Teaching display system and method based on knowledge graph, and medium

Abstract

The invention discloses a teaching display system and method based on a knowledge graph, and a medium, and relates to the technical field of artificial intelligence and education, and the method comprises the steps: regularly collecting multi-modal original teaching data, and constructing a teaching knowledge graph; dynamically updating the relation weight of the teaching knowledge graph; according to the potential capability coefficient of each learning terminal, the distinction degree coefficient and the difficulty coefficient of the plurality of questions corresponding to each knowledge point and the relationship weight between the knowledge points, obtaining a cognitive dependency coefficient between the knowledge points corresponding to each learning terminal; generating a personalized learning path of each learning terminal according to the cognitive dependency coefficient and the learning target; performing interaction attribute addition on nodes in the teaching knowledge graph according to the personalized learning path, and generating a VR virtual scene and visual data; and real-time learning monitoring is performed on the personalized learning path of each learning terminal, and real-time learning feedback operation is performed according to the real-time learning monitoring result, so that the overall teaching quality is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the fields of artificial intelligence and educational technology, and specifically to a teaching display system and method, and a medium based on a knowledge graph. Background Art

[0002] A Chinese patent with the publication number CN114896417A discloses a method for constructing a computer education knowledge graph based on a knowledge graph, including the following steps: Step 1: Construct an education knowledge graph, S1: Perform data acquisition, S2: Perform knowledge extraction, S3: Perform knowledge identification, S4: Perform knowledge storage, S5: Perform knowledge fusion, S6: Perform quality control; Step 2: Platform construction of the education knowledge graph, A: Build a web page, and the built web page includes an education knowledge graph display module, an intelligent question answering module, and a knowledge point query module.

[0003] A Chinese patent with the publication number CN119271339A discloses a teaching material content visualization system and method based on data analysis, including: By dynamically updating the knowledge graph and optimizing the learning path, combining reinforcement learning and the knowledge graph, the optimal learning path most suitable for the current learning situation of students can be found to help students strengthen in weak knowledge points.

[0004] In the current education and teaching process, there are problems such as scattered teaching resources, lack of effective correlation display between knowledge points, and difficulty in carrying out personalized teaching according to individual differences of students. At the same time, the knowledge graph is static, and traditional systems build graphs based on preset ontologies, making it difficult to incorporate new knowledge in a timely manner (such as cutting-edge disciplines, real-time feedback from students), and traditional teaching display methods are often limited to the presentation of single-course content. It is difficult for teachers to grasp the entire subject knowledge system from a macro perspective and conduct systematic teaching, and it is also difficult for students to clearly understand the internal logical relationship between knowledge points, resulting in poor learning effects. At the same time, due to the inability to accurately analyze the learning situation of students, it is difficult to provide targeted learning suggestions and resource recommendations, and it is impossible to meet the personalized learning needs of different students. The interaction form is single: mainly two-dimensional graphics, lacking an immersive learning experience. Summary of the Invention

[0005] In order to solve the above technical problems, the purpose of the present invention is to provide a teaching display system and method, and a medium based on a knowledge graph.

[0006] The first aspect of the present invention provides a teaching display system based on a knowledge graph, including a cloud, and the cloud is communicatively connected to a learning terminal, a knowledge graph construction module, a graph dynamic update module, a cognitive dependency analysis module, a personalized path planning module, a teaching visualization module, and a real-time monitoring and feedback module;

[0007] The knowledge graph construction module is used to regularly collect multi-modal original teaching data, perform entity naming and concern extraction on the multi-modal original teaching data, and construct a teaching knowledge graph;

[0008] The graph dynamic update module is used to construct a graph dynamic update model and perform dynamic update of the relationship weights of the teaching knowledge graph;

[0009] The cognitive dependence analysis module is used to obtain the potential ability coefficients of each learning terminal, the discrimination coefficients and difficulty coefficients of each question according to the answering records of each learning terminal, construct a directed graph of knowledge paths, and obtain the cognitive dependence coefficients between each knowledge point corresponding to each learning terminal according to the potential ability coefficients of each learning terminal, the discrimination coefficients, difficulty coefficients of several questions corresponding to each knowledge point, and the relationship weights between each knowledge point;

[0010] The personalized path planning module is used to generate personalized learning paths for each learning terminal according to the cognitive dependence coefficients between each knowledge point corresponding to each learning terminal and the learning objectives;

[0011] The teaching visualization module is used to add interactive attributes to the nodes in the teaching knowledge graph according to the personalized learning path, generate a VR virtual scene and visualization data, and feedback them to the learning terminal;

[0012] The real-time monitoring and feedback module is used to perform real-time learning monitoring on the personalized learning paths of each learning terminal and perform real-time learning feedback operations according to the real-time learning monitoring results.

[0013] Further, the process of the knowledge graph construction module regularly collecting multi-modal original teaching data, performing entity naming and concern extraction on the multi-modal original teaching data, and constructing a teaching knowledge graph includes:

[0014] Regularly collect multi-modal original teaching data through several channels and mark the collection period. In this embodiment, the collection period is one month. The multi-modal original teaching data includes textbooks and lesson plans, academic papers, online course resources, and student homework and exam data. Perform format conversion (including converting PDF format textbook documents into text format), word segmentation, sentence segmentation, and part-of-speech tagging preprocessing operations on the multi-modal original teaching data. For text data, perform word segmentation operations and mark the part of speech of each word;

[0015] Input the pre - processed multi - modal original teaching data into a preset entity naming determination model for entity determination, and obtain several entities included in the multi - modal original teaching data. The entity naming determination model is trained with a large number of labeled text data, and can automatically identify entities such as course names, knowledge points, teachers, etc. Based on machine learning, construct a relationship extraction model, input the multi - modal original teaching data and several entities into the relationship extraction model for relationship extraction, and obtain the domain semantic connection relationships between several entities;

[0016] Take several entities as the nodes of the knowledge graph, take the domain semantic connection relationships between several entities as the connection relationships between the nodes, construct a teaching knowledge graph, and set the relationship weights of the edges between the nodes in the teaching knowledge graph.

[0017] Further, the process of setting the relationship weights of the edges between the nodes in the teaching knowledge graph includes:

[0018] Scan the pre - processed multi - modal original teaching data, and count the number of times each pair of entities appears simultaneously. For example, if entity A and entity B appear together n times in num data records, then the relationship weight ω between them AB is obtained through the formula For example, in the teaching text data, the two entities "function" and "derivative" appear together 50 times, and the total number of data records is 200, then the relationship weight between them is 0.25. The more times two entities appear together in the multi - modal original teaching data, the stronger the association between the two entities may be, and the corresponding relationship weight can be set higher.

