A teaching material content visualization system and method based on data analysis

Abstract

The present invention relates to the field of data processing technology, and specifically discloses a teaching material content visualization system and method based on data analysis. By dynamically updating the knowledge graph and optimizing the learning path, combined with reinforcement learning and knowledge graph, it is possible to find the optimal learning path that best suits the student's current learning situation, and help students strengthen their weak knowledge points. In addition, by visually displaying the optimal learning path, it is possible to enhance students' learning interest and intuitive understanding. Teachers can also use visualization data to accurately grasp students' learning progress and weak links, further optimize teaching content and strategies, and thus improve the overall teaching quality.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a teaching material content visualization system and method based on data analysis. Background Art

[0002] With the development of information technology, the use of digital teaching materials is becoming more and more common. However, traditional teaching materials are usually text-based and lack intuitive visual display methods. It is difficult to fully mobilize students' learning interest and improve learning efficiency, and there is no visual display of students' mastery of knowledge points. At the same time, with the continuous increase in teaching content, teachers need to spend a lot of time on content organization and analysis during lesson preparation and teaching.

[0003] In view of this, there is a need to provide a teaching material content visualization system and method based on data analysis. Summary of the invention

[0004] The purpose of the present invention is to provide a textbook content visualization system based on data analysis to solve the following technical problems: through data analysis technology, combined with knowledge graphs and reinforcement learning, the textbook content is structured and processed, and the optimal learning path of knowledge points is displayed in a visual way to effectively improve learning efficiency and teaching quality.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A teaching material content visualization system based on data analysis, the system comprising:

[0007] A first acquisition module is used to acquire digital teaching material content;

[0008] A first data analysis module, used to perform data analysis based on the digital teaching material content to obtain semantic analysis results, knowledge point extraction results and structural processing results of the digital teaching material content;

[0009] A graph construction module, used to construct an initial knowledge graph based on the semantic analysis results, knowledge point extraction results and structured processing results;

[0010] The second acquisition module is used to acquire electronic test papers of multiple students;

[0011] The second data analysis module is used to perform data analysis based on the electronic test paper to obtain the scores of each student in the test questions corresponding to different knowledge points;

[0012] A graph updating module, used to update the initial knowledge graph based on the score to obtain a current knowledge graph;

[0013] A first determination module is used to determine the degree of topological change of the current knowledge graph compared to the initial knowledge graph in the topological structure;

[0014] The second determination module is used to determine the reward function of reinforcement learning based on the degree of topological change:

[0015] ;

[0016] Among them, R t is the reward function of the reinforcement learning, α is the learning rate, γ is the discount factor, is the student’s learning progress at the current time t, T cha (G t ) is the degree of topological change, E t is the student's error rate;

[0017] A learning path determination module, used to obtain the reward and punishment values ​​of the reinforcement learning reward function, and determine the optimal learning path through a preset Q value update formula;

[0018] A visualization module is used to determine and display visualized teaching content based on the current knowledge graph and the optimal learning path; wherein the visualized teaching content at least includes knowledge points corresponding to the optimal learning path displayed in a graphical structure.

[0019] As a further solution of the present invention, the preset Q value update formula is determined by the following method:

[0020] Based on the current knowledge graph, determine the state definition S for the reinforcement learning t =(K t , P t , E t , L t , G t ); where K t represents the set of knowledge points that students have mastered, P t Indicates the student’s current learning progress, E t represents the type and frequency of students’ errors, L t represents the confidence distribution of students on knowledge points, G t is the current knowledge graph;

[0021] Determine the action definition a of the reinforcement learning t =Q(S t , a1); where a t In state S t Next, select the Q value of action a1;

[0022] Based on the state definition S t and the action definition at , determine the preset Q value update formula as:

[0023]

[0024] Among them, α is the learning rate, γ is the discount factor, is the next state S t+1 The maximum Q value of all possible actions.

[0025] As a further solution of the present invention, the atlas updating module is further used for:

[0026] Get the student's learning progress P after each time step t t ;

[0027] Based on the student's learning progress t and the current action a t , for the current knowledge graph G t to update.

[0028] As a further solution of the present invention, the learning progress P of the student is based on the learning progress P of the student. t and the current action a t , for the current knowledge graph G t Updates include:

[0029] Based on the student's learning progress t , determine the new knowledge points mastered by the students;

[0030] Add the new knowledge point to the current knowledge graph G t ;

[0031] Based on the current action a t , determine existing knowledge points related to the new knowledge points;

[0032] Connect the new knowledge point with the existing knowledge point to obtain a new edge, and based on the student's learning progress P t , assign a weight coefficient to the newly added edge; wherein the weight coefficient is used to indicate the difficulty of students mastering the new knowledge point.

