to pre-process the student population time series data, search index time series data and skill demand time series data, and then input the pre-trained prediction model to obtain the student population estimation data, search index estimation data and skill demand estimation data within a future set time period;

[0059] The prediction model is obtained by training the multi-objective neural network based on the gradient similarity supernet with the historical student population time series data samples, the historical search index time series data samples and the historical skill demand time series data samples;

[0060] A generation module, used for adjusting the first course scheduled data based on the student number estimated data, the search index estimated data and the skill demand estimated data, to generate the second course scheduled data;

[0061] A construction module, used to obtain knowledge points from the second fixed course data according to prior knowledge, and to construct a knowledge graph with the knowledge points as nodes and the connections between the knowledge points as edges;

[0062] The adjustment module is used to learn the importance of each knowledge node using the attention mechanism and loss function of the Transformer model, obtain the prior importance of each knowledge node, and adjust the prior model structure according to each prior importance;

[0063] A fifth acquisition module is used to acquire the interaction data of each knowledge node, input the interaction data into the model to obtain the posterior importance of each knowledge node, and adjust the posterior model structure according to each posterior importance;

[0064] The recommendation module is used to input the knowledge nodes in the syllabus into the model according to the teacher's needs to obtain the corresponding course structure and the recommended order of knowledge nodes.

[0065] Compared with the prior art, the beneficial effects of the present application are as follows: the present application provides an auxiliary method for the teaching reform of secondary and higher vocational schools, which obtains the basic information of the teachers, creates a user information data set based on the basic information of the teachers; obtains the course information offered by the teachers, creates a course information data set based on the offered courses; obtains the course types in the course information data set, matches the course types with the teachers in the user information data set, and obtains the first scheduled course data; obtains the time series data of the number of students, the search index time series data, and the skill requirement time series data of the first scheduled course data within a set historical time period; After preprocessing the data, search index time series data and skill demand time series data, the pre-trained prediction model is input to obtain the estimated data of the number of students, the estimated data of the search index and the estimated data of the skill demand in the future set time period; the prediction model is obtained by training a multi-objective neural network based on a gradient similarity supernet through historical student number time series data samples, historical search index time series data samples and historical skill demand time series data samples; based on the estimated data of the number of students, the estimated data of the search index and the estimated data of the skill demand, the first scheduled course data is adjusted to generate the second scheduled course data; according to prior knowledge, from the first Second, obtain knowledge points from fixed course data, and construct a knowledge graph with knowledge points as nodes and the connections between knowledge points as edges; use the attention mechanism and loss function of the Transformer model to learn the importance of each knowledge node, obtain the prior importance of each knowledge node, and adjust the prior model structure according to each prior importance; obtain the interaction data of each knowledge node, input the interaction data into the model to obtain the posterior importance of each knowledge node, and adjust the posterior model structure according to each posterior importance; input the knowledge nodes in the syllabus into the model according to the teacher's needs, and obtain the corresponding course structure and the recommended order of knowledge nodes; Only by constructing the mechanism model of the prior model and the optimization mechanism of the posterior model, the a priori importance and posterior importance of the knowledge nodes can be obtained, and the recommended order of the course structure and knowledge nodes can be obtained based on the needs of teachers. This can provide more accurate, personalized and evidence-based courses to meet the needs of different teachers, and fully consider the importance and relevance of each demand variable, which not only improves the accuracy of data prediction results, but also provides a reliable explanation of the potential mechanism of course selection, thereby improving the decision-making quality of course reconstruction, helping to improve students' learning outcomes and the innovation of teaching, and playing a positive auxiliary effect on the reform of secondary and higher vocational education. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0067] Figure 1 A flowchart of a secondary and higher vocational education reform auxiliary method provided by this application is shown;

[0068] Figure 2 A schematic diagram of a secondary and higher vocational education reform auxiliary system provided by the present application is shown;

[0069] Figure 3 A schematic diagram of an electronic device provided by the present application is shown. DETAILED DESCRIPTION

[0070] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0071] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application should have the common meanings understood by technicians in the field to which this application belongs.