[0019] Further, the process of the graph dynamic update module constructing a graph dynamic update model to dynamically update the relationship weights of the teaching knowledge graph includes:

[0020] Divide the teaching knowledge graph according to the collection period in terms of time, and generate teaching knowledge graphs G t =(V, E t ), t = 1, 2,... N, where N represents the total number of collection periods. Here, V represents the node set of the teaching knowledge graph, and E t represents the set of relationship weights of the edges corresponding to the collection period t, and ω ij t represents the relationship weight between nodes i and j in the t - th collection period. Construct a time - series graph {G1, G2,... G N}, perform node feature analysis on the time series graph, and extract the node feature sequence. Among them, the node features include static features (including the subject of the knowledge point and the prior difficulty) and dynamic features (including the number of paper mentions of the node in the last three collection cycles and the collection cycle in which the node is embedded);

[0021] Adopt the DySAT architecture, combine graph convolution and temporal convolution to construct a graph dynamic update model. Use the time series graph and the node feature sequence as the training set and the validation set, iteratively train the graph dynamic update model, optimize the model parameters of the graph dynamic update model, and at the same time, when the knowledge graph construction module collects the multi-modal original teaching data of the latest collection cycle, perform incremental learning on the graph dynamic update model;

[0022] Obtain the predicted relationship weights according to the output layer of the graph dynamic update model, and dynamically update the teaching knowledge graph according to the predicted relationship weights. For example, according to academic platforms such as IEEE Xplore and arXiv, obtain the newly added papers every month through the API interface, set the keyword as quantum computing, and dynamically update the relationship weights according to the graph dynamic update model (such as the association strength between quantum computing and linear algebra).

[0023] Furthermore, the specific process of constructing the graph dynamic update model includes:

[0024] Adopt the DySAT architecture, select the mean square error loss function, select G1~G N-1 as the training set, G N as the validation set, iteratively train the graph dynamic update model, and optimize the model parameters W (l) , the graph dynamic update model includes:

[0025] Model input layer: time series graph {G1, G2,...G N} and node feature sequence;

[0026] Temporal embedding layer: E t =PositionalEncoding(t), t∈{1,...,N}; where, PositionalEncoding() represents the encoding function, which is used to encode the time information (month) into a continuous vector;

[0027] Graph convolution layer: Among them, H t (l+1) represents the node representation updated at the current time step, σ represents the non-linear activation function, represents the degree matrix, represents the normalized adjacency matrix, W (l) represents the weight matrix;

[0028] Temporal Convolution Layer: Perform one-dimensional convolution on the time series features of each node to capture long-term dependencies;

[0029] Predict the prediction relationship weight ω for the next acquisition period according to the output layer ij N+1 ; According to the prediction relationship weight ω ij N+1 Perform graph update, and the dynamic update rule: ω ij new =α·ω ij N+1 +(1 - α)·ω ij old , where α represents the forgetting coefficient. The above formulas are all calculated by removing the dimension and taking their numerical values. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters and preset thresholds in the formulas are set by those skilled in the art according to the actual situation or obtained by simulating a large amount of data;

[0030] When adding new multi-modal original teaching data for each new acquisition period, add G N to the training set and retrain.

[0031] Furthermore, the process by which the cognitive dependence analysis module obtains the potential ability coefficients of each learning terminal, the discrimination coefficients of each question, and the difficulty coefficients based on the answering records of each learning terminal includes:

[0032] Construct a joint answering matrix based on the answering records of each learning terminal. The rows of the joint answering matrix represent learning terminals (k learning terminals), the columns represent questions (m questions), and the elements are answering results (correct = 1, wrong = 0). Randomly initialize the potential ability coefficients of each learning terminal, the discrimination coefficients of each question, and the difficulty coefficients in the answering matrix. Obtain the correct answering probabilities of each learning terminal for each question according to the potential ability coefficients, discrimination coefficients, and difficulty coefficients. Construct a log-likelihood function based on the joint answering matrix and the correct answering probabilities of each learning terminal for each question. Determine the correct answering probabilities of all learning terminals for each question given the student ability and question parameters according to the log-likelihood function;

[0033] Among them, the specific formulas for obtaining the correct answering probabilities of each learning terminal for each question and the log-likelihood function are:

[0034]

[0035] Among them, P(θ p ,a j ,b j ) represents the correct answering probability of learning terminal p for question j, θp Denote the potential ability coefficient of the learning terminal p;

[0036]

[0037] Among them, lnL represents the log-likelihood function, and X pj represents the answering result (0 or 1) of the learning terminal p for question j;

[0038] Since the potential ability coefficient of the learning terminal is interdependent with the discrimination coefficient and difficulty coefficient of the questions, the potential ability coefficient of the learning terminal, the discrimination coefficient and difficulty coefficient of the questions are solved by separate iteration. Fix the discrimination coefficient and difficulty coefficient of each question, and perform iterative update of the potential ability coefficient on the log-likelihood function. Subsequently, fix the potential ability coefficient of each learning terminal, and perform iterative update of the discrimination coefficient and difficulty coefficient on the log-likelihood function. Repeat the above process until the potential ability coefficient of each learning terminal, the discrimination coefficient of each question, and the difficulty coefficient converge, and output the potential ability coefficient of each learning terminal, the discrimination coefficient of each question, and the difficulty coefficient.

[0039] Furthermore, the process of fixing the discrimination coefficient and difficulty coefficient of each question and performing iterative update of the potential ability coefficient on the log-likelihood function includes:

[0040]

[0041] And the process of fixing the potential ability coefficient of each learning terminal and performing iterative update of the discrimination coefficient and difficulty coefficient on the log-likelihood function includes:

[0042]

[0043] Among them, argmax represents the parameter that makes the function obtain the maximum value.

[0044] Furthermore, the process of the cognitive dependence analysis module constructing a knowledge path directed graph and obtaining the cognitive dependence coefficient between each knowledge point corresponding to each learning terminal according to the potential ability coefficient of each learning terminal, the discrimination coefficient, difficulty coefficient of several questions corresponding to each knowledge point, and the relationship weight between each knowledge point includes:

[0045] Extract the knowledge point set S = {S1, S2,... S from several entities of the teaching knowledge graph n1} and extract the domain semantic connection relationships between the various knowledge points in the knowledge point set. Use each knowledge point as a node and the domain semantic connection relationships between the nodes as directed edges between the nodes (for example, S1→S2 means that S1 is a prerequisite knowledge of S2) to construct a directed graph of knowledge paths. Annotate the relationship weights of the directed edges in the directed graph of knowledge paths according to the teaching knowledge graph. At the same time, obtain the potential ability coefficients of each learning terminal, the discrimination coefficients of each question, and the difficulty coefficients according to the answer records of each learning terminal, and obtain the corresponding question sets of each knowledge point in the directed graph of knowledge paths;

[0046] When receiving the learning goals of the learning terminal, according to the potential ability coefficients of the learning terminal, the discrimination coefficients, difficulty coefficients of each question corresponding to each knowledge point in the directed graph of knowledge paths, and the relationship weights between the knowledge points, obtain the conditional mastery probabilities between the nodes in the directed graph of knowledge paths, perform standardization processing on the conditional mastery probabilities between the nodes, and convert the conditional mastery probabilities between the nodes into cognitive dependence coefficients.

