A teaching optimization method based on big data informationization

Abstract

The present invention discloses a method for optimizing teaching based on information technology of big data, relates to the technical field of information technology teaching management, and is used to solve the problem of poor teaching management in the process of information technology teaching; the present invention integrates multimodal data collection and deep learning, analyzes students' behaviors, emotions and physiological states in real time, generates accurate learning state characteristics, constructs a dynamic knowledge graph according to a graph convolutional network, updates students' knowledge mastery, and optimizes learning paths, and strengthens learning to adjust personalized learning tasks through instant feedback to improve learning efficiency. At the same time, based on causal inference, the influence of teaching behaviors on learning effects is analyzed, the intervention effects are quantified, and teaching strategies are adjusted. Through the combination of multimodal fusion, graph convolutional networks and reinforcement learning, personalized learning support is provided, teaching strategies are optimized, and learning efficiency is improved, which has significant teaching effects and learning path optimization advantages.

Description

Technical Field

[0001] The present invention relates to the field of information-based teaching technology, and more specifically, to a method for optimizing information-based teaching based on big data. Background Art

[0002] Information-based teaching optimization is a method based on digitization and intelligence that optimizes the teaching process through big data technology, artificial intelligence, and data-driven decision-making. By collecting student learning data, teacher teaching data, and teaching resource usage data, it can accurately analyze problems in the teaching process and provide a scientific basis for personalized learning path recommendations, resource allocation optimization, and teaching strategy adjustments.

[0003] The shortcomings of existing technologies are: the ability to dynamically adjust students' personalized learning paths is limited. Most teaching systems lack in-depth mining and precise adaptation of students' real-time learning behaviors, making it difficult to dynamically adjust learning content and difficulty based on students' immediate performance. Knowledge graphs are mostly static models that cannot update students' knowledge mastery in real time and dynamically associate with subjects, resulting in inaccurate diagnosis of learning bottlenecks. The analysis of teaching behaviors and learning outcomes mostly stays at the correlation level, and causal inference technology is not used to deeply explore the direct impact of different teaching strategies on learning outcomes, making it difficult to provide a scientific basis for the optimization of teaching strategies, resulting in backward teaching strategies. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, there is the following solution to solve the problem of poor teaching management in the information-based teaching process in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: A teaching optimization method based on big data information technology includes the following steps: Collect students' behavior, emotions and physiological data, perform synchronous alignment and normalization, extract features and reduce dimensions based on deep learning, and fuse features of each modality through a multi-head attention mechanism; Construct a graph of knowledge points and dependencies, update students’ knowledge mastery status in real time, and adjust learning paths and knowledge mastery status based on student performance and knowledge propagation characteristics of graph convolutional networks; Define state space and action space, optimize personalized learning paths based on immediate feedback and reinforcement learning, and adjust tasks in real time; Based on causal inference, we analyze the impact of teaching behavior on learning outcomes, establish a causal relationship model to quantify the intervention effect, and adjust the teaching strategy.

[0006] In a preferred embodiment, student behavior, emotion and physiological data are collected, synchronized and normalized, and features are extracted and dimension reduced based on deep learning. The specific steps are as follows: The collected behavioral data include clickstream logs, task completion time, answer accuracy, and learning frequency; Collect emotional data, capture facial expressions and gestures through cameras, and collect speech through microphones; Collect physiological data and obtain students' physiological signals through wearable devices. Physiological signals include heart rate changes and skin conductance. For the collected data of different time granularities, the time window slicing method is used for alignment and the logarithmic scaling function is used for normalization; Perform behavioral data feature extraction to extract complex patterns in time series, including the correlation between task completion time and click behavior; Extract emotional data features and extract emotional features from facial expressions and voice signals through convolutional neural networks. Emotional features include changes in attention and stress. Perform physiological data feature extraction and use variational autoencoder to extract feature vectors of physiological signals; For the extracted features, nonlinear dimensionality reduction method is used for dimensionality reduction.

[0007] In a preferred embodiment, the features of each modality are fused through a multi-head attention mechanism, and the specific steps include: The collected data features are constructed into physiological data feature matrix, emotional data feature matrix, and behavioral data feature matrix, and the feature matrices are fused through a multi-head attention mechanism; The fused feature matrix is ​​input into the graph embedding model to generate a global embedding vector of the student’s learning status.