[0033] As a further solution of the present invention, the atlas updating module is further used for:

[0034] Obtaining N-hop neighbor nodes of the node of the newly added edge; wherein N is a positive integer less than a preset value;

[0035] Based on the association between the knowledge point corresponding to the N-hop neighbor node and the knowledge point corresponding to the node of the newly added edge;

[0036] Based on the association degree and the weight coefficient corresponding to the newly added edge, the weight coefficient of the edge between the N-hop neighbor nodes is updated.

[0037] As a further solution of the present invention, the step of determining and displaying the visual teaching content based on the current knowledge graph and the optimal learning path includes:

[0038] Obtaining multiple decision paths in the reinforcement learning process;

[0039] Based on the multiple decision paths, acquiring knowledge points corresponding to the multiple decision paths from the current knowledge graph;

[0040] Determining a change path of the knowledge point based on the knowledge points corresponding to the multiple decision paths and time nodes of the actions corresponding to the multiple decision paths;

[0041] The change path of the knowledge point is taken as a part of the visual teaching content.

[0042] As a further solution of the present invention, the visualization module is further used for:

[0043] Obtaining reward function values ​​at multiple time steps during the reinforcement learning process;

[0044] Based on the reward function values ​​at the multiple time steps, plotting a learning efficiency curve;

[0045] Establish and display the correlation between the learning efficiency curve and the corresponding knowledge points in the reinforcement learning process.

[0046] A teaching material content visualization method based on data analysis, the method comprising:

[0047] Access to digital textbook content;

[0048] Performing data analysis based on the digital textbook content to obtain semantic analysis results, knowledge point extraction results, and structural processing results of the digital textbook content;

[0049] Based on the semantic analysis results, knowledge point extraction results and structured processing results, construct an initial knowledge graph;

[0050] Get electronic test papers of multiple students;

[0051] Perform data analysis based on the electronic test paper to obtain the scores of each student in the test questions corresponding to different knowledge points;

[0052] Based on the score, the initial knowledge graph is updated to obtain a current knowledge graph;

[0053] Determine the degree of topological change of the current knowledge graph compared to the initial knowledge graph in the topological structure;

[0054] Based on the degree of topological change, the reward function of reinforcement learning is determined:

[0055]

[0056] Among them, R t is the reward function of the reinforcement learning, α is the learning rate, γ is the discount factor, is the student’s learning progress at the current time t, T cha (G t ) is the degree of topological change, E t for students’ error rates;

[0057] Obtaining the reward and punishment values ​​of the reinforcement learning reward function, and determining the optimal learning path through a preset Q-value update formula;

[0058] Based on the current knowledge graph and the optimal learning path, the visualized teaching content is determined and displayed; wherein the visualized teaching content at least includes knowledge points corresponding to the optimal learning path displayed in a graphical structure.

[0059] Beneficial effects of the present invention: The teaching material content visualization method based on data analysis provided by the present invention can realize personalized teaching, and can improve students' learning effects in a targeted manner by dynamically updating the knowledge graph and optimizing the learning path. At the same time, by combining reinforcement learning and knowledge graph, it is possible to find the optimal learning path that best suits the students' current learning situation, and help students strengthen their weak knowledge points. In addition, by graphically displaying the optimal learning path, it is possible to enhance students' learning interest and intuitive understanding. Teachers can also use visualized data to accurately grasp students' learning progress and weak links, and further optimize teaching content and strategies, thereby improving the overall teaching quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] The present invention will be further described below in conjunction with the accompanying drawings.

[0061] Figure 1 It is a structural schematic diagram of a teaching material content visualization system based on data analysis of the present invention;

[0062] Figure 2 It is a flow chart of the teaching material content visualization method based on data analysis of the present invention. DETAILED DESCRIPTION

[0063] 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.

[0064] See also Figure 1 As shown, the present invention is a teaching material content visualization system based on data analysis. In some embodiments, the system 100 may include the following modules.

[0065] The first acquisition module 101 is used to acquire digital teaching material content.

[0066] The first data analysis module 102 is used to perform data analysis based on the digital teaching material content to obtain semantic analysis results, knowledge point extraction results and structural processing results of the digital teaching material content.

[0067] The graph construction module 103 is used to construct an initial knowledge graph based on the semantic analysis results, knowledge point extraction results and structured processing results.

[0068] The second acquisition module 104 is used to acquire electronic test papers of multiple students.

[0069] The second data analysis module 105 is used to perform data analysis based on the electronic test paper to obtain the scores of each student corresponding to the test questions at different knowledge points.

[0070] The graph updating module 106 is used to update the initial knowledge graph based on the score to obtain the current knowledge graph.

[0071] The first determination module 107 is used to determine the degree of topological change of the current knowledge graph compared to the initial knowledge graph in the topological structure.