[0072] In addition, the terms "first" and "second" etc. are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or devices.

[0073] Please refer to Figure 1 , Figure 1 This is an auxiliary method for secondary and higher vocational education reform provided by the embodiment of the present application, comprising the following steps:

[0074] S101, obtaining basic information of the teacher, and creating a user information data set according to the basic information of the teacher;

[0075] S102, obtaining information about courses offered by the teacher, and creating a course information data set according to the courses offered;

[0076] S103, obtaining the course type in the course information data set, matching the course type with the teacher in the user information data set, and obtaining the first course data; the first course data may include information such as course ID, course name, course type, teacher name, course content, etc.;

[0077] S104, obtaining the time series data of the number of students, the time series data of the search index, and the time series data of the skill requirements of the first scheduled course data within a set historical time period;

[0078] S105, after pre-processing the student number time series data, search index time series data and skill requirement time series data, input the pre-trained prediction model to obtain the student number estimation data, search index estimation data and skill requirement estimation data in the future set time period;

[0079] The prediction model is obtained by training a multi-objective neural network based on a gradient similarity supernet with historical student population time series data samples, historical search index time series data samples, and historical skill demand time series data samples;

[0080] S106. Based on the estimated data on the number of students, the estimated data on the search index, and the estimated data on the skill requirements, the first scheduled course data is adjusted to generate the second scheduled course data. Specifically, weights may be assigned to the estimated data on the number of students, the estimated data on the search index, and the estimated data on the skill requirements, and the adjustment factor may be calculated by weighted fusion. The first scheduled course data is adjusted according to the adjustment factor. If the adjustment factor is higher than a preset threshold, increasing the course capacity or opening more similar courses may be considered. If the preset threshold is lower than the preset threshold, reducing the course capacity or merging courses may be considered. Based on the adjustment result, the second scheduled course data is generated, which may include information such as the course number, name, capacity, schedule, and course content.

[0081] S107, acquiring knowledge points from the second set course data according to prior knowledge, and constructing a knowledge graph with the knowledge points as nodes and the connections between the knowledge points as edges;

[0082] S108, using the attention mechanism and loss function of the Transformer model to learn the importance of each knowledge node, obtain the prior importance of each knowledge node, and adjust the prior model structure according to each prior importance;

[0083] S109, obtaining the interaction data of each knowledge node, inputting the interaction data into the model to obtain the posterior importance of each knowledge node, and adjusting the posterior model structure according to each posterior importance;

[0084] S110. Input the knowledge nodes in the syllabus into the model according to the teacher's needs to obtain the corresponding course structure and the recommended order of the knowledge nodes.

[0085] In step S107, the knowledge points are obtained from the second course data according to the prior knowledge, and the edge set E = {e1, e2, ..., e m}, where e is the abstract representation of the connection or relationship between knowledge points, and m represents the number of connections between knowledge points;

[0086] Represent each knowledge point as a d-dimensional vector v i , the dimension d is determined by the characteristics of the node;

[0087] According to the connection between knowledge points, the weight w of the corresponding connection is set ij , and w ij =weight(v i ,v j ), w ij Represents node v i and v j The weight of the edge between them;

[0088] Construct edge weight set W = {w ij};

[0089] According to the knowledge graph, the knowledge graph structure G = (V, E, W) is constructed, where V represents the node set, E represents the edge set, and W represents the edge weight set.

[0090] In step S108, the importance of each knowledge node is learned using the attention mechanism and loss function of the Transformer model to obtain the prior importance of each knowledge node, which specifically includes the following steps:

[0091] 1. Initialize node embedding: Initialize an embedding vector for each node in the knowledge graph.

[0092] 2. Input sequence construction: According to the structure of the knowledge graph, construct the input sequence. This can be the order of nodes, paths or subgraphs.

