Student mutual-assistance network construction and group division method and system based on knowledge graph

Abstract

The invention provides a knowledge graph-based student mutual assistance network construction and group division method and system, and belongs to the field of knowledge graphs, and the method comprises the steps: S1, obtaining student answer data information from an online learning platform or an intelligent learning system, and constructing a student learning portrait; s2, calculating the mastering degree of each knowledge point by the students according to the answer records of the students by using a Bayesian knowledge tracking model, and forming a knowledge point mastering matrix; s3, according to the knowledge point mastering matrix, calculating knowledge point mastering differences among the students by adopting a Manhattan distance, and constructing a knowledge point mastering difference matrix; constructing a student mutual assistance network according to the knowledge point mastering difference matrix; and S4, based on the student mutual assistance network, fully mining student features by using a graph convolutional neural network, and performing group division on the students by using a clustering method to form learning groups with different features. According to the method, personalized and accurate learning group division is realized, so that the learning efficiency and mutual assistance effect of students are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of knowledge graphs, and particularly relates to a method and system for constructing a student mutual assistance network and grouping based on a knowledge graph. Background Art

[0002] With the development of educational informatization, online learning platforms provide students with rich learning resources and personalized learning experiences, and at the same time accumulate a large amount of students' learning behavior data, such as answering time, submission time, correct or incorrect situation, etc. How to effectively use these data to evaluate students' learning status, establish a dynamic and accurate student mutual assistance network, and then group students with different learning characteristics to achieve data-driven personalized education has become an important research direction in the education field.

[0003] Currently, the existing learning status evaluation and student grouping methods have the following technical problems:

[0004] 1. It is difficult to effectively construct the knowledge difference and mutual assistance relationship between students: The current mutual assistance relationship construction methods often rely on the difference in the mastery of a single knowledge point, lack multi-level analysis of students' learning behavior data, and fail to effectively mine the knowledge differences between students. A simple mutual assistance relationship is difficult to form a reasonable learning network, and it is impossible to establish a targeted mutual assistance relationship between students with complementary learning needs, thus restricting the knowledge sharing and mutual learning between students.

[0005] 2. Lack of in-depth analysis of students' characteristics: Traditional clustering methods are mostly based on students' static characteristics or shallow learning behavior characteristics, and cannot deeply mine the potential characteristics of students in the dynamic learning process. And the characteristics of students in the student mutual assistance network not only depend on their own knowledge mastery, but also are affected by the mutual assistance relationship and network structure. Therefore, clustering methods based solely on surface characteristics are difficult to accurately group, resulting in inconsistent characteristics of students within the learning group and unsatisfactory mutual assistance effects.

[0006] 3. The accuracy of dynamic learning group division is low: Due to the lack of dynamic analysis of the knowledge structure, mastery degree, and mutual assistance relationship of students during the learning process, traditional group division methods cannot respond in real time to the changes in students' learning status, and the accuracy and flexibility of group division are insufficient, making it difficult to form an optimal learning group. Summary of the Invention

[0007] To solve the above technical problems, the present invention provides a method for constructing a student mutual assistance network and grouping based on a knowledge graph, including the following steps:

[0008] Including:

[0009] Step S1: Obtain students' answering data information from an online learning platform or an intelligent learning system, and construct students' learning portraits;

[0010] Step S2: Use the Bayesian knowledge tracing model to calculate students' mastery levels of each knowledge point based on their answering records, and form a knowledge point mastery matrix;

[0011] Step S3: According to the knowledge point mastery matrix, calculate the knowledge point mastery differences between students using the Manhattan distance, and construct a knowledge point mastery difference matrix; construct a student mutual assistance network based on the knowledge point mastery difference matrix;

[0012] Step S4: Based on the student mutual assistance network, use a graph convolutional neural network to fully mine students' features, and use the k-means clustering method to group students, forming learning groups with different features.

[0013] Beneficial effects:

[0014] 1. Effective construction of dynamic mutual assistance relationships: By combining the Bayesian knowledge tracing model with a knowledge graph, this invention dynamically calculates students' mastery levels of each knowledge point, and calculates the knowledge differences between students using the Manhattan distance, thereby automatically forming student mutual assistance relationships with complementary learning needs. In this way, a student mutual assistance network can be constructed more effectively to support mutual learning and knowledge sharing among students.