[0072] The second determination module 108 is used to determine the reward function of reinforcement learning based on the degree of topological change:

[0073] (1);

[0074] Among them, R t is the reward function of the reinforcement learning, α is the learning rate, γ is the discount factor, is the student’s learning progress at the current time t, T cha (G t ) is the degree of topological change, E t is the student's error rate.

[0075] The learning path determination module 109 is used to obtain the reward and punishment values ​​of the reinforcement learning reward function and determine the optimal learning path through a preset Q value update formula.

[0076] The visualization module 110 is used to determine and display the visualized teaching content based on the current knowledge graph and the optimal learning path; wherein the visualized teaching content at least includes knowledge points corresponding to the optimal learning path displayed in a graphical structure.

[0077] about Figure 1 For more information on the modules of the data analysis-based textbook content visualization system, see Figure 2 Detailed description.

[0078] Figure 2 FIG. 2 is a flow chart of a method for visualizing teaching material content based on data analysis of the present invention. In some embodiments, process 200 may be executed by a processor or the system 100. Figure 2 As shown, process 200 may include the following operations.

[0079] Step 201, obtaining digital teaching material content.

[0080] Digital teaching material content refers to teaching material content presented in electronic form, which may include multimedia forms such as text, images, videos, audio, etc. In some embodiments, the digital teaching material content may include course chapters, knowledge points, examples, exercises and related auxiliary materials.

[0081] In some embodiments, the system may obtain digital teaching material content by accessing a teaching material database or an online teaching material library through automatic crawling.

[0082] Step 202: Perform data analysis based on the digital teaching material content to obtain semantic analysis results, knowledge point extraction results and structured processing results of the digital teaching material content.

[0083] Data analysis refers to the processing and analysis of input data, such as statistics, mining, and analysis. Through data analysis, useful information can be extracted from the content of digital teaching materials.

[0084] The results of semantic analysis refer to the semantic understanding of the text in the digital teaching materials through natural language processing (NLP) technology, and the extraction of core concepts, relationships, structures and other information.

[0085] The knowledge point extraction results refer to the knowledge points extracted from the textbooks, for example, the content in the textbooks that has strong concepts and clear learning objectives.

[0086] Structuring the results means that the textbook content is organized according to certain rules or structures to make it more suitable for further analysis and processing. For example, the textbook content is converted into structured forms such as data tables, nodes, and charts.

[0087] In some embodiments, data analysis may include multiple operations such as preprocessing, semantic analysis, knowledge point extraction, and structured processing. Among them, semantic analysis can be the use of natural language processing (NLP) technology to identify core knowledge points, concepts, and relationships in teaching materials. For example, dependency syntax analysis is used to extract keywords and topics in sentences. Knowledge point extraction can be the extraction of knowledge points in teaching materials through rules or models (such as topic models, named entity recognition). Structured processing can be the conversion of the content of the teaching materials into structured data forms, such as tables or graphics.

[0088] Step 203, constructing an initial knowledge graph based on the semantic analysis results, knowledge point extraction results and structured processing results.

[0089] A knowledge graph is a semantic network that reveals the relationships between entities (or objects). Nodes in a graph represent entities. Nodes can be of multiple types, called node types, which are used to indicate various entities. Edges in a graph represent relationships. Edges can also be of multiple types, called edge types, which are used to indicate various relationships. Entities can refer to things in the real world, such as people, place names, concepts, drugs, companies, and so on. Relationships can be used to express the connection between different entities, for example, Zhang San and Li Si are "friends", social accounts and mobile terminals have a login relationship, and so on.

[0090] A knowledge graph can be a directed graph or an undirected graph, that is, the edges in a knowledge graph can be directed or undirected. Directed edges can be unidirectional or bidirectional to indicate the directionality of the relationship. When the knowledge graph is an undirected graph, the edges can indicate that the relationship has no directionality or that the relationship is bidirectional (such as a "friend" relationship). The edge pointing to a node can be called the incoming edge of the node, and the edge pointing out of a node (i.e. pointing to other nodes) can be called the outgoing edge of the node.

[0091] In this embodiment, the knowledge points and structured processing results in the knowledge point extraction results can be used as nodes of the knowledge graph, and edges can be added between the nodes according to the semantic analysis results to obtain the initial knowledge graph. The edges of the initial knowledge graph can be used to represent the relationship between different knowledge points, the difficulty of mastering from one knowledge point to another, the importance of mastering knowledge points, etc.

[0092] In some embodiments, the system can use the results of semantic analysis and knowledge point extraction to establish an initial knowledge graph. Each knowledge point is a node in the graph, and the dependencies and associations between knowledge points are used as edges to connect different nodes.

[0093] Step 204, obtaining electronic test papers of multiple students.

