Multi-scale course knowledge graph construction method and intelligent course teaching method

Abstract

The present invention discloses a multi-scale course knowledge graph construction method and an intelligent course teaching method, and relates to the field of intelligent teaching technology. The first layer of knowledge graph is established by extracting knowledge point entities and knowledge point entity relationships in the course textbook text, and the second layer of knowledge graph is formed by extracting questions and their corresponding answers in the exercise materials, and combining relevant knowledge points, and further summarizing the node relationships of the second layer of knowledge nodes to form a third layer of knowledge graph, and finally connecting the relevant nodes of each layer of knowledge graph through course objectives to form a multi-scale course knowledge graph, thereby providing learners with personalized and hierarchical knowledge recommendations based on the knowledge graph. An intelligent teaching method and teaching system based on a multi-scale course knowledge graph are also provided, which can generate learning paths according to the mastery of knowledge by different learners, thereby improving learning efficiency and learning effects.

Description

Technical Field

[0001] The present invention relates to the field of intelligent teaching technology, and in particular to a multi-scale course knowledge graph construction method and an intelligent course teaching method. Background Art

[0002] With the rapid development of artificial intelligence technology, especially the emergence of large language models, new opportunities have been brought to the field of education. Traditional teaching models often fail to meet the needs of personalized learning, while knowledge graph technology can improve teaching efficiency and learning effects through automation and intelligence. However, most of the AI-assisted teaching systems that have emerged, such as homework tutoring and knowledge quizzes, are based on preset learning plans and fixed teaching modes, and there is a problem of insufficient flexibility in teaching content. Different learners have different levels of mastery of the same knowledge points. It often takes a lot of time to find learning materials that match their own knowledge system, and there may be a lot of incompatible content, which affects learning efficiency and learning effects. Summary of the invention

[0003] The present invention provides a multi-scale course knowledge graph construction method and an intelligent course teaching method to solve the problem that the existing intelligent teaching system cannot provide appropriate learning materials for different learners, thus affecting the learners' learning efficiency and learning effect.

[0004] The present invention is achieved through the following technical solutions:

[0005] The first aspect of the present invention provides a method for constructing a multi-scale course knowledge graph, comprising:

[0006] Extracting knowledge point entities and knowledge point entity relationships from the course textbook text, and establishing a first-level knowledge graph based on the knowledge point entities and the knowledge point entity relationships;

[0007] Extracting questions and answers corresponding to the questions in the exercise materials, identifying knowledge nodes related to the questions in the first-layer knowledge graph, forming a knowledge graph microstructure with the knowledge nodes, and establishing a second-layer knowledge graph based on the knowledge graph microstructure;

[0008] Clustering the knowledge graph microstructures, generating summary nodes according to similar knowledge graph microstructures obtained through clustering, and establishing a third-layer knowledge graph based on the summary nodes;

[0009] Obtain course objectives, establish connections between the course objectives and the first-layer knowledge graph, the second-layer knowledge graph, and the third-layer knowledge graph to form a multi-scale course knowledge graph.

[0010] The present invention establishes a multi-scale course knowledge graph, which is applied to the intelligent teaching system. It can search for relevant knowledge in the knowledge graph according to the learning objectives. The multi-scale course knowledge graph includes a first-layer knowledge graph with course knowledge points as the main content, a second-layer knowledge graph formed by questions and answers connected to each knowledge point through the knowledge graph microstructure, and a third-layer knowledge graph formed by summarizing the knowledge graph microstructure. Finally, the fourth-layer knowledge graph is established with the course objectives as the guide, which connects the course objectives with each layer of knowledge nodes across layers, that is, the final multi-scale course knowledge graph is formed. The course objectives are set by the teacher or learner to form a knowledge graph with a personalized knowledge system, which has rich knowledge dimensions such as knowledge points and exercises, so as to recommend knowledge related to the learning objectives based on the knowledge graph in the intelligent teaching process, provide personalized learning plan customization for teachers, and provide personalized learning experience for learners, thereby improving learning effects and learning efficiency.

[0011] In some implementations of the first aspect of the present invention, the knowledge point entities include entity objects, abstract concepts and known rules.

[0012] In some implementations of the first aspect of the present invention, forming the knowledge nodes into a knowledge graph microstructure, and establishing a second-layer knowledge graph based on the knowledge graph microstructure, comprises:

[0013] According to the problem-solving process and comments, the knowledge nodes related to the problem are organized in the order of program operation steps to form a knowledge graph microstructure corresponding to the problem.

[0014] In some implementations of the first aspect of the present invention, clustering the knowledge graph microstructures, generating summary nodes according to similar knowledge graph microstructures obtained by clustering, and establishing a third-layer knowledge graph based on the summary nodes, comprises:

[0015] Performing K-means clustering on the knowledge graph microstructure, analyzing the clustering results, and generating relationships between the knowledge graph microstructures;

[0016] Hierarchical clustering method is used to further summarize and generalize the relationships between knowledge graph microstructures, and summary nodes of similar knowledge graph microstructures are generated based on the relationships between knowledge graph microstructures, and the third-level knowledge graph is generated based on the node relationships.