[0008] In a preferred implementation, a knowledge point and dependency graph is constructed to update the student's knowledge mastery status in real time. The specific steps are as follows: Initialize knowledge points and associations, extract knowledge points and dependencies between knowledge points from teaching resources; Define nodes, knowledge points It is the basic unit of the graph, each node Indicates specific teaching knowledge points; Define the edge and calculate the weight. The edge represents the relationship between knowledge points. Use the co-occurrence probability to mine the previous and next dependencies between knowledge points from teaching resources. The formula is: ,in, It is a knowledge point and The number of times they appear in the same question, the probability of co-occurrence Indicates the strength of association between knowledge points and serves as the weight of the edge; Initialize the knowledge point state, initial state Based on the matching calculation between student feature embedding and knowledge points, the dot product method is used to measure the student embedding vector and knowledge point embedding vector The matching degree: ,in, is the mapping matrix used for learning, is the bias term, T represents the transposition operation, is the Sigmoid function; Provide student performance feedback updates and dynamically update knowledge point status based on student performance in tests and assignments : ,in, Indicates that students are at the knowledge point The test scores of Representing knowledge points , The strength of the dependence between is the learning rate.

[0009] In a preferred embodiment, based on the student performance and according to the knowledge propagation characteristics of the graph convolutional network, the learning path and the knowledge mastery state are adjusted, including the following steps: Use the graph attention network to propagate the knowledge point features and update the feature representation of each knowledge point: , where the attention weight Through knowledge points and The association strength and current feature representation calculation: ,in, and is the student's learning state vector and the feature vector of the knowledge point, sim is the similarity measurement function, and N is the total number of knowledge points; Based on students' latest learning performance and communication characteristics, the node attributes and edge weights of the knowledge graph are adjusted in real time.

[0010] In a preferred embodiment, defining the state space and the action space, optimizing the personalized learning path based on immediate feedback and reinforcement learning, and adjusting the task in real time include the following steps: Define the state space. The state of the state space includes the student's current learning state and the node features of the dynamic knowledge graph. The student's learning state is represented by a feature vector, and the node features of the dynamic knowledge graph are represented by the updated knowledge point features. Define an action space, where the actions in the action space represent recommended learning tasks. The action set is dynamically generated based on the node relationships in the dynamic knowledge graph and the student's knowledge mastery status. Define the immediate reward as the trade-off between the improvement in knowledge mastery and the time cost after completing the learning task as the reward function; The state captures the student's current learning situation, the action provides task selection, and the reward function evaluates the learning task; Input the state into the policy function, output the action probability distribution, and use the PPO algorithm to optimize the policy function; Based on the optimized strategy, the recommended learning task is executed at each time step, and the students' learning feedback is collected and the task is adjusted according to the learning feedback.

[0011] In a preferred embodiment, the influence of teaching behavior on learning effect is analyzed based on causal inference, a causal relationship model is established to quantify the intervention effect, and the teaching strategy is adjusted, including the following steps: The causal variables were defined, and the independent variables represented the intervening variables of teaching behaviors, including the frequency of classroom interactions, task allocation strategies, and the complexity of teaching content; The dependent variables represent the students’ learning outcomes, including the mastery of knowledge points and the accuracy of answering questions; Confounding variables represent factors that affect both the independent and dependent variables; Use Bayesian networks to automatically learn causal graph structures from historical teaching data. Causal graph nodes represent variables, and edges represent causal relationships. The learning process determines the network structure based on the method of maximizing log-likelihood estimation. The constructed causal diagram is input into the causal inference stage to analyze the intervention effect of the teaching strategy and define the causal effect; According to the Do-Calculus rule, the estimated path of the intervention effect is separated from the causal diagram; Based on the analysis results of the intervention effects, optimize teaching strategies and dynamically adjust teaching behaviors.

[0012] The technical effects and advantages of the present invention are as follows: The present invention uses multimodal data collection and deep learning technology to accurately capture students' behaviors, emotions, and physiological states, combines multi-head attention mechanisms to achieve data fusion, generates accurate learning state features, and based on the constructed dynamic knowledge graph and graph convolutional network, updates students' knowledge mastery status in real time and adjusts the learning path to ensure personalized learning support, and then strengthens learning to optimize personalized learning paths, and adjusts learning tasks in real time with instant feedback to improve learning efficiency and experience. At the same time, the impact of teaching behavior on learning outcomes is analyzed through causal inference, the intervention effect is quantified, and the teaching strategy is scientifically adjusted to improve the teaching effect. This solution optimizes learning paths and teaching strategies through the combination of multimodal fusion, graph convolutional networks, and reinforcement learning, and has significant advantages in personalized learning support, improved teaching outcomes, and optimized learning efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The present invention is a flowchart of a method for optimizing teaching based on big data information technology. DETAILED DESCRIPTION

[0014] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0015] In order to achieve the above objectives, Figure 1 A structural schematic diagram of a teaching optimization method based on big data informationization of the present invention is given, which specifically includes the following steps: Collect students' behavior, emotions and physiological data, perform synchronous alignment and normalization, extract features and reduce dimensions based on deep learning, and fuse features of each modality through a multi-head attention mechanism; Construct a graph of knowledge points and dependencies, update students’ knowledge mastery status in real time, and adjust learning paths and knowledge mastery status based on student performance and knowledge propagation characteristics of graph convolutional networks; Define state space and action space, optimize personalized learning paths based on immediate feedback and reinforcement learning, and adjust tasks in real time; Based on causal inference, we analyze the impact of teaching behavior on learning outcomes, establish a causal relationship model to quantify the intervention effect, and adjust the teaching strategy.