[0094] Electronic test papers refer to test or assessment content generated and presented in electronic form. Electronic test papers may include multiple-choice questions, true-or-false questions, short-answer questions, etc., and record students' answers and scores.

[0095] In some embodiments, students' electronic test papers can be automatically collected by integrating with a student management system or a learning platform.

[0096] Step 205, performing data analysis based on the electronic test paper to obtain the scores of each student for test questions corresponding to different knowledge points.

[0097] The score situation refers to the answers of each student on different knowledge points in the electronic test paper. It includes information such as the student's score, wrong answers, etc. The score situation can reflect the student's mastery of each knowledge point.

[0098] In some embodiments, the system can perform data analysis based on the answers of the students' answer sheets and calculate the scores of each knowledge point.

[0099] Step 206: Update the initial knowledge graph based on the score to obtain the current knowledge graph.

[0100] In some embodiments, the system can mark the nodes in the initial knowledge graph according to the scores (for example, mark the knowledge point as mastered / not mastered and correspond to each student). Afterwards, the marks between knowledge points can be updated according to the scores of students on different knowledge points (for example, from "not mastered" to "mastered", etc.).

[0101] Step 207, determining the degree of topological change in the topological structure of the current knowledge graph compared to the initial knowledge graph.

[0102] The degree of topological change refers to the degree of structural change of nodes and edges in the knowledge graph. The degree of topological change can be reflected by the change of the relationship or connection between nodes. Topological change can reflect the progress of students' knowledge mastery and relationship adjustment in the learning process.

[0103] In some embodiments, the system can calculate the state changes of nodes in the knowledge graph (the transition between mastered and not mastered) and obtain the degree of topological change through a graph analysis algorithm (such as the edit distance of a graph).

[0104] Step 208: Determine a reward function for reinforcement learning based on the degree of topological change.

[0105] The reinforcement learning reward function refers to a function used to evaluate the quality of a student's learning behavior. The reinforcement learning reward function guides the student's learning by giving rewards and punishments. The reward function can generate a specific reward value based on factors such as the student's learning progress, error rate, topological changes, etc. In some embodiments, the reinforcement learning reward function can be shown as the following formula (1).

[0106] (1);

[0107] Among them, R t is the reward function of the reinforcement learning, α is the learning rate, γ is the discount factor, is the student’s learning progress at the current time t, T cha (G t ) is the degree of topological change, E t is the student's error rate.

[0108] Step 209, obtaining the reward and punishment values ​​of the reinforcement learning reward function, and determining the optimal learning path through a preset Q value update formula.

[0109] In the preset Q-value update formula, Q-value (Quality Value) is the expected return (e.g., the score of the test questions corresponding to the knowledge point) for a student to take a certain action (e.g., learning a certain knowledge point) in a certain state in reinforcement learning. The Q-value update formula is used to adjust the learning strategy according to the student's performance, thereby determining the optimal learning path.

[0110] In some embodiments, the Q-learning reinforcement learning algorithm can be used to update the Q value according to the feedback of the reward function, and the optimal learning path can be selected according to the maximum Q value.

[0111] In some embodiments, the preset Q value update formula can be determined in the following manner.

[0112] S10, based on the current knowledge graph, determining a state definition S for the reinforcement learning t =(K t , P t , E t , L t , G t ).

[0113] In the state definition S t In, K t represents the set of knowledge points that students have mastered, indicating the knowledge points that students have successfully understood and mastered at the current time point; P t Indicates the student's current learning progress, which refers to the student's progress in the textbook or learning path, which can be the chapters or knowledge points that have been learned, etc.; E tIndicates the types and frequency of students' mistakes, indicating the types and frequency of mistakes made by students in answering questions or in the learning process, which can reflect the students' weak links; L t It indicates the distribution of students’ confidence in knowledge points, and the confidence of students in mastering each knowledge point, which can be obtained through students’ self-evaluation or feedback; G t The current knowledge graph can reflect students' mastery of each knowledge point and the relationship between knowledge points.

[0114] In some embodiments, the system can calculate the student's current mastered knowledge points K based on the student's learning records, test paper data, wrong question analysis and other information. t , learning progress P t , error type and frequency E t , and the student confidence distribution L t , and extract the knowledge graph G at the current time point from the database or learning management platform t , get the current state S t .

[0115] S11, determine the action definition a of the reinforcement learning t =Q(S t , a1).

[0116] Among them, a t In state S t Next, select the Q value of action a1. Action definition a t In state S t Under the circumstance, the specific behavior selected based on the reinforcement learning strategy can be to recommend students to learn a specific knowledge point or take a certain learning path.

[0117] In some embodiments, in state S t In this case, the system can calculate all possible actions a1 through reinforcement learning algorithms (such as Q learning) and assign a Q value to each action. The action selection is based on the Q value function, that is, the action with the largest Q value is selected as the current optimal strategy.