[0015] 2. Deeper mining of students' features: This invention uses a graph convolutional neural network to extract deep features from the student mutual assistance network, representing students' features in the student mutual assistance network as vectors, which helps to capture the potential features of each student in the network, thereby improving the accuracy of student feature mining.

[0016] 3. More targeted grouping of learning groups: By extracting students' deep features and using the k-means clustering algorithm for grouping, this invention can divide different learning groups according to the similarity of students' features, ensuring the similarity of students' features within the groups, making mutual learning more efficient, and improving the scientificity and rationality of learning group division.

[0017] 4. Strong dynamic response ability and high adaptability: This invention can dynamically update the knowledge graph and the mutual assistance network according to the real-time data in the students' learning process, adapt to the changes in students' learning states, ensure the continuous optimization of the learning mutual assistance relationship, realize the dynamic adjustment of learning group division, and improve the adaptability and flexibility of the system. Description of the drawings

[0018] Figure 1 It is a schematic flowchart of a method for constructing a student mutual assistance network and grouping based on a knowledge graph according to this invention;

[0019] Figure 2 It is a schematic diagram of a student learning portrait;

[0020] Figure 3 It is a schematic diagram of a Bayesian knowledge tracing model;

[0021] Figure 4 It is a schematic diagram of a student mutual assistance network;

[0022] Figure 5 It is a schematic diagram of student group division;

[0023] Figure 6 It is a structural block diagram of a system for constructing a student mutual assistance network and dividing groups based on a knowledge graph according to the present invention. Specific implementation manners

[0024] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0025] Embodiment 1

[0026] As Figure 1 shown, a method for constructing a student mutual assistance network and dividing groups based on a knowledge graph provided by an embodiment of the present invention includes the following steps:

[0027] Step S1: Obtain student answer data information from an online learning platform or an intelligent learning system, and construct a student learning portrait;

[0028] Step S2: Use a Bayesian knowledge tracing model to calculate the mastery degree of each knowledge point by the student according to the student's answer records, and form a knowledge point mastery matrix;

[0029] Step S3: According to the knowledge point mastery matrix, calculate the difference in knowledge point mastery between students using the Manhattan distance, and construct a knowledge point mastery difference matrix; construct a student mutual assistance network based on the knowledge point mastery difference matrix;

[0030] Step S4: Based on the student mutual assistance network, use a graph convolutional neural network to fully mine student features, and use the k-means clustering method to divide students into groups to form learning groups with different features.

[0031] In one embodiment, the above step S1: The construction of the student mutual assistance network and group division based on the knowledge graph specifically includes:

[0032] Step S11: Obtain the online answering data of each student during the learning process through the learning platform as the original answering data for constructing the student learning portrait, including information such as the student's answering time, answering content, answering result, and knowledge point tags;

[0033] In this embodiment, the goal of data collection is to obtain the answering behavior data of students on the online learning platform and organize this data into a structured input matrix for subsequent analysis and processing. The minimum input data is set as a two-dimensional matrix, denoted as , and the dimension of this matrix is , where each row represents an answering record, recording the answering behavior and related information of the student. The specific data content is as follows:

[0034] 1. Data Structure Definition

[0035] In the matrix , each row represents an answering record of a student, including the following four fields:

[0036] 1) User ID ( ): Represents the unique identifier of the student for the th record. Through this field, it is possible to uniquely determine which student the answering record belongs to.

[0037] 2) Knowledge Point ID ( ): Represents the knowledge point involved in the th record. The Knowledge Point ID identifies the knowledge point corresponding to the question answered by the student, facilitating subsequent analysis of the student's mastery of different knowledge points.

[0038] 3) Answering Correctness ( ): Represents the answering correctness of the th record, with a value of 1 (correct) or 0 (incorrect). This field is used to evaluate the student's mastery of this knowledge point and helps calculate the degree of knowledge mastery.

[0039] 4) Timestamp ( ): Represents the answering time sequence of the th record, with the unit of serial number, starting from 1 and increasing incrementally. The timestamp is used to record the time sequence of answering behavior and helps analyze the student's learning progress and the dynamic changes in knowledge mastery.