[0016] Step 1: Perform multimodal data integration and feature modeling to obtain multimodal data (behavior, emotion, physiology, etc.), and transform these heterogeneous data into usable information through a unified feature modeling and fusion process. The specific steps are as follows: Collecting behavioral data , record students’ operation behaviors in the information-based teaching system, including clickstream logs, task completion time, answer accuracy, learning frequency, etc.; Collecting emotional data ,Using sentiment analysis technology, the camera captures facial expressions and gestures and the microphone collects the voice emotional features to extract the students’ learning emotions; Collecting physiological data , using wearable devices (such as smart bracelets) to obtain students’ physiological signals, including heart rate changes, skin conductance, etc., to determine the stress and concentration level of their learning state; Behavioral data is the core data that directly reflects students’ learning behavior and efficiency. Emotional data is used to capture students’ emotional fluctuations during the learning process, such as fatigue and anxiety, by analyzing the emotional characteristics of students’ expressions and voices. Physiological data provides students’ physiological status through wearable devices, such as heart rate changes and stress index, which indirectly reflects students’ concentration and learning load. The learning status of students is multi-dimensional. Single-mode data may be one-sided or misleading. For example, although behavioral data can record the number of tasks completed by students, it cannot determine their true learning engagement. Emotional and physiological data can make up for this shortcoming. Therefore, through multi-modal data collection, we can fully describe the learning status of students and provide an accurate information basis for teaching optimization. For data of different time granularities, the time window slicing method is used for alignment. Assuming that the sampling frequency of the behavior data is , the sampling frequency of sentiment data is , the sampling frequency of physiological data is , the aligned time window can be expressed as: , where lcm represents the least common multiple operation, mapping all data to the time window , and fill in missing values ​​to unify the time dimension; Normalization uses a logarithmic scaling function to avoid the sensitivity of traditional linear normalization to extreme values. The normalization formula is as follows: ,in is the original data sample, is the normalized result, max(X) is the maximum value of the feature dimension, X is the data set, expressed as .

[0017] Data at different time scales may result in data loss or redundancy. For example, the sampling frequency of emotion data is high, while the sampling frequency of behavior data is low. If they are not aligned, the behavior data in the time window will be too sparse, affecting the timeliness of the analysis. Similarly, different data modalities (such as heart rate and answering time) have different dimensions, and direct analysis may produce deviations.

[0018] Use self-supervised learning technology to extract high-dimensional features from different data modalities, and use different advanced algorithms to extract features from the collected multimodal data: For behavioral data feature extraction, TransformerEncoder is used to extract complex patterns in time series, such as the correlation between task completion time and click behavior; For emotional data feature extraction, convolutional neural networks (CNNs) are used to extract emotional features from facial expressions and voice signals, such as attention and stress changes. Extract physiological data features and use variational autoencoders (VAE) to extract the potential feature vectors of physiological signals to capture the concentration and fatigue states during the learning process; For the extracted high-dimensional features, nonlinear dimensionality reduction methods (such as t-SNE or UMAP) are used to reduce the dimensionality, retain the local manifold structure of the features, and retain the local manifold structure of the data to generate the feature matrix after dimensionality reduction: ,in, is the feature dimension after dimensionality reduction, L is the number of input samples of the data set, and R represents the number of dimensions of the feature matrix after dimensionality reduction; High-dimensional features may have collinearity and noise. Direct modeling will affect the computational efficiency and predictive performance of the model. Dimensionality reduction can remove irrelevant information and reduce data complexity. For example, some features in behavioral data (such as irrelevant click behaviors) have limited predictive significance for learning states. Dimensionality reduction methods retain core information and remove redundant features through nonlinear mapping.