[0118] For example, the system can recommend learning paths or practice tasks based on the student's current learning needs (such as weak knowledge points or confidence distribution). Each possible action (such as choosing to learn a certain knowledge point or do a certain exercise) is assigned a Q value.

[0119] S12, based on the state definition S t and the action definition a t , determine the preset Q value update formula (2) as:

[0120] (2);

[0121] Among them, α is the learning rate, γ is the discount factor, is the next state S t+1 The maximum Q value of all possible actions.

[0122] The Q value represents the t Next, take action a t After that, the total reward expected to be obtained (for example, how many points can be obtained by learning).

[0123] In some embodiments, the system can calculate the reward value R based on the student's current learning status and the action selected, and the system can calculate the reward value R through a feedback mechanism (such as the student's learning progress, knowledge point mastery, error type, etc.) t , and according to the Q value update formula, the current state, current action, feedback reward and the maximum Q value of possible future states are calculated to update the Q value. The updated Q value reflects the expected long-term benefits of taking an action in the current state and can be used to guide future learning path selection.

[0124] In this embodiment, by setting how to optimize the student's learning path through state definition, action selection and Q value update. By constructing the state definition St, the system can fully understand the student's learning progress, error type and knowledge mastery, so as to recommend the best learning strategy for the student. At the same time, through the Q value update formula, the system can continuously optimize its learning strategy and improve the teaching effect.

[0125] In some embodiments, during the process of reinforcement learning, the system can also update the current knowledge graph according to the results of reinforcement learning. For example, the system can obtain the learning progress P of the student after each time step t. t ; and, based on the student's learning progress P t and the current action a t , for the current knowledge graph G t to update.

[0126] The learning progress of students refers to the learning progress of students after each time step t. It can represent the number of knowledge points mastered by students in a specific time, the learning tasks completed, or the learning goals achieved. t It could be the number of exercises completed by students, the number of knowledge points mastered, or the learning progress measured according to the course schedule.

[0127] In some embodiments, the student's learning progress can be recorded in a course management system or learning platform. For example, the system can calculate and record the student's learning progress through data such as study time, number of completed tasks, and test scores.

[0128] Current action a t In state S t The system selects specific actions based on the reinforcement learning strategy. Action a t It may be a suggestion for students to learn a certain knowledge point, do a certain exercise, or skip certain content. In some embodiments, the current action a t It can be calculated through the reinforcement learning model. The system can t and the selected strategy to determine the most appropriate action. For example, if a student has difficulty learning a certain knowledge point, the student can be recommended to review the knowledge point.

[0129] In some embodiments, the current knowledge graph G t It can be continuously updated based on the student’s learning progress and error records. For example, if a student has mastered a certain knowledge point, the corresponding node in the graph will be updated to the mastered state. If a student frequently makes mistakes on a certain knowledge point, the node in the graph can be updated to not yet mastered.

[0130] In some embodiments, the learning progress of the student is based on t and the current action a t , for the current knowledge graph G t The update may include the following operations.

[0131] S20, based on the student's learning progress P t , determine the new knowledge points mastered by the students.

[0132] New knowledge points refer to the knowledge points that students currently master but have not yet been marked as “mastered” in the knowledge graph.

[0133] In some embodiments, new knowledge points can be determined by comparing with existing knowledge points and changes in learning progress. For example, the system can monitor the student's learning process, for example, based on the student's learning progress P t , track students’ learning activities (such as test scores, homework completion, etc.) in real time, and extract new knowledge points mastered by students. If students pass certain learning tasks (for example, pass a round of tests or exercises) and successfully master a certain knowledge point, then the knowledge point can be considered new.

[0134] S21, adding the new knowledge point to the current knowledge graph G t .

[0135] After the new knowledge point is determined, it can be added to the current knowledge graph as a new node. This node will establish a connection with other nodes in the graph, reflecting the relationship between the new knowledge point and the existing knowledge points.

[0136] S22, based on the current action a t , determine the existing knowledge points related to the new knowledge points.

[0137] In some embodiments, the current action a t Evaluate which existing knowledge points are related to new knowledge points. For example, if you recommend students to learn a new concept, you can determine which existing knowledge points are closely related to the new knowledge points based on the logical relationship between the concept and other concepts (such as dependencies, application relationships, etc.). Then, you can generate existing knowledge points related to the new knowledge points based on the logical relationship between the knowledge points (such as the course outline or teaching design).

[0138] S23, connecting the new knowledge point with the existing knowledge point to obtain a new edge, and based on the student's learning progress P t , assign a weight coefficient to the newly added edge, wherein the weight coefficient is used to indicate the difficulty for students to master the new knowledge point.