[0047] Among them, the calculation formula for obtaining the conditional mastery probabilities between the nodes in the directed graph of knowledge paths is:

[0048] Suppose the nodes of the directed graph of knowledge paths are S = {S1, S2, S i ,..., S j ,...S n1}, and the calculation formula for the conditional mastery probability between two nodes (i, j) in the knowledge points of the directed graph of knowledge paths is:

[0049]

[0050] Among them, P(j|i,θ) represents the conditional probability that the learning terminal masters knowledge point j under the condition of mastering knowledge point i, w ij represents the relationship weight between knowledge point i and knowledge point j, θ represents the potential ability coefficient of the learning terminal, represents the discrimination coefficient of knowledge point j, a j represents the discrimination coefficient of question j (a j > 0, the larger the value, the stronger the discrimination ability of the question for high- and low-ability students), b j represents the difficulty coefficient of question j (the larger the value, the more difficult the question), represents the discrimination coefficient of knowledge point i, b i represents the difficulty coefficient of question i, represents the discrimination coefficient of knowledge point j, M i represents the question set corresponding to knowledge point i, |NM i | represents the total number of questions corresponding to knowledge point i, M jDenote the set of questions corresponding to knowledge point j, |NM j | represents the total number of questions corresponding to knowledge point j, Δ ij represents the amount of cognitive promotion of knowledge point i to j. The higher the difficulty of the prerequisite knowledge point i, the greater the promotion to the subsequent knowledge point j. For example, if i is a prerequisite knowledge point directly connected to j, and then Δ ij > 0, indicating that after mastering i, the difficulty of j is relatively reduced and the probability P(j|i,θ) is increased. α represents the adjustment coefficient. The specific formula for converting the conditional mastery probability between each node into a cognitive dependence coefficient is:

[0051]

[0052] where dp ij represents the cognitive dependence coefficient between knowledge point i and knowledge point j, pre(j) represents the set of all prerequisite knowledge points of knowledge point j, and pathlength(i,j) represents the shortest path length from i to j in the knowledge path directed graph (reflecting the cognitive span). Among them, ∑ d∈pre(j) P(d|i,θ) ensures that the sum of the weights of all out-edges of the same node is 1, avoiding preference for long paths. At the same time, in order to punish long paths and give priority to direct dependence relationships, the path length pathlength(i,j) is divided in the formula. The above formulas are all dimensionless and take their numerical values for calculation. The formula is obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained by simulating a large amount of data.

[0053] Furthermore, the process of the personalized path planning module generating the personalized learning path for each learning terminal according to the cognitive dependence coefficient between each knowledge point corresponding to each learning terminal and the learning goal includes:

[0054] Obtain the start node and end node in the knowledge path directed graph according to the learning goal of the learning terminal, extract all paths between the start node and the end node in the knowledge path directed graph, perform cognitive dependence aggregation on all paths, obtain the cumulative cognitive dependence coefficient of each path, and screen out the path with the largest cumulative cognitive dependence coefficient to generate the personalized learning path. The specific process includes finding the path with the largest cumulative cognitive dependence coefficient (indicating the strongest cognitive dependence) from the start node to the end node. Define the state dpx[i] as the largest cumulative cognitive dependence coefficient reaching knowledge point i, then: dpx[j] = max i∈pre(j) (dpx[i] + w ij ), and the personalized learning path can be obtained by path backtracking.

[0055] Furthermore, the process of the teaching visualization module adding interaction attributes to the nodes in the teaching knowledge graph according to the personalized learning path, generating a VR virtual scene and visualization data, and feeding them back to the learning terminal includes:

[0056] Obtain the correct answering probability, prerequisite relationship, and subject attribute of each knowledge point in the personalized learning path. Add interaction attributes to the nodes to which each knowledge point in the teaching knowledge graph belongs according to the correct answering probability, prerequisite relationship, and subject attribute of each knowledge point. The interaction attributes include the color of the knowledge point: subject classification (red - mathematics, blue - physics); size: correct answering probability (the lower the correct answering probability, the larger the node); transparency: prerequisite relationship (the prerequisite knowledge points are displayed semi-transparently); node pop-up layer display: knowledge point definition, related examples, learning videos, historical error rate, gesture operations: double-finger sliding to rotate the graph, fist-clicking to expand sub-knowledge points, etc.; voice operations: query knowledge points, request help, or perform navigation operations through voice commands; head tracking and gaze interaction operations: utilize the head tracking function of the VR headset to achieve gaze interaction. When the user gazes at a knowledge point node, the system can display relevant detailed information or pop up an interaction menu;

[0057] Among them, the calculation formula for obtaining the correct answering probability of each knowledge point is:

[0058]

[0059] Create a VR virtual scene based on the nodes in the teaching knowledge graph with interaction attributes added. The specific process includes using 3D modeling software (such as Unity, UnrealEngine, etc.) to create a VR teaching scene, designing a suitable scene layout and environment according to the subject attribute of the knowledge point. For example, in geography teaching, a virtual globe model and related geographical landscapes can be created, constructing a knowledge graph visualization model, outputting the visualization data of the nodes with interaction attributes added according to the knowledge graph visualization model (such as representing nodes with spheres and relationships with lines). The knowledge graph visualization model presents the nodes and edges in the teaching knowledge graph in a visual form in the virtual scene, presenting nodes with different interaction attributes using different shapes, colors, and sizes, feeding back the VR virtual scene and visualization data to the learning terminal, and realizing the interaction between the user and the knowledge graph using the VR device of the learning terminal.

[0060] Furthermore, the process of the real-time monitoring and feedback module performing real-time learning monitoring on the personalized learning paths of each learning terminal and performing real-time learning feedback operations according to the real-time learning monitoring results includes:

[0061] Segment the progress according to each knowledge point in the personalized learning path. One knowledge point serves as one stage to form progress stages. Then, estimate the learning time for each knowledge point based on the cognitive dependence coefficient between the knowledge points in the personalized learning path and the previous knowledge point, forming the estimated time for each progress stage. Use the end timestamps of each progress stage as the detection time points, and set the judgment criteria for each progress stage according to the correct answering probabilities of each knowledge point in the personalized learning path.

[0062] At the detection time point, obtain the actual correct answering probability of the learning terminal in the current progress stage, and compare the actual correct answering probability with the judgment criteria of the current progress stage. If the actual correct answering probability is less than the judgment criteria, extract the answering records of the learning terminal in the current progress stage and input them into the cognitive dependence analysis module. Subsequently, regenerate the personalized learning path according to the personalized path planning module. If the actual correct answering probability is greater than or equal to the judgment criteria, automatically transfer to the comparison process of the judgment criteria for the next progress stage (compare the actual correct answering probability of the next progress stage with the judgment criteria of the next progress stage).

[0063] The second aspect of the present invention also provides a teaching display method based on a knowledge graph, including the following steps:

[0064] Step s1: Regularly collect multi-modal original teaching data, perform entity naming and concern extraction on the multi-modal original teaching data, construct a teaching knowledge graph, construct a graph dynamic update model, and dynamically update the relationship weights of the teaching knowledge graph.