[0040] 2. Data Collection Process

[0041] In the embodiment of the present invention, the data collection process is as follows:

[0042] 1) Log in to the online learning platform: During the learning process, students log in to their accounts through the online learning platform to conduct online answering exercises for knowledge points. Each answering behavior of the students and the corresponding knowledge point information will be automatically recorded by the system.

[0043] 2) Record answering behavior in real time: After the students answer the questions, the system generates and stores a answering record in real time, including information such as user ID, knowledge point ID, answering correctness, and timestamp. This information will accumulate in chronological order to form time series data, which is convenient for subsequent analysis of the students' learning progress and the dynamics of knowledge mastery.

[0044] 3) Construct an answering data matrix : The system integrates all answering behaviors into a two-dimensional matrix according to the above answering records, where each row represents a record, and each field is the user ID, knowledge point ID, answering correctness, and timestamp in turn, constituting the four column fields of the matrix.

[0045] 4) Data quality inspection: During the data collection process, data quality inspection is carried out on the input matrix to ensure that each record is complete and accurate, avoid missing key information such as user ID or knowledge point ID, and verify whether the value of the answering correctness field is within the specified range (1 or 0).

[0046] 3. Example data

[0047] The example data includes student ID, knowledge point ID, answering correctness, and timestamp. For example, student 1001 correctly answered knowledge point 102 at timestamp 1. Such data provides the original behavior records for the subsequent steps and supports the analysis of the students' knowledge mastery and the construction of the mutual assistance network.

[0048] Step S12: Preprocess the original answering data, including: data cleaning, formatting, and data standardization processing to ensure the consistency and accuracy of the data; at the same time, perform knowledge point mapping on the original answering data according to the knowledge point tags;

[0049] Step S13: Use the knowledge graph to associate the students' answering behaviors with knowledge points and construct a student learning portrait for each student. The specific methods include: associating each answering behavior of the students with the corresponding knowledge points according to the answering content and knowledge point tags; the graph contains three types of nodes: students, knowledge points, and questions; the question node embeds correct and timestamp features internally, representing whether the answer is correct and the answering order respectively.

[0050] Such as Figure 2As shown in the figure, the knowledge map consists of three types of nodes. Taking node 1001 as an example, its subordinate nodes are knowledge point nodes, including 101, 102, and 103 respectively. The subordinate nodes of these three knowledge point nodes correspond to the questions of the knowledge points. Since the relationship between the knowledge point nodes and the question nodes is one-to-many, there can be multiple subordinate question nodes for one knowledge point. Each question node embeds the correct and timestamp features, representing whether the answer is correct and the order of answering respectively. Figure 2 It shows a part of the student learning portrait. The complete graph structure of the embodiment of the present invention contains 27,066 student nodes, 266 knowledge point nodes, and 2,541,201 question nodes.

[0051] In one embodiment, the above step S2: Using the Bayesian knowledge tracing model, according to the student's answering records, calculate the mastery degree of each knowledge point by the student, and form a knowledge point mastery matrix, specifically including:

[0052] Step S21: Initialize the mastery state of the knowledge points for each student according to the student learning portrait: Assume that the initial value of the mastery degree is between 0 and 1, where 0 means not mastered at all and 1 means completely mastered;

[0053] Step S22: Based on the basic parameters of the Bayesian knowledge tracing model, define the learning transition probability of the knowledge points, including: guessing probability , wrong answer probability , learning probability and forgetting probability ; These four basic parameters are obtained by training the Bayesian model with historical answering data and are used for subsequent calculation of the update of the knowledge mastery state of the students during the learning process;

[0054] In the embodiment of the present invention, to accurately evaluate the mastery degree of each knowledge point by each student, the Bayesian knowledge tracing model is used to dynamically model the knowledge state of the students. The Bayesian knowledge tracing model continuously updates the mastery probability of the students for the knowledge points based on the correctness of the students' answers. The mastery degree of each knowledge point is modeled according to four parameters:

[0055] In the Bayesian knowledge tracing model, the mastery degree of a certain knowledge point by each student is determined by the following four parameters:

[0056] Learning probability : It represents the probability that a student changes from not mastering the knowledge point to mastering the knowledge point, and is used to measure the success rate of the student in learning the knowledge point.