[0093] 3. Attention layer calculation: For each node in the input sequence, the self-attention mechanism is used to calculate its correlation with other nodes. Then, the multi-head attention mechanism is used to learn the relationship between nodes from different subspaces.

[0094] 4. Feedforward network processing: The output of the attention layer is passed to the feedforward network sublayer for further nonlinear transformation.

[0095] 5. Loss function optimization: Calculate the loss between the model’s output and the true label, and use gradient descent or other optimization algorithms to update the model parameters to minimize the loss function.

[0096] 6. Iterative training: Repeat the above steps and train the model through multiple iterations until the model converges.

[0097] 7. Prior importance extraction: After the model training is completed, the attention weights or model outputs are extracted, which can be regarded as the prior importance of the nodes. The attention weights usually reflect the relative importance of the nodes in the graph, while the model outputs can be used for classification or regression tasks, indirectly reflecting the importance of the nodes.

[0098] In this way, the Transformer model can be adjusted and the adjusted prior importance of each knowledge node can be obtained.

[0099] In some implementations, inputting the interaction data into the model to obtain the posterior importance of each knowledge node, and adjusting the posterior model structure according to each posterior importance, includes:

[0100] Obtain the interaction data of each knowledge node and pre-process the interaction data;

[0101] The preprocessed interaction data are input into the prior model;

[0102] Use the softmax function in the Transformer model to output the posterior importance of each knowledge node.

[0103] Optionally, the knowledge nodes in the syllabus are input into the model according to the teacher's needs to obtain the corresponding course structure and the recommended order of the knowledge nodes, including:

[0104] Create a list List = [P1, P2, ..., P l ], where P is the knowledge node in the syllabus and l is the number of knowledge nodes;

[0105] Extract knowledge nodes from the teacher's needs and record the indexes corresponding to the knowledge nodes in a list, forming a knowledge node list Node=[N1,N2,…,N g ], where g≤l, Node represents the index of the knowledge node, N g When the value is 1, the knowledge node is a required subject. g When the value is 0, the knowledge node is a learnable subject;

[0106] Select the node set V'' corresponding to the knowledge node in the knowledge node list = {v1'', v2'', ..., v g ''}, and select the dimension vector corresponding to each knowledge node in the set to form a node vector P = [w1'', w2'', ..., w g ''] T , and input the node vector P into the encoder of the Transformer model to obtain the corresponding node embedding P''=[w1",w2",…,w g ”];

[0107] Input the node embedding and knowledge node list into the decoder of the Transformer model and get the decoding result S = [s1", s2", ..., s l ”];

[0108] Input the decoding result into the softmax model to obtain the probability of each knowledge node p = [p1", p2", ..., p l ”], where p l " represents the probability of this knowledge node entering the course structure, and the knowledge node corresponding to the knowledge node index with the largest probability is selected as the prior knowledge node.

[0109] In some embodiments, adjusting the prior model structure according to each prior importance comprises:

[0110] The node height of each knowledge node is set according to the prior importance, and the hierarchy of each knowledge node is drawn according to the node height and the connection between the nodes; specifically, it can be achieved by a mapping function, which converts the prior importance into the node height, and the mapping function can be linear or nonlinear, depending on how to better represent the relationship between the importance and the height; the node height and the connection between the nodes (for example, dependency, association strength, etc.) are used to draw a hierarchical structure diagram of the knowledge nodes, and the hierarchical structure diagram can be a tree diagram, a directed acyclic graph, or other graphics suitable for representing the hierarchy and dependency relationship;

[0111] Set the knowledge node border of each knowledge node;

[0112] If the knowledge node is a required subject, the knowledge node border of the knowledge node is set to a solid line and bold. If the knowledge node is a learnable subject, the knowledge node border of the knowledge node is set to a dotted line. These steps will help to visually represent the importance of each node in the knowledge graph and the relationship between them, so as to better understand and analyze the knowledge structure.