[0065] Step s2: Obtain the potential ability coefficients of each learning terminal, the discrimination coefficients and difficulty coefficients of each question according to the answering records of each learning terminal, construct a knowledge path directed graph, and obtain the cognitive dependence coefficients between the corresponding knowledge points of each learning terminal according to the potential ability coefficients of each learning terminal, the discrimination coefficients, difficulty coefficients of several questions corresponding to each knowledge point, and the relationship weights between each knowledge point.

[0066] Step s3: Generate the personalized learning paths of each learning terminal according to the cognitive dependence coefficients between the corresponding knowledge points of each learning terminal and the learning objectives.

[0067] Step s4: Add interaction attributes to the nodes in the teaching knowledge graph according to the personalized learning path, generate a VR virtual scene and visualization data, and feedback them to the learning terminal.

[0068] Step s5: Perform real-time learning monitoring on the personalized learning paths of each learning terminal, and perform real-time learning feedback operations according to the real-time learning monitoring results.

[0069] In the third aspect of the present invention, there is also provided a computer-readable storage medium storing computer program instructions, which when executed, implement the steps of the knowledge graph-based teaching display system as described in any one of the first aspects.

[0070] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0071] 1. Multi-modal data utilization: The knowledge graph construction module regularly collects multi-modal original teaching data, covering various forms such as text, images, and audio, which can comprehensively and deeply reflect the teaching content. Compared with the traditional method that only relies on single text data, it can provide a richer and more three-dimensional knowledge source, making the knowledge graph more complete and accurate in presenting the knowledge system in the teaching field. For example, in science courses, not only can theoretical knowledge described in words be obtained, but also videos of experimental operations and experimental data charts can be incorporated, making the knowledge graph more vivid and comprehensive.

[0072] 2. Dynamic update: The graph dynamic update model constructed by the graph dynamic update module can divide and update the teaching knowledge graph according to the collection period. It can timely integrate new teaching content, academic achievements, etc. into the knowledge graph to maintain the timeliness of knowledge. Taking the field of computer science as an example, with the rapid development of technology, new algorithms, programming language features, etc. can be updated into the knowledge graph in a timely manner to ensure that students learn the latest knowledge. At the same time, through the analysis and learning of the time series graph and node feature sequence, the relationship weight between nodes can be accurately adjusted to reflect the dynamic changes in the association between knowledge and optimize the structure of the knowledge graph.

[0073] 3. Precise cognitive dependence analysis: The cognitive dependence analysis module can accurately obtain the potential ability coefficient, item discrimination degree, and difficulty coefficient of students by deeply mining the answer records of learning terminals. On this basis, combined with the relationship weight of knowledge points in the knowledge graph, a directed graph of knowledge paths is constructed and the cognitive dependence coefficient is calculated. This enables the system to deeply understand the mastery of each student on different knowledge points and the learning dependence relationship between knowledge points. For example, in mathematics learning, the cognitive association between different knowledge sections such as functions and geometry can be clarified, providing a precise basis for personalized learning path planning.

[0074] 4. Personalized path planning: The personalized path planning module generates a personalized learning path based on the cognitive dependence coefficient and the learning goals of students. Each student can obtain a learning route that suits their own knowledge mastery situation and learning needs. For students with weak foundations, the system plans a path that gradually deepens from basic knowledge; for students with spare capacity, a more challenging and expansive learning path is provided. This greatly improves learning efficiency, avoids students wasting time on learning content that is not suitable for them, and enhances learning enthusiasm and effects.

[0075] 5. Immersive learning experience: After adding interactive attributes to the nodes of the teaching knowledge graph, the teaching visualization module generates VR virtual scenarios and visualization data. Students can learn in an immersive VR environment, enhancing the fun and sense of participation in learning. For example, in history teaching, students can be immersed in historical scenes through VR and interact with the knowledge nodes in the virtual environment, such as clicking on historical figures to learn about their life stories and touching historical relics to view relevant introductions, making the learning process more intuitive and vivid and deepening the understanding and memory of knowledge.

[0076] 6. Real-time monitoring and feedback: The real-time monitoring and feedback module monitors the personalized learning path of students in real time, segments by knowledge points and estimates the learning time, and sets judgment criteria. At the detection time point, it compares the actual correct answering probability with the judgment criteria to timely discover problems in students' learning. If a student does not meet the standard, the system automatically re-plans the learning path. This realizes the dynamic tracking and adjustment of the students' learning process, ensuring that students always learn at a pace and difficulty suitable for themselves and continuously optimizing the learning effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 It is a schematic diagram of the teaching display system based on the knowledge graph according to the embodiment of the present application.

[0078] Figure 2 It is a schematic diagram of the teaching display method based on the knowledge graph according to the embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0079] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0080] As Figure 1 shown, the first aspect of the present invention provides a teaching display system based on a knowledge graph, including a cloud, and the cloud is communicatively connected to a learning terminal, a knowledge graph construction module, a graph dynamic update module, a cognitive dependency analysis module, a personalized path planning module, a teaching visualization module, and a real-time monitoring and feedback module;

[0081] The knowledge graph construction module is used to regularly collect multi-modal original teaching data, perform entity naming and concern extraction on the multi-modal original teaching data, and construct a teaching knowledge graph;

[0082] The graph dynamic update module is used to construct a graph dynamic update model and perform dynamic update of the relationship weights of the teaching knowledge graph;

[0083] The cognitive dependence analysis module is used to obtain the potential ability coefficients of each learning terminal, the discrimination coefficients of each question, and the difficulty coefficients according to the answering records of each learning terminal, construct a directed graph of knowledge paths, and obtain the cognitive dependence coefficients between each knowledge point corresponding to each learning terminal according to the potential ability coefficients of each learning terminal, the discrimination coefficients, difficulty coefficients of several questions corresponding to each knowledge point, and the relationship weights between each knowledge point;

[0084] The personalized path planning module is used to generate personalized learning paths for each learning terminal according to the cognitive dependence coefficients between each knowledge point corresponding to each learning terminal and the learning objectives;

[0085] The teaching visualization module is used to add interactive attributes to the nodes in the teaching knowledge graph according to the personalized learning path, generate a VR virtual scene and visualization data and feedback them to the learning terminal;

[0086] The real-time monitoring and feedback module is used to perform real-time learning monitoring on the personalized learning paths of each learning terminal, and perform real-time learning feedback operations according to the real-time learning monitoring results.