[0057] Forgetting probability : It represents the probability that a student forgets after mastering the knowledge point, and is used to reflect the possible forgetting phenomenon of the student on the already mastered knowledge.

[0058] Guessing probability : It represents the probability that a student answers a question correctly when not mastering the knowledge point, reflecting the possibility of accidental guessing.

[0059] Probability of wrong answer : It represents the probability that a student answers a question wrong when having mastered the knowledge point, used to handle the situation where a student masters the knowledge point but accidentally answers wrong.

[0060] Step S23: After each student answers a question, calculate the student's mastery level of the relevant knowledge point based on the Bayesian knowledge tracing model, and update the status according to the information of correct or wrong answer: If the answer is correct, increase the mastery level of the knowledge point; if the answer is wrong, then decrease the mastery level of the knowledge point; construct a knowledge point mastery matrix P with dimension U×S based on the mastery level of the knowledge point, where U is the number of students and S is the number of knowledge points, and each element of this matrix represents the mastery probability of student i for knowledge point j, and the mastery probability calculation formula is as follows:

[0061] ;

[0062] ;

[0063] where, represents the probability that a student answers a question correctly, represents the probability that a student answers a question wrong, and the guessing probability is the probability that student i answers correctly without mastering knowledge point j; the probability of wrong answer is the probability that student i answers wrong while mastering knowledge point j;

[0064] Update the knowledge point mastery matrix P based on the changes of learning and forgetting. The learning and forgetting adjustment formulas are as follows:

[0065] ;

[0066] where, the learning probability is the probability that student i has not mastered knowledge point j; the forgetting probability is the probability that student i forgets after mastering knowledge point j; this formula considers the dual effects of learning and forgetting, and ensures that the student's mastery level reflects the dynamic changes in the learning process through the increase and decrease correction of the mastery probability.

[0067] The knowledge point mastery matrix P is as follows:

[0068] For each student and each knowledge point , store their final knowledge mastery probabilities in the knowledge point mastery matrix P, where the element represents the mastery level of knowledge point j by student i.

[0069] .

[0070] For example, assume the initial mastery probability of a certain student is , the learning probability is , the forgetting probability is , the guessing probability is , and the wrong answer probability is .

[0071] If the first answer record of the student shows that the student answers the question correctly ( ), then the updated mastery probability is calculated according to the above formula. Then, substitute the updated mastery probability into the learning and forgetting adjustment formulas to correct it to the mastery probability for the next round. Process all the answer records of the student in turn to gradually form the student's mastery level of each knowledge point.

[0072] Through the above method, this embodiment can dynamically evaluate the student's mastery of each knowledge point, generate the knowledge point mastery matrix, and provide data support for the subsequent construction of mutual assistance relationships and group division.

[0073] As Figure 3 shows, it presents the structure of the Bayesian knowledge tracing model, where each time step includes the student's knowledge mastery state and answer result. The upper figure shows the state evolution process of the knowledge tracing model at different time steps. There are two nodes in each time step (Time Step): the knowledge mastery state and the answer result . The knowledge mastery state is a hidden variable, representing the student's mastery level of knowledge point at time , which is affected by the mastery state at the previous time step (the arrow indicates the dependency relationship). The answer result is an observable variable, indicating whether the student's answer is correct or wrong at this time step, depending on the current mastery state . Through the chain structure of these time steps, the model can infer how the student's knowledge mastery progresses over time. The lower figure shows the four transition probabilities between the mastered and unmastered states: P(T) is the learning probability, P(S) is the forgetting probability, P(G) is the probability of guessing correctly when not mastered, and P(S) is the probability of answering wrong when mastered. This model is used to track the change of the student's knowledge mastery over time.