[0113] In some embodiments, adjusting the posterior model structure according to each posterior importance includes:

[0114] The thickness of the knowledge node border of each of the knowledge nodes is set according to each a posteriori importance; specifically, a mapping function may be defined to convert the a posteriori importance into the thickness of the border; the mapping function may be linear or nonlinear, depending on how to better represent the relationship between the importance and the thickness of the border;

[0115] If the knowledge node is a required subject, the corresponding knowledge node border is set to bold. If the knowledge node is a learnable subject, the corresponding knowledge node border is set to not bold. The above steps provide educators or learners with more intuitive information to facilitate course design and learning path planning.

[0116] In this embodiment, the Transformer model specifically includes an encoder and a decoder. The structure of the encoder is composed of n hierarchical layers, each of which includes a self-attention module SA, a first residual sublayer, and a first feedforward network sublayer; the structure of the decoder is composed of n hierarchical layers, each of which includes a masked multi-head attention layer MHSA, a second residual sublayer, and a second feedforward network sublayer. The above model has the following advantages:

[0117] 1. Being able to understand the contextual relationship between course content through the self-attention mechanism is crucial for identifying dependencies and correlations between courses;

[0118] 2. Ability to effectively process sequence data and help restructure courses to better fit the learning path;

[0119] 3. These multi-source information, such as students’ prior knowledge, course content, and teaching objectives, can be integrated to generate a more comprehensive and coordinated course structure;

[0120] 4. Highly customizable and flexible, model parameters can be adjusted to optimize course design according to specific educational goals and student needs;

[0121] 5. In a large-scale course system, there may be long-distance dependencies between courses. The Transformer model of this application can capture such long-distance dependencies and help maintain the coherence of course content during reconstruction;

[0122] 6. It can speed up the course reconstruction process, especially when dealing with a large amount of course data;

[0123] 7. By analyzing students’ learning history and performance, it can help customize personalized learning paths to improve learning efficiency and satisfaction;

[0124] 8. Educational needs and student groups are constantly changing. The above Transformer model can be used to dynamically adjust the course structure to adapt to these changes.

[0125] In step S105, the multi-objective neural network based on the gradient similarity supernet includes an input layer, a shared feature extraction layer, a supernet layer, a task-specific layer, an attention mechanism layer, a fusion layer and an output layer.

[0126] The input layer is used to receive input data from different tasks; the shared feature extraction layer may include multiple convolutional layers (for image data) or fully connected layers (for non-image data) to extract feature representations shared by all tasks; the supernet layer is used to generate task-specific weights, specifically including multiple fully connected layers or other types of layers, which adjust the generated weights according to the gradient similarity of the input tasks; the task-specific layer uses the weights generated by the supernet, which are fine-tuned for each task and may include convolutional layers, fully connected layers, recurrent layers (for sequence data), etc.; the attention mechanism layer can be used to focus on certain specific parts of the input data; the fusion layer is used to merge features from different sources; the output layer has an output layer for each task, usually a fully connected layer, followed by an activation function suitable for the task type (for example, Sigmoid is used for binary classification and Softmax is used for multi-classification).

[0127] In this embodiment, the method described herein pre-trains the prediction model in the following manner:

[0128] Obtain time series data samples of the historical number of students, historical search index and historical skill demand of the first scheduled course data; specifically, obtain the historical number of students, historical search index and historical skill demand, and statistically generate time series with a resolution of each preset time length to obtain time series data samples of the historical number of students, historical search index and historical skill demand; wherein, the historical number of students is the number of students who choose the first scheduled course data, the historical search index is the search frequency and user interest in choosing the first scheduled course data, and the historical skill demand is the market demand for the skills involved in the first scheduled course data.

[0129] After preprocessing the historical student population time series data samples, the historical search index time series data samples, and the historical skill demand time series data samples, the historical student population time series data samples, the historical search index time series data samples, and the historical skill demand time series data samples are converted into supervised data of student population, supervised data of search index, and supervised data of skill demand, respectively;

[0130] The supervised data of student numbers, search index and skill requirements are input into a multi-objective neural network based on gradient similarity hypernet for training, and the prediction model is obtained after reaching the preset training cutoff condition.