[0139] In some embodiments, the system can calculate the weight coefficient based on the student's performance when learning a new knowledge point. For example, if the student takes a long time to master the new knowledge point or has a low success rate in completing related tasks, the weight coefficient of the knowledge point is high, indicating that the knowledge point is difficult to master. The system can assign the calculated weight coefficient to the newly added edge to reflect the strength of the association between the new knowledge point and the existing knowledge point and the difficulty of mastering it.

[0140] In some embodiments, the system may also update the weight coefficients of the associated knowledge points according to the weight coefficients. Exemplarily, the specific method of updating may be implemented through the following operations.

[0141] S30, obtaining N-hop neighbor nodes of the node of the newly added edge; wherein N is a positive integer less than a preset value.

[0142] The nodes with newly added edges refer to the nodes newly added to the knowledge graph. The N-hop neighbor nodes refer to all nodes that can be reached from the nodes with newly added edges through at most N hops.

[0143] In some embodiments, the system may use a graph traversal algorithm, for example, starting from the node with the newly added edge, and using a graph traversal algorithm (such as breadth-first search) to find all neighbor nodes that can be reached through a maximum of N hops.

[0144] S31, based on the association between the knowledge point corresponding to the N-hop neighbor node and the knowledge point corresponding to the node of the newly added edge.

[0145] The knowledge points corresponding to the N-hop neighbor nodes refer to all the knowledge points that can be reached after N hops from the newly added edge nodes.

[0146] The knowledge points corresponding to the nodes of the newly added edges refer to the new knowledge points connected by the newly added edges, for example, the new content currently mastered by the students.

[0147] In some embodiments, the system can calculate the association between the N-hop neighbor nodes and the newly added edge nodes based on the relationship between the knowledge points. For example, the association can be calculated by semantic similarity calculation, co-occurrence analysis, dependency analysis, etc.

[0148] The degree of association is used to measure the strength or closeness of the association between two knowledge points.

[0149] S32: Update the weight coefficient of the edge between the N-hop neighbor nodes based on the association degree and the weight coefficient corresponding to the newly added edge.

[0150] The weight coefficient corresponding to the newly added edge is used to indicate the difficulty or importance of students mastering the new knowledge point.

[0151] The edge between N-hop neighbor nodes refers to the relationship edge between N-hop neighbor nodes, which can be used to represent the learning association between nodes.

[0152] In some embodiments, the system may update the weight coefficients of the edges between N-hop neighbor nodes according to the following formula (3).

[0153] (3)

[0154] Where W is the weight of the updated edge, α and β are weight coefficients set according to actual needs, and W 新增边 is the weight coefficient of the newly added edge, T 关联度 It is the association degree between the N-hop neighbor node and the newly added edge node.

[0155] The system can dynamically adjust the weights of edges based on the difficulty of students mastering new knowledge points and the learning relationship between N-hop neighbor nodes, so that the knowledge graph can more accurately reflect students' learning progress and the interdependence between knowledge points.

[0156] Step 210, based on the current knowledge graph and the optimal learning path, determine and display the visualized teaching content.

[0157] Visualized teaching content refers to teaching content displayed in the form of graphics, charts, videos, etc. Visualization means that students can perceive the corresponding teaching content through vision. In some embodiments, the visualized teaching content at least includes knowledge points corresponding to the optimal learning path displayed in a graphical structure. Displaying knowledge points in a graphical structure can more intuitively express the relationship between different knowledge points and provide students with a more understandable learning guide.

[0158] In some embodiments, the system can generate and display visual teaching content based on the updated knowledge graph and the optimal learning path.

[0159] In some embodiments, determining and displaying the visual teaching content based on the current knowledge graph and the optimal path may include the following operations.

[0160] S40, obtaining multiple decision paths in the reinforcement learning process.

[0161] The reinforcement learning process refers to the process in which students continuously choose learning paths and obtain feedback through interaction with the system during the learning process. Through the reinforcement learning algorithm, the system can record the decision paths taken by students.

[0162] Multiple decision paths refer to the node sequences of the knowledge graph that represent the different choices students make in different learning situations.

[0163] In some embodiments, the system may record all decision paths made by students during the learning process through a reinforcement learning algorithm, thereby obtaining the multiple decision paths.

[0164] S41, based on the multiple decision paths, obtaining knowledge points corresponding to the multiple decision paths from the current knowledge graph.

[0165] The knowledge points corresponding to multiple decision paths refer to the specific knowledge points involved in each decision path.

[0166] In some embodiments, the system may map multiple decision paths with the current knowledge graph to identify the knowledge points involved in each path. Alternatively, the system may traverse the knowledge graph to extract the knowledge points that appear on each path and record the relevant information of these knowledge points (such as learning order, difficulty, mastery status, etc.).