[0074] The pseudo-code of the algorithm for calculating the student's knowledge point mastery degree through Bayesian knowledge tracing is as follows: Input: Student answer data, Bayesian Knowledge Tracing model parameters (learn_prob, forget_prob, guess_prob, slip_prob), breakpoint file path checkpoint_file function BKTSystem(learn_prob, forget_prob, guess_prob, slip_prob, checkpoint_file) Initialize student_skill_models and processed_timestamps if checkpoint_file exists then load_checkpoint(checkpoint_file) end function function get_or_create_model(user_id, skill_id) if model does not exist then create an instance of BKTModel return model end function function update_from_data(data) for each record row in data do Skip processed timestamp Get user_id, skill_id, correct model ← get_or_create_model(user_id, skill_id) model.update_knowledge(correct) end for end function function get_knowledge_level(user_id, skill_id) return get_or_create_model(user_id, skill_id).get_knowledge_level() end function function get_knowledge_matrix(student_ids, skill_ids) Initialize the knowledge_matrix for user_id in student_ids, skill_id in skill_ids do knowledge_matrix[user_id][skill_id] ← get_knowledge_level(user_id, skill_id) end for return knowledge_matrix end function。

[0075] In one embodiment, the above step S3: According to the knowledge point mastery matrix, use the Manhattan distance to calculate the difference in knowledge point mastery between students, and construct a knowledge point mastery difference matrix; construct a student mutual assistance network based on the knowledge point mastery difference matrix, specifically including:

[0076] Step S31: Set a predetermined knowledge point mastery difference threshold T to determine whether the knowledge difference between students meets the conditions for a mutual assistance relationship;

[0077] Step S32: Use the mastery degree data in the knowledge point mastery matrix and calculate the difference degree in knowledge point mastery between each pair of students using the Manhattan distance. The specific method is as follows:

[0078] ;

[0079] where the element represents the difference in knowledge point mastery between student and student . Here, it is assumed that knowledge points with a knowledge point mastery level of 0 are not included in the difference calculation. is an indicator function, which takes 1 when the condition in the parentheses is satisfied, otherwise it takes 0. When the mastery degrees of two students on knowledge point are both non-zero, the difference is included in the sum;

[0080] Step S33: Record the difference in knowledge point mastery between all students in the knowledge point mastery difference matrix; assuming there are n students, this matrix is n×n, and the diagonal elements are zero, that is, the self-difference of students is zero, and other elements represent the difference in knowledge point mastery between any two students;

[0081] Step S34: According to the knowledge point mastery difference matrix, determine whether the knowledge point mastery difference between each pair of students exceeds the difference threshold T; if the difference between two students is greater than T, then these two students meet the mutual assistance condition, and record this pair of students as a mutual assistance relationship binary group; each binary group represents a pair of students with complementary learning needs.

[0082] Set a difference threshold . If , then establish an edge between student and . This edge may be bidirectional or unidirectional. Bidirectional means that both parties can get help from each other, and unidirectional means that only one party can provide help to the other party, and vice versa. The characteristic of the edge is a set of knowledge points, indicating that there is a large difference between them in this knowledge point.

[0083] Step S35: In the knowledge graph, construct mutual assistance edges for student pairs that meet the mutual assistance condition. Each mutual assistance edge represents the learning mutual assistance relationship between two students, and records the student mutual assistance network in the form of a graph structure.

[0084] As Figure 4 shown, the student mutual assistance network is constructed based on the student learning portraits. Taking student 1001 as an example, his mastery degrees of knowledge points 101, 102, and 103 are 0.9, 0.3, and 0.2 respectively. Through the calculation of the Manhattan distance, the mastery differences between student 1001 and student 1002 and student 1003 both exceed the difference threshold T. Therefore, a mutual assistance relationship is formed between student 1001 and student 1002 and student 1003. Specifically: student 1001 can help student 1002 learn knowledge point 101, student 1002 can help student 1001 learn knowledge point 102, and student 1001 can help student 1003 learn knowledge point 101. Figure 4 Shows a part of the student mutual assistance network. The complete graph structure contains 27,066 student nodes, 266 knowledge point nodes, and 2,541,201 question nodes. There are also 11,294 mutual assistance edges.

[0085] In one embodiment, the above step S4: Based on the student mutual assistance network, use the graph convolutional neural network to fully mine student features, and use the k-means clustering method to group students to form learning groups with different features, specifically including:

[0086] Step S41: Initialize the feature vector for each student node in the student mutual assistance network. The initial feature vector includes the basic information of the student and the key learning features extracted from the knowledge graph; use the initial features as the input of the graph convolutional neural network for in-depth feature extraction in the network.