[0087] It should be further noted that in the specific implementation process, the knowledge graph construction module regularly collects multi-modal original teaching data, performs entity naming and concern extraction on the multi-modal original teaching data. The process of constructing the teaching knowledge graph includes:

[0088] Regularly collect multi-modal original teaching data through several channels and mark the collection period. The collection period of this embodiment is one month. The multi-modal original teaching data includes textbooks and teaching plans, academic papers, online course resources, and student homework and exam data. Perform format conversion (including converting PDF format textbook documents into text format), word segmentation and sentence splitting, and part-of-speech tagging preprocessing operations on the multi-modal original teaching data. For text data, perform word segmentation operations and mark the part of speech of each word;

[0089] Input the multi-modal original teaching data after the preprocessing operation into a preset entity naming determination model for entity determination to obtain several entities included in the multi-modal original teaching data. The entity naming determination model is trained with a large number of labeled text data and can automatically identify entities such as course names, knowledge points, and teachers. Based on machine learning, construct a relationship extraction model, and input the multi-modal original teaching data and several entities into the relationship extraction model for relationship extraction to obtain the domain semantic connection relationships between several entities;

[0090] Among them, the process of constructing a relationship extraction model based on deep learning includes:

[0091] Define the domain semantic connection relationships in the teaching field. The domain semantic connection relationships include knowledge hierarchy relationships: such as inclusion and being included relationships. For example, "mathematics" includes sub-disciplines such as "algebra" and "geometry"; "algebra" further includes knowledge points such as "functions" and "equations"; sequential relationships: such as prerequisite relationships. "Advanced Mathematics" is a prerequisite course for "Probability Theory and Mathematical Statistics"; causal relationships: in the physics discipline, "the action of force" will cause "the change of the motion state of an object"; association relationships: certain connections existing between different knowledge points, such as "literary works" being associated with "the background of the times". Subsequently, collect a large amount of labeled sample text data. Each sample contains two entities and the domain semantic connection relationship between them. Extract features related to the domain semantic connection relationship from the text, such as the context information, part of speech, syntactic structure, etc. of the entities. Use neural networks, etc. to train the sample text data and features to obtain a relationship extraction model.

[0092] Take several entities as the nodes of the knowledge graph, take the domain semantic connection relationships between several entities as the connection relationships between the nodes, construct a teaching knowledge graph, and set the relationship weights of the edges between the nodes in the teaching knowledge graph.

[0093] It should be further noted that in the specific implementation process, the process of setting the relationship weights of the edges between the nodes in the teaching knowledge graph includes:

[0094] Scan the preprocessed multi-modal original teaching data, and count the number of times each pair of entities appear simultaneously. For example, if entity A and entity B appear together n times in num data records, then the relationship weight ω between them AB is obtained through the formula For example, in the teaching text data, the two entities "function" and "derivative" appear together 50 times, and the total number of data records is 200. Then the relationship weight between them is 0.25. The more times two entities appear together in the multi-modal original teaching data, the stronger the association between these two entities may be, and the corresponding relationship weight can be set higher.

[0095] It should be further noted that in the specific implementation process, the process of the graph dynamic update module constructing a graph dynamic update model and dynamically updating the relationship weights of the teaching knowledge graph includes:

[0096] Divide the teaching knowledge graph according to the collection period in terms of time to generate teaching knowledge graphs G t =(V, E t ), t = 1, 2,... N, where N represents the total number of collection periods. Among them, V represents the node set of the teaching knowledge graph, and E t represents the set of relationship weights of the edges corresponding to the collection period t, ωij t represents the relationship weight between nodes i and j in the t-th collection period. According to the teaching knowledge graph of different collection periods, a time series graph {G1, G2,... G N} is constructed. Node feature analysis is performed on the time series graph to extract the node feature sequence. Among them, the node features include static features (including the subject of the knowledge point and the prior difficulty) and dynamic features (including the number of paper mentions of the node in the last three collection periods and the collection period in which the node is embedded);

[0097] Adopt the DySAT architecture, combine graph convolution and temporal convolution to construct a graph dynamic update model. Use the time series graph and the node feature sequence as the training set and the validation set, iteratively train the graph dynamic update model, optimize the model parameters of the graph dynamic update model. At the same time, when the knowledge graph construction module collects the multi-modal original teaching data of the latest collection period, perform incremental learning on the graph dynamic update model;

[0098] Obtain the predicted relationship weight according to the output layer of the graph dynamic update model, and dynamically update the teaching knowledge graph according to the predicted relationship weight. For example, according to academic platforms such as IEEE Xplore and arXiv, obtain the newly added papers every month through the API interface, set the keyword as quantum computing, and dynamically update the relationship weight according to the graph dynamic update model (such as the association strength between quantum computing and linear algebra).

[0099] It should be further noted that the specific process of constructing the graph dynamic update model includes:

[0100] Adopt the DySAT architecture, select the mean square error loss function, select G1~G N-1 as the training set, G N as the validation set, iteratively train the graph dynamic update model, and optimize the model parameter W (l) of the graph dynamic update model. The graph dynamic update model includes:

[0101] Model input layer: the time series graph {G1, G2,... G N} and the node feature sequence;

[0102] Temporal embedding layer: E t = PositionalEncoding(t), t ∈ {1,..., N}; where, PositionalEncoding() represents the encoding function for encoding the time information (month) into a continuous vector;

[0103] Graph convolution layer: where, H t (l+1)represents the node representation updated at the current time step, and σ represents the non-linear activation function. represents the degree matrix. represents the normalized adjacency matrix, and W (l) represents the weight matrix.

[0104] Temporal convolutional layer: performs one-dimensional convolution on the time series features of each node to capture long-term dependencies.

[0105] Predict the predicted relationship weight ω for the next acquisition period according to the output layer. ij N+1 ; According to the predicted relationship weight ω ij N+1 Perform graph spectrum update, and the dynamic update rule: ω ij new = α·ω ij N+1 +(1 - α)·ω ij old , where α represents the forgetting coefficient. The above formulas are all calculated by removing the dimension and taking their numerical values. The formulas are obtained by software simulation of a large amount of data to get a formula closest to the real situation. The preset parameters and preset thresholds in the formulas are set by those skilled in the art according to the actual situation or obtained by simulation of a large amount of data.

[0106] When new multi-modal original teaching data for each new acquisition period is added each time, add G N to the training set and retrain.

[0107] It should be further noted that in the specific implementation process, the process of the cognitive dependence analysis module obtaining the potential ability coefficients of each learning terminal, the discrimination coefficients of each question, and the difficulty coefficients includes:

[0108] Construct a joint answer matrix according to the answer records of each learning terminal. The rows of the joint answer matrix represent learning terminals (k learning terminals), the columns represent questions (m questions), and the elements are answer results (correct = 1, wrong = 0). Randomly initialize the potential ability coefficients of each learning terminal, the discrimination coefficients of each question, and the difficulty coefficients in the answer matrix. Obtain the correct answering probabilities of each learning terminal for each question according to the potential ability coefficients, discrimination coefficients, and difficulty coefficients. Construct a log-likelihood function according to the joint answer matrix and the correct answering probabilities of each learning terminal for each question. Determine the correct answering probabilities of all learning terminals for each question given the student ability and question parameters according to the log-likelihood function.