[0167] S42, determining a change path of the knowledge point based on the knowledge points corresponding to the multiple decision paths and time nodes of the actions corresponding to the multiple decision paths.

[0168] Time nodes refer to different time points in the learning process of students, which can be associated with learning activities or actions.

[0169] The change path of a knowledge point refers to the mastery progress or status changes of a knowledge point at different time nodes.

[0170] In some embodiments, the system can track the changes of each knowledge point at these time nodes based on the action time nodes in each decision path. For example, from "not learned" to "partially mastered" and then to "fully mastered". Afterwards, the system can calculate the change path of the knowledge point in the learning process by analyzing the time nodes of each action in each decision path. For example, when a student first learns a certain knowledge point, the status of the knowledge point is "not exposed", and then at a certain point in time it enters "learned", and then after certain review steps, it gradually becomes "fully mastered".

[0171] S43, taking the change path of the knowledge point as a part of the visualized teaching content.

[0172] In some embodiments, the system can generate a visual chart or dynamic display based on the calculated knowledge point change path, for example, using a timeline or progress bar to show the student's knowledge point mastery status at different time points.

[0173] The display interface may be an interactive interface, through which students can view the changing trajectory of each knowledge point during their learning process.

[0174] In some embodiments, the system can also obtain reward function values ​​at multiple time steps in the reinforcement learning process; draw a learning efficiency curve based on the reward function values ​​at the multiple time steps; and establish and display a correlation relationship between the learning efficiency curve and the corresponding knowledge points in the reinforcement learning process.

[0175] For example, the system can calculate the effect of the student's learning decision within each time step based on the end time of each time step. After each time step, the reward function value is calculated and recorded based on the student's learning progress (such as test scores, knowledge points mastered, etc.). By collecting the reward function values ​​under all time steps, a sequence of time-reward function values ​​is constructed.

[0176] Furthermore, the system can plot the reward function value and the corresponding time step into a learning efficiency curve through visualization techniques (such as a line graph or a curve graph).

[0177] Through the learning efficiency curve, the system can identify time periods with higher or lower learning efficiency within certain time steps by analyzing the fluctuations of the learning efficiency curve. According to the changes in learning efficiency, these changes can be associated with specific knowledge points mastered by students during the learning process. For example, during the time period of improved efficiency, students may have mastered certain difficult knowledge points. Presenting these associations to users in a visual interface through dynamic charts or labels can help students or teachers understand the relationship between learning efficiency and knowledge point mastery.

[0178] Working principle of the present invention: In the present invention, by dynamically updating the knowledge graph and optimizing the learning path, the learning effect of students can be improved in a targeted manner. At the same time, by combining reinforcement learning and knowledge graph, the optimal learning path that best suits the student's current learning situation can be found to help students strengthen their weak knowledge points. In addition, by graphically displaying the optimal learning path, students' learning interest and intuitive understanding can be enhanced. Teachers can also use visual data to accurately grasp students' learning progress and weak links, further optimize teaching content and strategies, and thus improve the overall teaching quality.

[0179] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A teaching material content visualization system based on data analysis, characterized in that: The system comprises: A first acquisition module is used to acquire digital teaching material content; A first data analysis module, used to perform data analysis based on the digital teaching material content to obtain semantic analysis results, knowledge point extraction results and structural processing results of the digital teaching material content; A graph construction module, used to construct an initial knowledge graph based on the semantic analysis results, knowledge point extraction results and structured processing results; The second acquisition module is used to acquire electronic test papers of multiple students; The second data analysis module is used to perform data analysis based on the electronic test paper to obtain the scores of each student in the test questions corresponding to different knowledge points; A graph updating module, used to update the initial knowledge graph based on the score to obtain a current knowledge graph; A first determination module is used to determine the degree of topological change of the current knowledge graph compared to the initial knowledge graph in the topological structure; The second determination module is used to determine the reward function of reinforcement learning based on the degree of topological change: ； Among them, R t is the reward function of the reinforcement learning, α is the learning rate, γ is the discount factor, P t is the student’s learning progress at the current time t, T cha (G t ) is the degree of topological change, G t is the current knowledge graph, E t is the student's error rate; A learning path determination module, used to obtain the reward and punishment values ​​of the reinforcement learning reward function, and determine the optimal learning path through a preset Q value update formula; A visualization module, used to determine and display visualized teaching content based on the current knowledge graph and the optimal learning path; wherein the visualized teaching content at least includes knowledge points corresponding to the optimal learning path displayed in a graphical structure; The preset Q value update formula is determined in the following way: Based on the current knowledge graph, determine the state definition S for the reinforcement learning t =(K t , P t , H t , L t , G t ); where K t represents the set of knowledge points that students have mastered, P t Indicates the student’s current learning progress, H t represents the type and frequency of students’ errors, L t represents the confidence distribution of students on knowledge points, G t is the current knowledge graph; Determine the action definition of the reinforcement learning a t =Q(S t , a 1); among which, a t In state S t Next select action a A Q value of 1; Based on the state definition S t and the action definition a t , determine the preset Q value update formula as: ； Among them, α is the learning rate, γ is the discount factor, is the next state S t+1 The maximum Q value of all possible actions under a given state S t Next, take action a t After that, the total reward expected to be obtained.