[0087] Step S42: Input the student mutual assistance network into the graph convolutional neural network. Through the graph node convolution operation of the graph convolutional neural network, extract the deep features of students in the mutual assistance network; the graph convolutional neural network updates the feature vector of each student node by passing and aggregating the information of adjacent nodes layer by layer, so that the finally extracted feature vector can reflect the position and feature relationship of students in the mutual assistance network.

[0088] The specific steps are as follows:

[0089] Use the graph convolutional neural network to analyze the student mutual assistance network. Let the initial feature vector be each row feature in the knowledge mastery matrix After two layers of graph convolutional neural networks:

[0090] The first layer of convolution: ;

[0091] The second layer of convolution: ;

[0092] Among them: is the adjacency matrix, which defines the mutual assistance relationship; is the learnable weight matrix; is the initial feature vector of the student, that is, the knowledge mastery matrix ; is used as the input for group division. Through k-means clustering, students are divided into different group matrices .

[0093] Step S43: According to the feature vector, set the feature division standard so as to divide students into different learning groups according to feature similarity during the clustering process.

[0094] Step S44: Input the feature vector extracted from the graph convolutional neural network into the k-means clustering algorithm. According to the preset number of groups k, perform clustering analysis. The k-means clustering method assigns nodes with similar student features to the same group through iterative optimization, and finally forms learning groups with different features.

[0095] The specific steps are as follows:

[0096] After completing the construction of the mutual assistance network and the graph convolution operation, a new feature matrix will be obtained. Each row of this matrix represents the features of students in terms of knowledge mastery and mutual assistance potential.

[0097] Set the feature matrix as:

[0098] ;

[0099] Among them, is the feature dimension (obtained from the output of the graph convolutional neural network).

[0100] k-means is a commonly used clustering algorithm whose goal is to divide samples into groups such that the distance between each sample and the center of its assigned group is minimized. The clustering process is as follows:

[0101] 1. Initialize cluster centers: Randomly select samples as the initial cluster centers, denoted as .

[0102] 2. Assign samples to the nearest cluster center: For each sample , calculate its Euclidean distance to all cluster centers and assign it to the nearest cluster center:

[0103] ;

[0104] where represents the Euclidean distance.

[0105] 3. Update cluster centers: Calculate the mean of each cluster and update the cluster centers:

[0106]

[0107] where, represents the sample index belonging to the -th cluster.

[0108] 4. Iteration: Repeat steps 2 and 3 until the cluster centers no longer change (converge) or reach the set maximum number of iterations.

[0109] The final output is the group partition matrix with dimension where each element indicates which group a certain student belongs to.

[0110]

[0111] The pseudocode for this process is as follows: function GCN_Model(input_dim, hidden_dim, output_dim) Initialize two-layer GCN convolutional layers end function function forward_pass(x, edge_index) x ← conv1(x, edge_index) → relu(x) x ← conv2(x, edge_index) return x end function function train_model(data, model, optimizer, loss_fn, epochs) for epoch from 1 to epochs do Calculate the loss loss (assuming the target is a zero vector) Backpropagate and update the weights if (epoch is a multiple of 10) then Print epoch and loss end for end function function perform_clustering(node_features, num_groups) Return the k - means clustering labels end function function main() Initialize the model parameters and graph data Train the model train_model() Group labels ← perform_clustering(model output, num_groups = 3) Print the group labels of each student end function

[0112] As Figure 5 shown, it shows the clustering of students and the formed student group division matrix, where the students are divided into 6 groups according to characteristics such as learning ability, knowledge point mastery, and mutual assistance activity. Student 1001 belongs to group_1, student 1002 belongs to group_4, and student 1003 belongs to group_0. Each group has different learning characteristics. By analyzing the personalized characteristics of each group, teachers can give targeted teaching guidance to achieve precise teaching.