[0109] Among them, the specific formulas for obtaining the correct answering probabilities of each learning terminal for each question and the log-likelihood function are:

[0110]

[0111] Among them, P(θ p , a j , b j ) represents the correct answering probability of learning terminal p for question j, and θ p represents the potential ability coefficient of learning terminal p;

[0112]

[0113] Among them, lnL represents the log-likelihood function, and X pj represents the answering result (0 or 1) of learning terminal p for question j;

[0114] Since the potential ability coefficient of the learning terminal is interdependent with the discrimination coefficient and difficulty coefficient of the question, the potential ability coefficient of the learning terminal, the discrimination coefficient and difficulty coefficient of the question are solved by separate iteration. Fix the discrimination coefficient and difficulty coefficient of each question, and perform iterative update of the potential ability coefficient on the log-likelihood function. Subsequently, fix the potential ability coefficient of each learning terminal, and perform iterative update of the discrimination coefficient and difficulty coefficient on the log-likelihood function. Repeat the above process until the potential ability coefficient of each learning terminal, the discrimination coefficient of each question, and the difficulty coefficient converge, and output the potential ability coefficient of each learning terminal, the discrimination coefficient of each question, and the difficulty coefficient.

[0115] It should be further noted that in the specific implementation process, the process of fixing the discrimination coefficient and difficulty coefficient of each question and performing iterative update of the potential ability coefficient on the log-likelihood function includes:

[0116]

[0117] And the process of fixing the potential ability coefficient of each learning terminal and performing iterative update of the discrimination coefficient and difficulty coefficient on the log-likelihood function includes:

[0118]

[0119] Among them, argmax represents the parameter that makes the function obtain the maximum value.

[0120] It should be further noted that in the specific implementation process, the process of the cognitive dependence analysis module constructing a knowledge path directed graph and obtaining the cognitive dependence coefficient between each knowledge point corresponding to each learning terminal according to the potential ability coefficient of each learning terminal, the discrimination coefficient, difficulty coefficient of several questions corresponding to each knowledge point, and the relationship weight between each knowledge point includes:

[0121] Extract a knowledge point set S = {S1, S2,... S n1} from several entities in the teaching knowledge graph, and extract the domain semantic connection relationships between the knowledge points in the knowledge point set. Use each knowledge point as a node and the domain semantic connection relationship between the nodes as a directed edge between the nodes (for example, S1 → S2 means that S1 is a prerequisite knowledge of S2) to construct a directed graph of knowledge paths. Annotate the relationship weights of the directed edges in the directed graph of knowledge paths according to the teaching knowledge graph. At the same time, obtain the potential ability coefficients of each learning terminal, the discrimination coefficients of each question, and the difficulty coefficients according to the answer records of each learning terminal, and obtain the corresponding question sets of each knowledge point in the directed graph of knowledge paths;

[0122] When receiving the learning goal of the learning terminal, according to the potential ability coefficient of the learning terminal, the discrimination coefficient, the difficulty coefficient of each question corresponding to each knowledge point in the directed graph of knowledge paths, and the relationship weights between each knowledge point, obtain the conditional mastery probabilities between each node in the directed graph of knowledge paths, perform standardization processing on the conditional mastery probabilities between each node, and convert the conditional mastery probabilities between each node into cognitive dependence coefficients.

[0123] Among them, the calculation formula for obtaining the conditional mastery probabilities between each node in the directed graph of knowledge paths is:

[0124] Suppose the nodes of the directed graph of knowledge paths are S = {S1, S2, S i ,..., S j ,... S n1}, and the calculation formula for the conditional mastery probability between two nodes (i, j) in the knowledge points of the directed graph of knowledge paths is:

[0125]

[0126] Among them, P(j|i,θ) represents the conditional probability that the learning terminal masters knowledge point j under the condition of mastering knowledge point i, w ij represents the relationship weight between knowledge point i and knowledge point j, θ represents the potential ability coefficient of the learning terminal, represents the discrimination coefficient of knowledge point j, a j represents the discrimination coefficient of question j (a j > 0, the larger the value, the stronger the discrimination ability of the question for high- and low-ability students), b j represents the difficulty coefficient of question j (the larger the value, the more difficult the question), represents the discrimination coefficient of knowledge point i, b i represents the difficulty coefficient of question i, represents the discrimination coefficient of knowledge point j, M i represents the question set corresponding to knowledge point i, |NMi represents the total number of questions corresponding to knowledge point i, M j represents the set of questions corresponding to knowledge point j, |NM j represents the total number of questions corresponding to knowledge point j, Δ ij represents the cognitive promotion amount of knowledge point i to j. The higher the difficulty of the prerequisite knowledge point i, the greater the promotion to the subsequent knowledge point j. For example, if i is the prerequisite knowledge point directly connected to j, and then Δ ij > 0, indicating that after mastering i, the difficulty of j is relatively reduced and the probability P(j|i,θ) is increased. α represents the adjustment coefficient. The specific formula for converting the conditional mastery probability between each node into the cognitive dependence coefficient is:

[0127]

[0128] where dp ij represents the cognitive dependence coefficient between knowledge point i and knowledge point j. pre(j) represents the set of all prerequisite knowledge points of knowledge point j. pathlength(i,j) represents the shortest path length from i to j in the knowledge path directed graph (reflecting the cognitive span). Among them, ∑ d∈pre(j) P(d|i,θ) ensures that the sum of the weights of all out-edges of the same node is 1, avoiding preference for long paths. At the same time, in order to punish long paths and give priority to direct dependence relationships, the path length pathlength(i,j) is divided in the formula. The above formulas are all calculated by removing the dimension and taking their numerical values. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters and preset thresholds in the formulas are set by those skilled in the art according to the actual situation or obtained by a large amount of data simulation.

[0129] It should be further noted that in the specific implementation process, the process of the personalized path planning module generating the personalized learning path for each learning terminal according to the cognitive dependence coefficient between each knowledge point corresponding to each learning terminal and the learning goal includes:

[0130] Obtain the start node and end node in the knowledge path directed graph according to the learning goal of the learning terminal, extract all paths between the start node and the end node in the knowledge path directed graph, perform cognitive dependence aggregation on all paths, obtain the cumulative cognitive dependence coefficient of each path, and screen out the path with the largest cumulative cognitive dependence coefficient to generate the personalized learning path. The specific process includes finding the path with the largest cumulative cognitive dependence coefficient (indicating the strongest cognitive dependence) from the start node to the end node. Define the state dpx[i] as the largest cumulative cognitive dependence coefficient to reach knowledge point i. Then: dpx[j] = max i∈pre(j) (dpx[i] + w ij ), and the personalized learning path can be obtained by path backtracking.