2. A teaching material content visualization system based on data analysis according to claim 1, characterized in that: The graph updating module is further used for: Get the student's learning progress P after each time step t t ; Based on the student's learning progress P t and current action a t , for the current knowledge graph G t to update.

3. A teaching material content visualization system based on data analysis according to claim 2, characterized in that: Based on the student's learning progress P t and current action a t , for the current knowledge graph G t Updates include: Based on the student's learning progress P t , determine the new knowledge points mastered by the students; Add the new knowledge point to the current knowledge graph G t ; Based on the current action a t , determine existing knowledge points related to the new knowledge points; Connect the new knowledge point with the existing knowledge point to obtain a new edge, and based on the student's learning progress P t , assign a weight coefficient to the newly added edge; wherein the weight coefficient is used to indicate the difficulty of students mastering the new knowledge point.

4. A teaching material content visualization system based on data analysis according to claim 3, characterized in that: The graph updating module is further used for: Obtaining N-hop neighbor nodes of the node of the newly added edge; wherein N is a positive integer less than a preset value; Obtaining the degree of association between the knowledge point corresponding to the N-hop neighbor node and the knowledge point corresponding to the node of the newly added edge; Based on the association degree and the weight coefficient corresponding to the newly added edge, the weight coefficient of the edge between the N-hop neighbor nodes is updated.

5. The teaching material content visualization system based on data analysis according to claim 1, characterized in that: The determining and displaying of visual teaching content based on the current knowledge graph and the optimal learning path includes: Obtaining multiple decision paths in the reinforcement learning process; Based on the multiple decision paths, acquiring knowledge points corresponding to the multiple decision paths from the current knowledge graph; Determining a change path of the knowledge point based on the knowledge points corresponding to the multiple decision paths and time nodes of the actions corresponding to the multiple decision paths; The change path of the knowledge point is taken as a part of the visual teaching content.

6. The teaching material content visualization system based on data analysis according to claim 1, characterized in that: The visualization module is further used for: Obtaining reward function values ​​at multiple time steps during the reinforcement learning process; Based on the reward function values ​​at the multiple time steps, plotting a learning efficiency curve; Establish and display the correlation between the learning efficiency curve and the corresponding knowledge points in the reinforcement learning process.

7. A teaching material content visualization method based on data analysis, characterized in that: The method comprises: Access to digital textbook content; Performing data analysis based on the digital textbook content to obtain semantic analysis results, knowledge point extraction results, and structural processing results of the digital textbook content; Based on the semantic analysis results, knowledge point extraction results and structured processing results, construct an initial knowledge graph; Get electronic test papers of multiple students; Perform data analysis based on the electronic test paper to obtain the scores of each student in the test questions corresponding to different knowledge points; Based on the score, the initial knowledge graph is updated to obtain a current knowledge graph; Determine the degree of topological change of the current knowledge graph compared to the initial knowledge graph in the topological structure; Based on the degree of topological change, the reward function of reinforcement learning is determined: ； Among them, R t is the reward function of the reinforcement learning, α is the learning rate, γ is the discount factor, P t is the student’s learning progress at the current time t, T cha (G t ) is the degree of topological change, G t is the current knowledge graph, E t is the student's error rate; Obtaining the reward and punishment values ​​of the reinforcement learning reward function, and determining the optimal learning path through a preset Q-value update formula; Based on the current knowledge graph and the optimal learning path, determine and display the visual teaching content; wherein the visual teaching content at least includes the knowledge points corresponding to the optimal learning path displayed in a graphical structure; The preset Q value update formula is determined in the following way: Based on the current knowledge graph, determine the state definition S for the reinforcement learning t =(K t , P t , H t , L t , G t ); where K t represents the set of knowledge points that students have mastered, P t Indicates the student’s current learning progress, H t represents the type and frequency of students’ errors, L t represents the confidence distribution of students on knowledge points, G t is the current knowledge graph; Determine the action definition of the reinforcement learning a t =Q(S t , a 1); among which, a t In state S t Next select action a A Q value of 1; Based on the state definition S t and the action definition a t , determine the preset Q value update formula as: ； Among them, α is the learning rate, γ is the discount factor, is the next state S t+1 The maximum Q value of all possible actions under a given state S t Next, take action a t After that, the total reward expected to be obtained.

Images