[0113] Example Two

[0114] As Figure 6 shown, this embodiment of the present invention provides a student mutual assistance network construction and group division system based on a knowledge graph, including the following modules:

[0115] Build a student learning profile module 51, which is used to obtain student answer data information from an online learning platform or an intelligent learning system and build a student learning profile;

[0116] Build a knowledge point mastery matrix module 52, which is used to use the Bayesian knowledge tracing model to calculate the mastery degree of each knowledge point by students according to the students' answer records and form a knowledge point mastery matrix;

[0117] Build a student mutual assistance network module 53, which is used to calculate the difference in knowledge point mastery between students using the Manhattan distance according to the knowledge point mastery matrix and build a knowledge point mastery difference matrix; build a student mutual assistance network based on the knowledge point mastery difference matrix;

[0118] Build a learning group module 54, which is used to fully mine student features based on the student mutual assistance network using a graph convolutional neural network and use the k-means clustering method to divide students into groups to form learning groups with different features.

[0119] A device for constructing a student mutual assistance network and group division based on a knowledge graph, including one or more electronic devices, where one or more electronic devices are used to implement a method, system, and device for constructing a student mutual assistance network and group division based on a knowledge graph.

[0120] An electronic device, including: one or more processors; a memory for storing one or more programs, where when the one or more programs are executed by the one or more processors, the one or more processors implement a method, system, and device for constructing a student mutual assistance network and group division based on a knowledge graph.

[0121] A computer-readable storage medium, on which executable instructions are stored, and when the instructions are executed by a processor, the processor implements a method, system, and device for constructing a student mutual assistance network and group division based on a knowledge graph.

[0122] A non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements a method, system, and device for constructing a student mutual assistance network and group division based on a knowledge graph.

[0123] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present invention will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for constructing a student mutual assistance network and group division based on a knowledge graph, characterized in that, Including: Step S1: Obtain the student answer data information from an online learning platform or an intelligent learning system, and construct a student learning portrait; Step S2: Use the Bayesian knowledge tracing model to calculate the mastery level of each knowledge point by the student based on the student's answer records, and form a knowledge point mastery matrix; Step S3: According to the knowledge point mastery matrix, use the Manhattan distance to calculate the difference in knowledge point mastery among students, and construct a knowledge point mastery difference matrix; construct a student mutual assistance network based on the knowledge point mastery difference matrix; Step S4: Based on the student mutual assistance network, use a graph convolutional neural network to fully mine student features, and use the k-means clustering method to group students to form learning groups with different features.

2. The method for constructing a student mutual assistance network and group division based on a knowledge graph according to claim 1, characterized in that The above-mentioned Step S1: Construction and group division of a student mutual assistance network based on a knowledge graph specifically includes: Step S11: Obtain the online answer data of each student during the learning process through the learning platform as the original answer data for constructing the student learning portrait, including information such as the student's answer time, answer content, answer result, and knowledge point label; Step S12: Preprocess the original answer data, including data cleaning, formatting, and data standardization processing to ensure data consistency and accuracy; at the same time, perform knowledge point mapping on the original answer data according to the knowledge point label; Step S13: Use the knowledge graph to associate the student's answer behavior with knowledge points and construct a student learning portrait. The specific method includes: according to the answer content and knowledge point label, associate each answer behavior of the student with the corresponding knowledge point; the graph contains three types of nodes: students, knowledge points, and questions; the question node internally embeds correct and timestamp features, representing whether the answer is correct and the answer order respectively.