[0131] It should be further noted that in the specific implementation process, the process of the teaching visualization module adding interaction attributes to the nodes in the teaching knowledge graph according to the personalized learning path, generating a VR virtual scene and visualization data and feeding them back to the learning terminal includes:

[0132] Obtain the correct answering probability, prerequisite relationship, and subject attributes of each knowledge point in the personalized learning path, and add interaction attributes to the nodes to which each knowledge point in the teaching knowledge graph belongs according to the correct answering probability, prerequisite relationship, and subject attributes of each knowledge point. The interaction attributes include the color of the knowledge point: subject classification (red - mathematics, blue - physics); size: correct answering probability (the lower the correct answering probability, the larger the node); transparency: prerequisite relationship (the prerequisite knowledge point is semi-transparent); node pop-up layer display: knowledge point definition, related examples, learning videos, historical error rate, gesture operations: double-finger sliding to rotate the graph, fist-clicking to expand sub-knowledge points, etc.; voice operations: querying knowledge points, requesting help, or performing navigation operations through voice commands; head tracking and gaze interaction operations: using the head tracking function of the VR helmet to achieve gaze interaction. When the user gazes at a knowledge point node, the system can display relevant detailed information or pop up an interaction menu;

[0133] Among them, the calculation formula for obtaining the correct answering probability of each knowledge point is:

[0134]

[0135] Create a VR virtual scene according to the nodes in the teaching knowledge graph that have added interaction attributes. The specific process includes using 3D modeling software (such as Unity, Unreal Engine, etc.) to create a VR teaching scene, designing a suitable scene layout and environment according to the subject attributes of the knowledge points. For example, in geography teaching, a virtual earth model and related geographical landscapes can be created, constructing a knowledge graph visualization model, and outputting the visualization data of the nodes with added interaction attributes according to the knowledge graph visualization model (for example, using spheres to represent nodes and lines to represent relationships). The knowledge graph visualization model presents the nodes and edges in the teaching knowledge graph in a visual form in the virtual scene, presenting nodes with different interaction attributes using different shapes, colors, and sizes, feeding back the VR virtual scene and visualization data to the learning terminal, and realizing the interaction between the user and the knowledge graph using the VR device of the learning terminal.

[0136] It should be further noted that in the specific implementation process, the process of the real-time monitoring and feedback module performing real-time learning monitoring on the personalized learning paths of each learning terminal and performing real-time learning feedback operations according to the real-time learning monitoring results includes:

[0137] Segment the progress according to each knowledge point in the personalized learning path. One knowledge point serves as one stage to form progress stages. Then, estimate the learning time for each knowledge point based on the cognitive dependence coefficient between the knowledge point and the previous knowledge point in the personalized learning path to form the estimated time for each progress stage. Use the end timestamp of each progress stage as the detection time point, and set the judgment criteria for each progress stage according to the correct answering probability of each knowledge point in the personalized learning path.

[0138] Obtain the actual correct answering probability of the learning terminal at the current progress stage at the detection time point, and compare the actual correct answering probability with the judgment criteria of the current progress stage. If the actual correct answering probability is less than the judgment criteria, extract the answering record of the learning terminal at the current progress stage and input it into the cognitive dependence analysis module. Subsequently, regenerate the personalized learning path according to the personalized path planning module. If the actual correct answering probability is greater than or equal to the judgment criteria, automatically transfer to the comparison process of the judgment criteria for the next progress stage (compare the actual correct answering probability of the next progress stage with the judgment criteria of the next progress stage).

[0139] As Figure 2 shown, the second aspect of the present invention also provides a teaching display method based on a knowledge graph, including the following steps:

[0140] Step S1: Regularly collect multi-modal original teaching data, perform entity naming and concern extraction on the multi-modal original teaching data, construct a teaching knowledge graph, construct a graph dynamic update model, and dynamically update the relationship weights of the teaching knowledge graph.

[0141] Step S2: Obtain the potential ability coefficient of each learning terminal, the discrimination coefficient of each question, and the difficulty coefficient according to the answering records of each learning terminal, construct a knowledge path directed graph, and obtain the cognitive dependence coefficient between each knowledge point corresponding to each learning terminal according to the potential ability coefficient of each learning terminal, the discrimination coefficient, the difficulty coefficient of several questions corresponding to each knowledge point, and the relationship weights between each knowledge point.

[0142] Step S3: Generate the personalized learning path of each learning terminal according to the cognitive dependence coefficient between each knowledge point corresponding to each learning terminal and the learning goal.

[0143] Step S4: Add interactive attributes to the nodes in the teaching knowledge graph according to the personalized learning path, generate a VR virtual scene and visualization data, and feedback them to the learning terminal.

[0144] Step S5: Perform real-time learning monitoring on the personalized learning path of each learning terminal, and perform real-time learning feedback operations according to the real-time learning monitoring results.

[0145] The third aspect of the present invention further provides a computer-readable storage medium, which stores computer program instructions. When the computer program instructions are executed, the steps of the teaching display system based on the knowledge graph as described in any one of the first aspects are implemented.

[0146] The above embodiments are only used to illustrate the technical method of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A teaching display system based on a knowledge graph, characterized in that It includes a cloud, which is communicatively connected to a learning terminal, a knowledge graph construction module, a graph dynamic update module, a cognitive dependency analysis module, a personalized path planning module, a teaching visualization module, and a real-time monitoring and feedback module; The knowledge graph construction module is used to regularly collect multi-modal original teaching data, perform entity naming and concern extraction on the multi-modal original teaching data, and construct a teaching knowledge graph; The graph dynamic update module is used to construct a graph dynamic update model and perform dynamic update of the relationship weights of the teaching knowledge graph; The cognitive dependency analysis module is used to obtain the potential ability coefficients of each learning terminal, the discrimination coefficients of each question, and the difficulty coefficients according to the answering records of each learning terminal, construct a knowledge path directed graph, and obtain the cognitive dependency coefficients between each knowledge point corresponding to each learning terminal based on the potential ability coefficients of each learning terminal, the discrimination coefficients, difficulty coefficients of several questions corresponding to each knowledge point, and the relationship weights between each knowledge point; The personalized path planning module is used to generate personalized learning paths for each learning terminal according to the cognitive dependency coefficients between each knowledge point corresponding to each learning terminal and the learning objectives; The teaching visualization module is used to add interactive attributes to the nodes in the teaching knowledge graph according to the personalized learning path, generate a VR virtual scene and visualization data, and feedback them to the learning terminal; The real-time monitoring and feedback module is used to perform real-time learning monitoring on the personalized learning paths of each learning terminal and perform real-time learning feedback operations according to the real-time learning monitoring results.

2. The teaching display system based on a knowledge graph according to claim 1, wherein The process of the knowledge graph construction module regularly collecting multi-modal original teaching data, performing entity naming and concern extraction on the multi-modal original teaching data, and constructing a teaching knowledge graph includes: Regularly collecting multi-modal original teaching data and setting a collection period, and performing preprocessing operations such as format conversion, word segmentation, sentence splitting, and part-of-speech tagging on the multi-modal original teaching data; Inputting the multi-modal original teaching data after the preprocessing operation into a preset entity naming determination model for entity determination, obtaining several entities included in the multi-modal original teaching data, constructing a relationship extraction model based on machine learning, and inputting the multi-modal original teaching data and several entities into the relationship extraction model for relationship extraction to obtain the domain semantic connection relationships between several entities; Taking several entities as the nodes of the knowledge graph, taking the domain semantic connection relationships between several entities as the connection relationships between the nodes, constructing a teaching knowledge graph, and setting the relationship weights of the edges between each node in the teaching knowledge graph.