[0019] For feature fusion and embedding modeling, the features of different modalities are complementary and need to be fused through the multi-head attention mechanism (MHA). MHA calculates the correlation weights between modalities to achieve dynamic interaction and integration of cross-modal features. The specific steps are as follows: The extracted features are constructed into feature matrices, namely, physiological data feature matrix, emotional data feature matrix, and behavioral data feature matrix. Given the feature matrices of the three modalities , and , respectively represent the learning characteristics of students in three different modalities, such as physiological data, emotional data, and behavioral data. The fusion process applies the MHA mechanism multiple times, handles the interaction between modalities in each operation, and retains the important information of each modality: , where the fusion of the first pair of modal features, Indicates that from physiological data ( ) and sentiment data ( ), is the fusion of the second pair of modal features, representing the sentiment data ( ) and behavioral data ( ), is the fusion of the third pair of modal features, indicating that ) and physiological data ( ) The fusion of modalities can effectively capture the similarities and differences between modalities through the interaction of queries, keys, and values; The fused feature matrix is ​​input into the GraphEmbeddingModel to generate a global embedding vector E of the student’s learning status, thereby representing the student’s comprehensive learning status based on multimodal features throughout the learning process: ,in, is the feature matrix after MHA fusion, which contains information from multiple modalities; G is the knowledge graph structure, which represents the relationship between different knowledge points in the learning task. It can be a graph structure based on subject knowledge, such as the hierarchical relationship between mathematical knowledge points; N is the number of students, that is, the number of students in the data set; is the dimension of the embedding vector, i.e., the representation of each student’s learning state in a low-dimensional space; The goal of the graph embedding model is to embed the multimodal features of students into a low-dimensional space so that the embedded features can effectively represent the students’ learning status and knowledge mastery. The embedding calculation process is as follows: Initialize the feature matrix. The initial features of each student are composed of the fused modal features, which represent the learning status of the student in different dimensions. Construct an adjacency matrix. The adjacency matrix A of the knowledge graph is used to represent the relationship between knowledge points or the relationship between students. Perform feature propagation and update student features through graph convolution operations (GCN) or other graph neural network techniques: ,in represents the feature matrix of the kth layer, W is the learned weight matrix, is the activation function; After multiple layers of graph convolution, the final embedding matrix E is the comprehensive learning state of each student.

[0020] Graph embedding can utilize the relationship between nodes in the knowledge graph to capture the mutual influence of students in the process of knowledge mastery and learning, thereby generating a low-dimensional embedding vector for each student. This vector contains the student's overall learning status, which not only reflects his performance in a single modality, but also integrates the interaction information between different modalities and knowledge points.

[0021] Through data collection and preprocessing, feature extraction and dimensionality reduction, feature fusion and embedding modeling, a complete process from raw data to student learning status representation is constructed to solve the heterogeneous problems of multimodal data in time, dimension and information interaction.

[0022] Step 2: construct and update the dynamic knowledge graph in real time. The dynamic knowledge graph is constructed and updated in real time to dynamically capture the knowledge points mastered by students through their learning data, and model the relationship between knowledge points as a graph structure. The specific steps are as follows: Initialize knowledge points and associations, extract knowledge points and dependencies between knowledge points from teaching resources; Teaching resources refer to various materials and tools that support teaching activities, such as textbooks, courseware, exercise banks, syllabus, etc. Knowledge points refer to specific content or concepts that need to be learned in a course, which can be a single concept, principle, theorem, etc. For example, "algebraic equations" in mathematics; Define nodes, knowledge points It is the basic unit of the graph, each node Represents a specific teaching knowledge point M. For example, "Solution of quadratic equations" is a mathematical knowledge point, and the node set is obtained from the course textbook, question bank, and structured analysis of the syllabus; Define the edge and calculate the weight. The edge represents the relationship between knowledge points. Use the co-occurrence probability to mine the previous and next dependencies between knowledge points from the question bank or textbooks. The formula is: ,in, It is a knowledge point and The number of times the same topic or chapter appears at the same time, the probability of co-occurrence Indicates the strength of association between knowledge points, as the weight of the edge, which is the knowledge point The number of occurrences, It is a knowledge point Number of occurrences; The definition of nodes reflects the completeness of the knowledge point set, while the weight of edges describes the association strength and order of knowledge points.

[0023] Model the students’ knowledge mastery status and calculate their mastery status of each knowledge point through their learning behavior data , used to initialize and update node attributes in the graph in real time; Initialize the knowledge point state, initial state Based on the matching calculation between student feature embedding and knowledge points, the dot product method is used to measure the student embedding vector and knowledge point embedding vector The matching degree: ,in, is a learnable mapping matrix, is the bias term, T represents the transposition operation, is the Sigmoid function; Provide student performance feedback updates and dynamically update knowledge point status based on student performance in tests and assignments: ,in, Indicates that students are at the knowledge point The test scores of Indicates the dependency strength between knowledge points, is the learning rate; By matching the knowledge point embedding with the student status embedding, the student’s knowledge mastery level is initialized, and the mastery status is dynamically adjusted to reflect the real-time learning results. For example, a student’s high score on “function graph” will improve his or her mastery status, and vice versa.