3. The method for constructing a student mutual assistance network and group division based on a knowledge graph according to claim 2, characterized in that, The above-mentioned Step S2: Use the Bayesian knowledge tracing model to calculate the mastery level of each knowledge point by the student based on the student's answer records, and form a knowledge point mastery matrix, specifically including: Step S21: Initialize the mastery state of knowledge points for each student according to the student learning portrait: Assume that the initial value of the mastery level is between 0 and 1, where 0 means completely not mastered and 1 means completely mastered; Step S22: Define the learning transition probability of knowledge points based on the basic parameters of the Bayesian knowledge tracing model, including: guessing probability , wrong answer probability , learning probability and forgetting probability . The parameters are obtained by training the Bayesian model with historical answering data; Step S23: After each student answers a question, calculate the student's mastery level of relevant knowledge points based on the Bayesian knowledge tracing model, and update the status according to the information of correct or incorrect answers; construct a knowledge point mastery matrix P with dimensions U×S based on the mastery level of knowledge points, where U is the number of students and S is the number of knowledge points, and each element of this matrix represents the mastery probability of student i for knowledge point j: ； 。 4. The method for constructing a student mutual assistance network and group division based on a knowledge graph according to claim 3, wherein The above-mentioned Step S3: According to the knowledge point mastery matrix, use the Manhattan distance to calculate the difference in knowledge point mastery among students, and construct a knowledge point mastery difference matrix; construct a student mutual assistance network based on the knowledge point mastery difference matrix, specifically including: Step S31: Set a predetermined knowledge point mastery difference threshold T to determine whether the knowledge difference degree among students meets the conditions for a mutual assistance relationship; Step S32: Use the mastery level data in the knowledge point mastery matrix and use the Manhattan distance to calculate the difference degree in knowledge point mastery for each pair of students. The specific method is: ； Among them, the element represents the difference in knowledge mastery between students and students and is an indicator function; is the indicator function; Step S33: Record the knowledge point mastery differences among all students in the knowledge point mastery difference matrix; assume there are n students, then this matrix is n×n; Step S34: Based on the knowledge point mastery difference matrix, determine whether the knowledge point mastery difference between each pair of students exceeds the difference threshold T; if the difference between two students is greater than T, then these two students meet the mutual assistance condition, and record this pair of students as a mutual assistance relationship binary group. Step S35: In the knowledge graph, construct mutual assistance edges for student pairs that meet the mutual assistance condition. Each mutual assistance edge represents the learning mutual assistance relationship between two students, and record the student mutual assistance network in the form of a graph structure.

5. The method for constructing a student mutual assistance network and group division based on a knowledge graph according to claim 4, characterized in that Step S4: Based on the student mutual assistance network, use a graph convolutional neural network to fully mine student features, and adopt the k-means clustering method to group students, forming learning groups with different features, specifically including: Step S41: Initialize the feature vector for each student node in the student mutual assistance network. The initial feature vector includes the basic information of the student and the key learning features extracted from the knowledge graph; use the initial features as the input of the graph convolutional neural network for in-depth feature extraction in the network. Step S42: Input the student mutual assistance network into the graph convolutional neural network, and extract the deep features of students in the mutual assistance network through the graph node convolution operation of the graph convolutional neural network; the graph convolutional neural network updates the feature vector of each student node by passing and aggregating information of adjacent nodes layer by layer, so that the finally extracted feature vector can reflect the position and feature relationship of the student in the mutual assistance network. Step S43: Set the feature division criteria according to the feature vector, so as to divide students into different learning groups according to feature similarity during the clustering process. Step S44: Input the feature vector extracted from the graph convolutional neural network into the k-means clustering algorithm, and perform clustering analysis according to the preset number of groups k. The k-means clustering method assigns nodes with similar student features to the same group through iterative optimization, and finally forms learning groups with different features.

6. A student mutual assistance network construction and group division system based on a knowledge graph, characterized in that, including the following modules: A module for constructing a student learning portrait, which is used to obtain student answer data information from an online learning platform or an intelligent learning system and construct a student learning portrait. A module for constructing a knowledge point mastery matrix, which is used to calculate the mastery degree of each knowledge point of the student according to the student's answer record by using the Bayesian knowledge tracing model, and form a knowledge point mastery matrix. A module for constructing a student mutual assistance network, which is used to calculate the knowledge point mastery difference between students by using the Manhattan distance according to the knowledge point mastery matrix, and construct a knowledge point mastery difference matrix; construct a student mutual assistance network based on the knowledge point mastery difference matrix. A module for constructing learning groups, which is used to fully mine student features based on the student mutual assistance network by using a graph convolutional neural network, and adopt the k-means clustering method to group students, forming learning groups with different features.

7. A device for constructing a student mutual assistance network and grouping based on a knowledge graph, characterized in that, including one or more electronic devices, where the one or more electronic devices are used to implement the method according to any one of claims 1 to 5.

8. An electronic device, characterized in that, including: one or more processors; A memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 5.

9. A computer-readable storage medium, characterized in that, There are executable instructions stored thereon, and when the instructions are executed by a processor, the processor implements the method according to any one of claims 1 to 5.

10. A non-transitory computer-readable storage medium, characterized in that, There is a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images