[0131] In this embodiment, the method further includes:

[0132] The first scheduled course data is adjusted based on the estimated data of the number of students, the estimated data of the search index and the estimated data of the skill demand to obtain an adjustment factor;

[0133] According to each prior importance, the prior model structure is adjusted to generate a first adjustment feature; the first adjustment feature may include student background features, course content features, teaching resource features, and social demand features;

[0134] According to each posterior importance, the posterior model structure is adjusted to generate a second adjustment feature; the second adjustment feature may include a learning effectiveness feature, a course feedback feature, a teaching strategy feature, and a resource utilization feature;

[0135] Processing the adjustment factor based on the first adjustment feature and the second adjustment feature;

[0136] Whether to generate the second set of course data is determined based on the processing results. The above method can more effectively respond to market changes and student needs and improve the relevance and quality of courses.

[0137] Compared with the prior art, the beneficial effects of the present application are as follows: the present application provides an auxiliary method for the teaching reform of secondary and higher vocational schools, which obtains the basic information of the teachers, creates a user information data set based on the basic information of the teachers; obtains the course information offered by the teachers, creates a course information data set based on the offered courses; obtains the course types in the course information data set, matches the course types with the teachers in the user information data set, and obtains the first scheduled course data; obtains the time series data of the number of students, the search index time series data, and the skill requirement time series data of the first scheduled course data within a set historical time period; After preprocessing the data, search index time series data and skill demand time series data, the pre-trained prediction model is input to obtain the estimated data of the number of students, the estimated data of the search index and the estimated data of the skill demand in the future set time period; the prediction model is obtained by training a multi-objective neural network based on a gradient similarity supernet through historical student number time series data samples, historical search index time series data samples and historical skill demand time series data samples; based on the estimated data of the number of students, the estimated data of the search index and the estimated data of the skill demand, the first scheduled course data is adjusted to generate the second scheduled course data; according to prior knowledge, from the first Second, obtain knowledge points from fixed course data, and construct a knowledge graph with knowledge points as nodes and the connections between knowledge points as edges; use the attention mechanism and loss function of the Transformer model to learn the importance of each knowledge node, obtain the prior importance of each knowledge node, and adjust the prior model structure according to each prior importance; obtain the interaction data of each knowledge node, input the interaction data into the model to obtain the posterior importance of each knowledge node, and adjust the posterior model structure according to each posterior importance; input the knowledge nodes in the syllabus into the model according to the teacher's needs, and obtain the corresponding course structure and the recommended order of knowledge nodes; Only by constructing the mechanism model of the prior model and the optimization mechanism of the posterior model, the a priori importance and posterior importance of the knowledge nodes can be obtained, and the recommended order of the course structure and knowledge nodes can be obtained based on the needs of teachers. This can provide more accurate, personalized and evidence-based courses to meet the needs of different teachers, and fully consider the importance and relevance of each demand variable, which not only improves the accuracy of data prediction results, but also provides a reliable explanation of the potential mechanism of course selection, thereby improving the decision-making quality of course reconstruction, helping to improve students' learning outcomes and the innovation of teaching, and playing a positive auxiliary effect on the reform of secondary and higher vocational education.

[0138] Please refer to Figure 2 The present application embodiment also provides a secondary and higher vocational teaching reform auxiliary system, including:

[0139] The first acquisition module 101 is used to acquire basic information of the teacher and create a user information data set according to the basic information of the teacher;

[0140] The second acquisition module 102 is used to obtain the course information offered by the teacher and create a course information data set according to the offered courses;

[0141] The third acquisition module 103 is used to obtain the course type in the course information data set, match the course type with the teacher in the user information data set, and obtain the first scheduled course data;

[0142] The fourth acquisition module 104 is used to obtain the time series data of the number of students, the time series data of the search index and the time series data of the skill requirements of the first scheduled course data within a set historical time period;