[0024] It should be noted that the students’ mastery of each knowledge point is dynamic and needs to be modeled in combination with historical data and real-time learning behavior. The current knowledge status of students can be quantified by matching the embedding vector of students and the embedding vector of knowledge points. In the test feedback, the students’ performance on tasks related to knowledge points (such as the correct answer rate and error distribution) is used to update the mastery status. In the update formula, Considering the degree of dependency between knowledge points: if yes The prerequisite knowledge points, The mastery of The update range will be limited.

[0025] According to the students' mastery of the knowledge points and changing dynamically with the completion of the learning tasks, the matching degree of the embedding vector provides the initial state, while the dynamic feedback update makes the state more reflective of the students' actual mastery.

[0026] There is a strong dependency between knowledge points. Mastering a certain knowledge point may have a positive impact on the mastery of related knowledge points. Therefore, a graph attention network (GAT) is used to propagate knowledge point features and update the feature representation of each knowledge point: , where the attention weight Through knowledge points and The association strength and current feature representation calculation: ,in, and It can be the student's learning state vector, the feature vector of a knowledge point, or an embedding vector generated by a neural network. Sim is a similarity measurement function used to calculate the similarity between the input feature representation (such as the student's state vector) and the knowledge point representation, such as dot product, cosine similarity, etc. N is the total number of knowledge points; The graph attention network dynamically adjusts the weight of feature propagation by calculating the contribution of neighboring nodes to the current node. For example, if a student has a good grasp of "function extreme value" and "derivative application" is its directly related knowledge point, then the "function extreme value" state will have a positive impact on the "derivative application" state; In the information-based teaching scenario, knowledge point feature propagation simulates the "knowledge flow" of teaching content. For example, students gradually master a set of interrelated knowledge points during the learning process. This process can be modeled through feature propagation, which enhances the dynamic adaptability of the knowledge graph. Based on students' latest learning performance and propagation characteristics, the node attributes and edge weights of the knowledge graph are adjusted in real time. The students' real-time performance is combined with the propagation characteristics of knowledge points to achieve a comprehensive assessment of the degree of mastery of knowledge points. The dynamic adjustment of edge weights reflects the students' learning trajectory on different knowledge points. For example, if students repeatedly involve "function graph" and "derivative formula" in certain questions, the system will automatically increase the edge weight between these two knowledge points, prompting the system to pay more attention to this type of knowledge point combination in the future. The finally updated knowledge graph will be used for subsequent learning path optimization.

[0027] In the construction and updating of dynamic knowledge graphs, from knowledge point initialization to feature propagation and then to real-time updating, the logic clearly covers the structured management of teaching content and the dynamic adjustment of students' learning status. This process transforms the static teaching knowledge structure into a dynamic graph that can reflect learning performance in real time, providing support for the optimization of information-based teaching.

[0028] Step 3: Perform personalized learning path optimization driven by reinforcement learning. By modeling the relationship between students' dynamic learning status and knowledge points, an adaptive learning path planning mechanism is designed. The dynamic knowledge graph is used as input, combined with students' real-time learning feedback, to generate a personalized learning path that can be continuously optimized. The specific steps are as follows: The design of a reinforcement learning model requires clear components, including state space, action space, reward function, and strategy update method. Each component is closely related to the scenario. The specific steps are as follows: State Space: State Contains the student's current learning status and node features of the dynamic knowledge graph. The student's learning status is represented by its feature vector Indicates that the node features of the dynamic knowledge graph are updated by the knowledge point features in step 2 Therefore, the spatial state is defined as: , where K represents the knowledge point set of the current learning path; Action Space: Action Indicates the recommended learning tasks, which can be knowledge points , a set of related exercises, or a teaching video. The action set is dynamically generated based on the node relationship in the dynamic knowledge graph and the student's knowledge mastery status. For example, if a knowledge point The mastery level is low and is not consistent with the knowledge points If there is a strong dependency, knowledge points are recommended first. As a learning task; The reward function is used to measure the effect of the learning task and define the immediate reward The trade-off between the improvement of knowledge mastery and time cost after completing the learning task: ,in, and Represents the knowledge points after and before the task respectively The mastery state, T represents the time required to complete the task, and is the weight coefficient, which is used to balance learning efficiency and time cost; The reinforcement learning model captures the student's current learning situation through state, provides flexible task selection through action, and evaluates the effectiveness of learning tasks through reward function, forming a dynamic closed loop. The state and action are input into the reinforcement learning framework by the dynamic knowledge graph and the student's learning performance.