3. The teaching display system based on a knowledge graph according to claim 2, wherein The process of the graph dynamic update module constructing a graph dynamic update model and performing dynamic update of the relationship weights of the teaching knowledge graph includes: Performing time division on the teaching knowledge graph according to the collection period, generating teaching knowledge graphs of different collection periods, constructing a time series graph according to the teaching knowledge graphs of different collection periods, performing node feature analysis on the time series graph, and extracting node feature sequences; Build a graph dynamic update model, use the time series graph and node feature sequence as the training set and validation set, iteratively train the graph dynamic update model, optimize the model parameters of the graph dynamic update model, and perform incremental learning on the graph dynamic update model when the knowledge graph construction module collects multi-modal original teaching data in the latest collection cycle; Obtain the predicted relationship weights according to the output layer of the graph dynamic update model, and update the relationship weights of the edges between the nodes in the teaching knowledge graph according to the predicted relationship weights.

4. The teaching display system based on a knowledge graph according to claim 3, wherein The process by which the cognitive dependence analysis module obtains the potential ability coefficients of each learning terminal, the discrimination coefficients of each question, and the difficulty coefficients includes: Construct a joint answering matrix based on the answering records of each learning terminal, and randomly initialize the potential ability coefficients of each learning terminal, the discrimination coefficients of each question, and the difficulty coefficients in the answering matrix; Obtain the correct answering probabilities of each learning terminal for each question according to the potential ability coefficients, discrimination coefficients, and difficulty coefficients, and construct a log-likelihood function based on the joint answering matrix and the correct answering probabilities of each learning terminal for each question; Fix the discrimination coefficients and difficulty coefficients of each question, perform iterative update of the potential ability coefficients on the log-likelihood function, then fix the potential ability coefficients of each learning terminal, and perform iterative update of the discrimination coefficients and difficulty coefficients on the log-likelihood function. Repeat the above process until the potential ability coefficients of each learning terminal, the discrimination coefficients of each question, and the difficulty coefficients converge, and output the potential ability coefficients of each learning terminal, the discrimination coefficients of each question, and the difficulty coefficients.

5. The teaching display system based on a knowledge graph according to claim 4, wherein The process by which the cognitive dependence analysis module constructs a knowledge path directed graph and obtains the cognitive dependence coefficients between the corresponding knowledge points of each learning terminal according to the potential ability coefficients of each learning terminal, the discrimination coefficients of several questions corresponding to each knowledge point, the difficulty coefficients, and the relationship weights between each knowledge point includes: Extract the knowledge point set from several entities in the teaching knowledge graph and extract the domain semantic connection relationships between the knowledge points in the knowledge point set. Use each knowledge point as a node and the domain semantic connection relationship between the nodes as a directed edge between the nodes to construct a knowledge path directed graph. Perform relationship weight annotation on the directed edges in the knowledge path directed graph according to the teaching knowledge graph, and obtain the corresponding question set of each knowledge point in the knowledge path directed graph; Obtain the conditional mastery probabilities between the nodes in the knowledge path directed graph according to the potential ability coefficients of the learning terminal, the discrimination coefficients and difficulty coefficients of each question corresponding to each knowledge point in the knowledge path directed graph, and the relationship weights between each knowledge point. Standardize the conditional mastery probabilities between the nodes, and convert the conditional mastery probabilities between the nodes into cognitive dependence coefficients.

6. The teaching display system based on the knowledge graph according to claim 5, characterized in that, The process by which the personalized path planning module generates the personalized learning paths of each learning terminal according to the cognitive dependence coefficients between the corresponding knowledge points of each learning terminal and the learning goals includes: Obtain the start node and end node in the knowledge path directed graph according to the learning objectives of the learning terminal, extract all paths between the start node and the end node in the knowledge path directed graph, perform cognitive dependency aggregation on all paths, obtain the cumulative cognitive dependency coefficient of each path, and screen out the path with the largest cumulative cognitive dependency coefficient to generate a personalized learning path.

7. The teaching display system based on a knowledge graph according to claim 6, wherein The process of the teaching visualization module adding interaction attributes to the nodes in the teaching knowledge graph according to the personalized learning path, generating a VR virtual scene and visualization data and feeding them back to the learning terminal includes: Obtain the correct answering probability, prerequisite relationship, and subject attribute of each knowledge point in the personalized learning path, and add interaction attributes to the nodes belonging to each knowledge point in the teaching knowledge graph according to the correct answering probability, prerequisite relationship, and subject attribute of each knowledge point; Create a VR virtual scene according to the nodes in the teaching knowledge graph that have completed the addition of interaction attributes, construct a knowledge graph visualization model, and output the visualization data of the nodes that have completed the addition of interaction attributes according to the knowledge graph visualization model, and feed the VR virtual scene and visualization data back to the learning terminal.

8. The teaching display system based on a knowledge graph according to claim 7, wherein The process of the real-time monitoring feedback module performing real-time learning monitoring on the personalized learning paths of each learning terminal and performing real-time learning feedback operations according to the real-time learning monitoring results includes: Segment the progress according to each knowledge point in the personalized learning path, with one knowledge point as one stage to form a progress stage. Then, estimate the learning time of the knowledge points according to the cognitive dependency coefficient between the knowledge points and the previous knowledge point in the personalized learning path to form the estimated time of the progress stage. Use the end time stamp of each progress stage as the detection time point, and set the judgment criteria for each progress stage according to the correct answering probability of each knowledge point in the personalized learning path; Obtain the actual correct answering probability of the learning terminal in the current progress stage at the detection time point, compare the actual correct answering probability with the judgment criteria of the current progress stage. If the actual correct answering probability is less than the judgment criteria, extract the answering record of the learning terminal in the current progress stage and input it into the cognitive dependency analysis module, and then regenerate the personalized learning path according to the personalized path planning module. If the actual correct answering probability is greater than or equal to the judgment criteria, automatically transfer to the comparison process of the judgment criteria for the next progress stage.

9. A teaching display method based on a knowledge graph, specifically applied to the teaching display system based on a knowledge graph according to any one of claims 1 to 8, characterized in that, It includes the following steps: Step s1: Regularly collect multi-modal original teaching data, perform entity naming and concern extraction on the multi-modal original teaching data, construct a teaching knowledge graph, construct a graph dynamic update model, and perform dynamic update of the relationship weights of the teaching knowledge graph; Step s2: Obtain the potential ability coefficient of each learning terminal, the discrimination coefficient and difficulty coefficient of each question according to the answering records of each learning terminal, construct a knowledge path directed graph, and obtain the cognitive dependency coefficient between each knowledge point corresponding to each learning terminal according to the potential ability coefficient of each learning terminal, the discrimination coefficient, difficulty coefficient of several questions corresponding to each knowledge point, and the relationship weights between each knowledge point; Step s3: Generate personalized learning paths for each learning terminal according to the cognitive dependence coefficients and learning objectives among the knowledge points corresponding to each learning terminal; Step s4: Add interaction attributes to the nodes in the teaching knowledge graph according to the personalized learning paths, generate VR virtual scenarios and visualization data, and feedback them to the learning terminals; Step s5: Conduct real-time learning monitoring on the personalized learning paths of each learning terminal, and perform real-time learning feedback operations according to the real-time learning monitoring results.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, which, when executed, implement the steps of the teaching display system based on the knowledge graph according to any one of claims 1 to 8.

Images