[0143] The prediction module 105 is used to pre-process the student number time series data, the search index time series data and the skill requirement time series data, and input the pre-trained prediction model to obtain the student number estimation data, the search index estimation data and the skill requirement estimation data within a future set time period;

[0144] The prediction model is obtained by training the multi-objective neural network based on the gradient similarity supernet with the historical student population time series data samples, the historical search index time series data samples and the historical skill demand time series data samples;

[0145] A generating module 106, configured to adjust the first course schedule data based on the student number estimation data, the search index estimation data and the skill demand estimation data, and generate second course schedule data;

[0146] A construction module 107, for acquiring knowledge points from the second course data according to prior knowledge, and constructing a knowledge graph with the knowledge points as nodes and the connections between the knowledge points as edges;

[0147] An adjustment module 108 is used to learn the importance of each knowledge node using the attention mechanism and loss function of the Transformer model, obtain the prior importance of each knowledge node, and adjust the prior model structure according to each prior importance;

[0148] The fifth acquisition module 109 is used to acquire the interaction data of each knowledge node, input the interaction data into the model to obtain the posterior importance of each knowledge node, and adjust the posterior model structure according to each posterior importance;

[0149] The recommendation module 110 is used to input the knowledge nodes in the syllabus into the model according to the teacher's needs to obtain the corresponding course structure and the recommended order of the knowledge nodes.

[0150] The system provided in the embodiment of the present application and the auxiliary method provided in the embodiment of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, operated or implemented therein.

[0151] The embodiment of the present application also provides an electronic device corresponding to the method provided in the above embodiment, and the electronic device can be an electronic device used for a client, such as a mobile phone, a laptop computer, a tablet computer, a desktop computer, etc., to execute the above method.

[0152] Please refer to Figure 3 , which shows a schematic diagram of an electronic device provided by some embodiments of the present application. Figure 3 As shown, the electronic device 20 includes: a processor 200, a memory 201, a bus 202 and a communication interface 203, and the processor 200, the communication interface 203 and the memory 201 are connected via the bus 202; the memory 201 stores a computer program that can be run on the processor 200, and the processor 200 executes the method provided in any of the aforementioned embodiments of the present application when running the computer program.

[0153] The memory 201 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 203 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

[0154] The bus 202 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The memory 201 is used to store programs, and the processor 200 executes the programs after receiving execution instructions. The method disclosed in any implementation of the above-mentioned embodiment of the present application may be applied to the processor 200, or implemented by the processor 200.

[0155] The processor 200 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 200. The above processor 200 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a readily available programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application can be directly embodied as a hardware decoding processor to be executed, or the hardware and software modules in the decoding processor can be executed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 201, and the processor 200 reads the information in the memory 201 and completes the steps of the above method in combination with its hardware.

[0156] The electronic device provided in the embodiment of the present application and the auxiliary method provided in the embodiment of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, operated or implemented therein.

[0157] The present application also provides a computer-readable storage medium corresponding to the auxiliary method provided in the aforementioned embodiment, on which a computer program (i.e., a program product) is stored. When the computer program is executed by a processor, it will execute the auxiliary method provided in any of the aforementioned embodiments.

[0158] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical or magnetic storage media, which are not listed here one by one.

[0159] The computer-readable storage medium provided in the above-mentioned embodiments of the present application and the auxiliary method provided in the embodiments of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0160] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application, and they should all be included in the scope of the claims and specification of the present application.