[0029] Perform learning path optimization strategy. In reinforcement learning, strategy The probability distribution of selecting action a in state s is defined, and the optimization goal of the strategy is to maximize the cumulative return Improve the effectiveness of learning paths; The policy function is represented by a neural network, and the input is the state , the output is the action probability distribution The Xavier method is used to initialize the parameters of the policy network to avoid initial gradient explosion or disappearance; The PPO algorithm (Proximal Policy Optimization) is used to optimize the policy network. PPO maintains the stability and sample efficiency of policy updates by limiting the magnitude of policy updates. The objective function is: ,in, represents the expected value of the samples sampled at different time steps t, is the probability ratio between the current strategy and the old strategy, is the advantage function, which means that in the state Take action ϵ is a hyperparameter that limits the magnitude of policy updates relative to how good the behavior is on average. Through back propagation, the PPO algorithm updates the parameters of the policy network to maximize the objective function. The optimization process includes: Sample the actions taken by the student in each state from the environment, as well as the corresponding rewards and state transition information, and calculate the advantage function to capture the relative merits of the actions; By maximizing the objective function , update the parameters of the policy network, using the Adam optimizer or other optimizers for parameter update.

[0030] Strategy optimization maximizes the cumulative return of learning effects while maintaining the stability and efficiency of model training. For example, in a teaching scenario, the system dynamically adjusts the recommended learning tasks to maximize the efficiency of students' knowledge mastery. The optimized strategy provides a recommended task sequence for a personalized learning path.

[0031] Based on the optimized strategy , execute the recommended learning tasks at each time step and collect students’ learning feedback; The recommended tasks include task content (such as question sets, teaching videos) and task objectives (such as improving the mastery of knowledge points). After students complete the task, the task completion time T, the correct answer rate R, and the status changes of related knowledge points are recorded in real time. ; Feedback is recorded and the student's task completion data is used as the basis for updating the state of the next time step. For example, if the knowledge point after the task If the mastery of the knowledge point is significantly improved, the recommendation task in the next time step will give priority to the knowledge point Related advanced knowledge points; task completion feedback Used to update knowledge graph and reinforcement learning status .

[0032] After each strategy is executed, the state and node state of the knowledge graph can be recalculated. The priority of the recommended tasks can be adjusted according to real-time data. For example, for nodes that students have not yet mastered but are highly relevant to the current knowledge points, the system will improve their ranking in the recommended task sequence; If a student's performance in a task is significantly below expectations (for example, the completion time is too long or the error rate is too high), the system will trigger a jump mechanism to directly recommend review tasks for related basic knowledge points to make up for knowledge blind spots, and provide the adjusted learning path sequence as the final output to students and the teaching system.

[0033] Through the design of reinforcement learning models, path optimization strategies, task execution and feedback, and dynamic path adjustment, a personalized learning path system capable of real-time optimization is constructed. Combining students' dynamic learning status and structured information of knowledge graphs, the accuracy and efficiency of teaching task recommendations are achieved through strategy updating and feedback mechanisms.

[0034] Step 4: Optimize teaching strategies driven by causal inference. By building a causal relationship model, we analyze the causal effect between teaching behaviors (such as teaching methods and task allocation) and learning outcomes (such as knowledge mastery and learning interest), thereby optimizing teaching strategies. The specific steps are as follows: The construction of the causal relationship model requires clarifying the causal path between teaching behavior and learning outcomes, and modeling the relationship between these variables through a structured approach. The causal variables are defined as follows: The independent variable X represents the intervening variable of teaching behavior, such as the frequency of classroom interaction, task allocation strategy, and the complexity of teaching content; The dependent variable Y represents the student's learning outcomes, such as the degree of mastery of knowledge points. , correct answer rate R, and learning emotion index E; Confounding variables Z represent factors that may affect both X and Y, such as students’ learning ability and the quality of teaching resources; Use Bayesian Network to automatically learn the causal graph structure G from historical teaching data, where nodes represent variables and edges represent causal relationships. The learning process uses the method of maximizing log-likelihood estimation to determine the network structure: ,in, Representation Node The parent node set of , N is the total number of nodes; The causal model models teaching behavior, learning outcomes, and potential confounding factors as a directed acyclic graph, clarifying the potential impact paths of different teaching strategies on learning outcomes. For example, the frequency of classroom interaction (X) indirectly affects the mastery of knowledge points by improving student participation (Z); The constructed causal graph G is input into the causal inference stage to quantitatively analyze the intervention effect of the teaching strategy.

[0035] Conduct quantitative analysis of intervention effects, using causal inference techniques to quantitatively analyze the intervention effects of teaching behaviors, thereby evaluating the contribution of different strategies to learning outcomes; To define the causal effect, assume Indicates the implementation of a certain teaching strategy. means that the strategy is not implemented, and the intervention effect is defined as: ,in, Indicates external forced intervention rather than an observed state of nature; For example, suppose the frequency of classroom questioning (X) has two settings: a lower frequency ( ) and higher frequencies ( ), by calculating the change in the correct rate of students’ answers (Y) after the frequency of questions is increased through intervention inference, we can know the specific impact of questions on students’ learning outcomes; Using the Do-Calculus rule, separate the estimated path of the intervention effect from the causal graph G. For example, if the path in the causal graph is X to Z and then to Y, the intervention effect can be decomposed into: , where P(Y|Z=z, X=x) represents the probability of Y under the conditions of Z=z and X=x; The intervention effect is broken down into direct effect and indirect effect. The direct effect is from X to Y, such as asking questions in class directly stimulating students' thinking ability; Indirect effect X to Z to Y, such as indirectly improving learning outcomes (Y) by increasing student concentration (Z); Quantitative analysis of intervention effects clarifies the advantages and disadvantages of different teaching strategies in mathematical form. For example, if a strategy does not significantly improve learning outcomes, it should be optimized or replaced.

[0036] According to the analysis results of the intervention effect, the teaching strategy is optimized and the teaching behavior is adjusted dynamically. The strategy optimization rules are: like (Effect threshold), it indicates a significant improvement in effect, and the strategy should be retained and promoted. , it indicates that the effect is not significant or negative, and the strategy is replaced or adjusted. If the analysis finds that high interaction frequency significantly improves the mastery of the knowledge point "determination of function monotonicity" ( ), but the effect on "drawing function graphs" is not obvious ( ), then reduce the intensity of interaction in the teaching of the next knowledge point; During the teaching process, the causal model is updated according to real-time data. For example, for students with a good foundation, the proportion of low-level tasks is reduced; for weak students, review and practice of basic knowledge points are increased.

[0037] Strategy optimization and dynamic adjustment ensure the flexibility and adaptability of teaching behavior, implement differentiated strategies for different groups of students, and make the allocation of teaching resources more targeted. The optimized strategies are used to generate new teaching plans and dynamically updated to the teaching system.

[0038] It should be noted that the threshold information in this embodiment is pre-set by professionals and will not be explained in detail here. Some parameter English letters in the embodiments have the same situation, but different meanings are explained when used, which will not be explained one by one here.

[0039] The present invention uses multimodal data collection and deep learning technology to accurately capture students' behaviors, emotions, and physiological states, combines multi-head attention mechanisms to achieve data fusion, generates accurate learning state features, and based on the constructed dynamic knowledge graph and graph convolutional network, updates students' knowledge mastery status in real time and adjusts the learning path to ensure personalized learning support, and then strengthens learning to optimize personalized learning paths, and adjusts learning tasks in real time with instant feedback to improve learning efficiency and experience. At the same time, the impact of teaching behavior on learning outcomes is analyzed through causal inference, the intervention effect is quantified, and the teaching strategy is scientifically adjusted to improve the teaching effect. This solution optimizes learning paths and teaching strategies through the combination of multimodal fusion, graph convolutional networks, and reinforcement learning, and has significant advantages in personalized learning support, improved teaching outcomes, and optimized learning efficiency.

[0040] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0041] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0042] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0043] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0044] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0045] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A teaching optimization method based on big data informationization, characterized by: The steps include: Collect students' behavior, emotions and physiological data, perform synchronous alignment and normalization, extract features and reduce dimensions based on deep learning, and fuse features of each modality through a multi-head attention mechanism; Construct a graph of knowledge points and dependencies, update students’ knowledge mastery status in real time, and adjust learning paths and knowledge mastery status based on student performance and knowledge propagation characteristics of graph convolutional networks; Define state space and action space, optimize personalized learning paths based on immediate feedback and reinforcement learning, and adjust tasks in real time; Based on causal inference, we analyze the impact of teaching behavior on learning outcomes, establish a causal relationship model to quantify the intervention effect, and adjust the teaching strategy.

2. The method for optimizing teaching based on big data informationization according to claim 1, characterized in that: Collect student behavior, emotion, and physiological data, perform synchronous alignment and normalization, extract features and reduce dimensions based on deep learning. The specific steps are as follows: The collected behavioral data include clickstream logs, task completion time, answer accuracy, and learning frequency; Collect emotional data, capture facial expressions and gestures through cameras, and collect speech through microphones; Collect physiological data and obtain students' physiological signals through wearable devices. Physiological signals include heart rate changes and skin conductance. For the collected data of different time granularities, the time window slicing method is used for alignment and the logarithmic scaling function is used for normalization; Perform behavioral data feature extraction to extract complex patterns in time series, including the correlation between task completion time and click behavior; Extract emotional data features and extract emotional features from facial expressions and voice signals through convolutional neural networks. Emotional features include changes in attention and stress. Perform physiological data feature extraction and use variational autoencoder to extract feature vectors of physiological signals; For the extracted features, nonlinear dimensionality reduction method is used for dimensionality reduction.