Claims

1. A method for assisting the reform of secondary and higher vocational education, characterized in that: The following steps are involved: Obtain the basic information of the teacher, and create a user information dataset based on the basic information of the teacher; Obtain the course information offered by the teacher, and create a course information dataset based on the courses offered; Obtain the course type in the course information data set, match the course type with the teacher in the user information data set, and obtain the first scheduled course data; Obtain the time series data of the number of students, search index and skill requirements of the first scheduled course data within a set historical time period; After preprocessing the student population time series data, search index time series data, and skill demand time series data, the pre-trained prediction model is input to obtain the student population estimate data, search index estimate data, and skill demand estimate data within a future set time period; The prediction model is obtained by training the multi-objective neural network based on the gradient similarity supernet with the historical student population time series data samples, the historical search index time series data samples and the historical skill demand time series data samples; Based on the estimated data of the number of students, the estimated data of the search index, and the estimated data of the skill demand, the first scheduled course data is adjusted to generate the second scheduled course data; According to prior knowledge, knowledge points are obtained from the second set of course data, and a knowledge graph is constructed with the knowledge points as nodes and the connections between the knowledge points as edges; The Transformer model’s attention mechanism and loss function are used to learn the importance of each knowledge node, obtain the prior importance of each knowledge node, and adjust the prior model structure according to each prior importance. Obtain the interaction data of each knowledge node, input the interaction data into the prior model structure to obtain the posterior importance of each knowledge node, and adjust the posterior model structure according to each posterior importance; According to the teacher's needs, the knowledge nodes in the syllabus are input into the posterior model structure to obtain the corresponding course structure and the recommended order of knowledge nodes.

2. The method according to claim 1, characterized in that The method of acquiring knowledge points from the second fixed course data based on prior knowledge and constructing a knowledge graph with the knowledge points as nodes and the connections between the knowledge points as edges includes: According to the prior knowledge, we construct a node set V = {v1, v2, ..., v n }, where n represents the number of knowledge nodes; The edge set E = {e1, e2, ..., e m }, where e is the abstract representation of the connection or relationship between knowledge points, and m represents the number of connections between knowledge points; Represent each knowledge point as a d-dimensional vector v i , the dimension d is determined by the characteristics of the node; According to the connection between knowledge points, the weight w of the corresponding connection is set ij , and w ij =weight(v i ,v j ), w ij Represents node v i and v j The weight of the edge between them; Construct edge weight set W = {w ij }; According to the knowledge graph, the knowledge graph structure G = (V, E, W) is constructed, where V represents the node set, E represents the edge set, and W represents the edge weight set.

3. The method according to claim 1, characterized in that The step of inputting the interaction data into the prior model structure to obtain the a posteriori importance of each knowledge node, and adjusting the a posteriori model structure according to each a posteriori importance, includes: Obtain the interaction data of each knowledge node and pre-process the interaction data; The preprocessed interaction data are input into the prior model; Use the softmax function in the Transformer model to output the posterior importance of each knowledge node.

4. The method according to claim 1, characterized in that The method of inputting the knowledge nodes in the syllabus into the posterior model structure according to the teacher's needs to obtain the corresponding course structure and the recommended order of knowledge nodes includes: Create a list List = [P1, P2, ..., P l ], where P is the knowledge node in the syllabus and l is the number of knowledge nodes; Extract knowledge nodes from the teacher's needs and record the indexes corresponding to the knowledge nodes in a list, forming a knowledge node list Node=[N1,N2,…,N g ], where g≤l, Node represents the index of the knowledge node, N g When the value is 1, the knowledge node is a required subject. g When the value is 0, the knowledge node is a learnable subject; Select the node set V'' corresponding to the knowledge node in the knowledge node list = {v1'', v2'', ..., v g ''}, and select the dimension vector corresponding to each knowledge node in the set to form a node vector P = [w1'', w2'', ..., w g ''] T , and input the node vector P into the encoder of the Transformer model to obtain the corresponding node embedding P''=[w1",w2",…,w g ”]; Input the node embedding and knowledge node list into the decoder of the Transformer model and get the decoding result S = [s1", s2", ..., s l ”]; Input the decoding result into the softmax model to obtain the probability of each knowledge node p = [p1", p2", ..., p l ”], where p l " represents the probability of this knowledge node entering the course structure, and the knowledge node corresponding to the knowledge node index with the largest probability is selected as the prior knowledge node.