3. The method for optimizing teaching based on big data informationization according to claim 2 is characterized in that: The multi-head attention mechanism is used to fuse the features of each modality. The specific steps include: The collected data features are constructed into physiological data feature matrix, emotional data feature matrix, and behavioral data feature matrix, and the feature matrices are fused through a multi-head attention mechanism; The fused feature matrix is ​​input into the graph embedding model to generate a global embedding vector of the student’s learning status.

4. The method for optimizing teaching based on big data informationization according to claim 3 is characterized in that: Construct a knowledge point and dependency graph to update students’ knowledge mastery status in real time. The specific steps are as follows: Initialize knowledge points and associations, extract knowledge points and dependencies between knowledge points from teaching resources; Define nodes, knowledge points It is the basic unit of the graph, each node Indicates specific teaching knowledge points; Define the edge and calculate the weight. The edge represents the relationship between knowledge points. Use the co-occurrence probability to mine the previous and next dependencies between knowledge points from teaching resources. The formula is: ,in, It is a knowledge point and The number of times it appears in the same question at the same time, It is a knowledge point The number of occurrences, It is a knowledge point Number of occurrences, co-occurrence probability Indicates the strength of association between knowledge points and serves as the weight of the edge; Initialize the knowledge point state, initial state Based on the matching calculation between student feature embedding and knowledge points, the dot product method is used to measure the student embedding vector and knowledge point embedding vector The matching degree: ,in, is the mapping matrix used for learning, is the bias term, T represents the transposition operation, is the Sigmoid function; Provide student performance feedback updates and dynamically update knowledge point status based on student performance in tests and assignments : ,in, Indicates that students are at the knowledge point The test scores of Representing knowledge points , The strength of the dependence between is the learning rate.

5. The method for optimizing teaching based on big data informationization according to claim 4 is characterized in that: Based on the student performance and the knowledge propagation characteristics of the graph convolutional network, the learning path and knowledge mastery status are adjusted, including the following steps: Use the graph attention network to propagate the knowledge point features and update the feature representation of each knowledge point: , where the attention weight Through knowledge points and The association strength and current feature representation calculation: ,in, and is the student's learning state vector and the feature vector of the knowledge point, sim is the similarity measurement function, and N is the total number of knowledge points; Based on students' latest learning performance and communication characteristics, the node attributes and edge weights of the knowledge graph are adjusted in real time.

6. The method for optimizing teaching based on big data informationization according to claim 5 is characterized by: Define the state space and action space, optimize the personalized learning path based on immediate feedback and reinforcement learning, and adjust the task in real time, including the following steps: Define the state space. The state of the state space includes the student's current learning state and the node features of the dynamic knowledge graph. The student's learning state is represented by a feature vector, and the node features of the dynamic knowledge graph are represented by the updated knowledge point features. Define an action space, where the actions in the action space represent recommended learning tasks. The action set is dynamically generated based on the node relationships in the dynamic knowledge graph and the student's knowledge mastery status. Define the immediate reward as the trade-off between the improvement in knowledge mastery and the time cost after completing the learning task as the reward function; The state captures the student's current learning situation, the action provides task selection, and the reward function evaluates the learning task; Input the state into the policy function, output the action probability distribution, and use the PPO algorithm to optimize the policy function; Based on the optimized strategy, the recommended learning task is executed at each time step, and the students' learning feedback is collected and the task is adjusted according to the learning feedback.

7. The method for optimizing teaching based on big data informationization according to claim 6 is characterized by: Based on causal inference, we analyze the impact of teaching behavior on learning outcomes, establish a causal relationship model to quantify the intervention effect, and adjust the teaching strategy, including the following steps: The causal variables were defined, and the independent variables represented the intervening variables of teaching behaviors, including the frequency of classroom interactions, task allocation strategies, and the complexity of teaching content; The dependent variables represent the students’ learning outcomes, including the mastery of knowledge points and the accuracy of answering questions; Confounding variables represent factors that affect both the independent and dependent variables; Use Bayesian networks to automatically learn causal graph structures from historical teaching data. Causal graph nodes represent variables, and edges represent causal relationships. The learning process determines the network structure based on the method of maximizing log-likelihood estimation. The constructed causal diagram is input into the causal inference stage to analyze the intervention effect of the teaching strategy and define the causal effect; According to the Do-Calculus rule, the estimated path of the intervention effect is separated from the causal diagram; Based on the analysis results of the intervention effects, optimize teaching strategies and dynamically adjust teaching behaviors.

Images