5. The method according to claim 1, characterized in that The adjusting the prior model structure according to each prior importance comprises: Setting the node height of each knowledge node according to the prior importance, and drawing the hierarchy of each knowledge node according to the node height and the connection between the nodes; Set the knowledge node border of each knowledge node; If the knowledge node is a required subject, the knowledge node border of the knowledge node is set to be a solid line and bold; if the knowledge node is a learnable subject, the knowledge node border of the knowledge node is set to be a dotted line.

6. The method according to claim 1, characterized in that Also includes: The first scheduled course data is adjusted based on the estimated data of the number of students, the estimated data of the search index and the estimated data of the skill demand to obtain an adjustment factor; Adjust the prior model structure according to each prior importance to generate a first adjustment feature; Adjust the posterior model structure according to each posterior importance to generate a second adjustment feature; Processing the adjustment factor based on the first adjustment feature and the second adjustment feature; Determine whether to generate the second fixed course data based on the processing result.

7. The method according to claim 1, characterized in that The Transformer model includes an encoder and a decoder. The structure of the encoder is composed of n hierarchical layers, each of which includes a self-attention module SA, a first residual sublayer and a first feedforward network sublayer; The structure of the decoder is composed of n hierarchical layers, each of which includes a masked multi-head attention layer MHSA, a second residual sublayer and a second feedforward network sublayer.

8. The method according to claim 1, characterized in that The multi-objective neural network based on gradient similarity supernet includes an input layer, a shared feature extraction layer, a supernet layer, a task-specific layer, an attention mechanism layer, a fusion layer and an output layer.

9. The method according to claim 1, characterized in that: The prediction model is pre-trained as follows: Obtain the historical student number time series data samples, historical search index time series data samples and historical skill demand time series data samples of the first scheduled course data; After preprocessing the historical student population time series data samples, the historical search index time series data samples, and the historical skill demand time series data samples, the historical student population time series data samples, the historical search index time series data samples, and the historical skill demand time series data samples are converted into supervised data of student population, supervised data of search index, and supervised data of skill demand, respectively; The supervised data of student numbers, search index and skill requirements are input into a multi-objective neural network based on gradient similarity hypernet for training, and the prediction model is obtained after reaching the preset training cutoff condition.

10. A secondary and higher vocational teaching reform auxiliary system, characterized in that: include: The first acquisition module is used to acquire the basic information of the teacher and create a user information data set based on the basic information of the teacher; The second acquisition module is used to obtain the course information offered by the teacher and create a course information data set according to the offered courses; The third acquisition module is used to obtain the course type in the course information data set, match the course type with the teacher in the user information data set, and obtain the first scheduled course data; A fourth acquisition module is used to obtain the time series data of the number of students, the time series data of the search index and the time series data of the skill requirements of the first scheduled course data within a set historical time period; The prediction module is used to pre-process the student population time series data, search index time series data and skill demand time series data, and then input the pre-trained prediction model to obtain the student population estimation data, search index estimation data and skill demand estimation data within a future set time period; The prediction model is obtained by training the multi-objective neural network based on the gradient similarity supernet with the historical student population time series data samples, the historical search index time series data samples and the historical skill demand time series data samples; A generation module, used for adjusting the first course scheduled data based on the student number estimated data, the search index estimated data and the skill demand estimated data, to generate the second course scheduled data; A construction module, used to obtain knowledge points from the second fixed course data according to prior knowledge, and to construct a knowledge graph with the knowledge points as nodes and the connections between the knowledge points as edges; The adjustment module is used to learn the importance of each knowledge node using the attention mechanism and loss function of the Transformer model, obtain the prior importance of each knowledge node, and adjust the prior model structure according to each prior importance; A fifth acquisition module is used to acquire the interaction data of each knowledge node, input the interaction data into the prior model structure to obtain the posterior importance of each knowledge node, and adjust the posterior model structure according to each posterior importance; The recommendation module is used to input the knowledge nodes in the syllabus into the posterior model structure according to the teacher's needs, and obtain the corresponding course structure and the recommended order of knowledge nodes.

Images