[0017] The second aspect of the present invention provides a multi-scale course knowledge graph construction device, including:

[0018] The first agent is used to construct the first layer of knowledge graph: extract the knowledge point entities and knowledge point entity relationships in the course textbook text, and establish the first layer of knowledge graph according to the knowledge point entities and the knowledge point entity relationships;

[0019] The second agent is used to construct a second-layer knowledge graph: extract questions and answers corresponding to the questions in the exercise materials, identify knowledge nodes related to the questions in the first-layer knowledge graph, form a knowledge graph microstructure with the knowledge nodes, and establish a second-layer knowledge graph based on the knowledge graph microstructure;

[0020] A third intelligent agent is used to construct a third-layer knowledge graph: cluster the knowledge graph microstructure, generate summary nodes according to similar knowledge graph microstructures obtained by clustering, and establish a third-layer knowledge graph based on the summary nodes;

[0021] The fourth intelligent agent is used to construct a multi-scale course knowledge graph: obtain the course objectives, establish a connection relationship between the course objectives and the first-layer knowledge graph, the second-layer knowledge graph and the third-layer knowledge graph, and form a multi-scale course knowledge graph.

[0022] In some implementations of the second aspect of the present invention, the multi-scale course knowledge graph construction device further includes:

[0023] The knowledge graph operation agent is used to generate query statements or modification statements based on natural language, so as to perform knowledge query or knowledge modification in the multi-scale course knowledge graph through query statements or modification statements.

[0024] A third aspect of the present invention provides an intelligent course teaching method, comprising:

[0025] Before learners start to study the course objectives, test their mastery of knowledge points through dialogue guidance;

[0026] According to the mastery status, searching for a corresponding knowledge graph subgraph in a multi-scale course knowledge graph, wherein the knowledge graph subgraph includes knowledge that the learner has mastered and related knowledge that the learner has not mastered, and the multi-scale course knowledge graph is constructed by the multi-scale course knowledge graph construction method described in any implementation mode of the first aspect of the present invention;

[0027] The network structure of the knowledge graph subgraph is linearly expanded to form multiple progressive linear substructures, and a knowledge graph learning path is generated according to the progressive relationship of the linear substructures for learners to learn.

[0028] In some embodiments of the third aspect of the present invention, the method further includes, after the learner has learned a linear substructure, assessing the learner's learning effect through questions and answers corresponding to the knowledge graph subgraph, and updating the learner's knowledge point mastery.

[0029] A fourth aspect of the present invention provides an intelligent course teaching system, comprising:

[0030] A multi-scale course knowledge graph constructed by the multi-scale course knowledge graph construction method described in any embodiment of the first aspect of the present invention;

[0031] A knowledge graph operation agent is used to generate query statements or modification statements according to natural language, so as to perform knowledge query or knowledge modification in a multi-scale course knowledge graph through the query statements or modification statements;

[0032] Pre-course test agent, used to test learners’ mastery of knowledge points through dialogue before they start learning the course objectives;

[0033] The learning path planning agent is used to generate a knowledge graph learning path based on the learner's mastery.

[0034] In some embodiments of the fourth aspect of the present invention, further comprising:

[0035] The course assessment agent is used to assess the learner's learning effect through the questions and answers corresponding to the knowledge graph subgraph after the learner has learned a linear substructure, and update the learner's knowledge point mastery.

[0036] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0037] A multi-scale course knowledge graph with course knowledge points, exercises and answers, and course objectives as the main contents has been established. It can be used in intelligent teaching systems to customize personalized learning content and recommend personalized knowledge, thereby improving learning efficiency and learning outcomes.

[0038] Use large language models to build multiple specialized agents that work together to create a multi-level knowledge graph. When new course materials or exercise materials are obtained, the knowledge graph can be automatically expanded and updated.

[0039] An intelligent teaching method and teaching system based on a multi-scale course knowledge graph has been established. The pre-school test results and the multi-scale knowledge graph are combined to generate a learning path that suits the learners' learning situation. The learning path is updated according to the assessment results, and the learning content is optimized according to the learners' mastery of the learning process. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative work. In the drawings:

[0041] Figure 1 It is a flowchart of a multi-scale course knowledge graph construction method according to an embodiment of the present invention;

[0042] Figure 2 It is a structural block diagram of a multi-scale course knowledge graph construction device according to an embodiment of the present invention;

[0043] Figure 3 It is a structural diagram of an intelligent course teaching system according to an embodiment of the present invention;

[0044] Figure 4 It is a structural schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0045] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0046] It should be noted that the terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to or inherent to other steps or units of the device.

[0047] The terms used in the various embodiments of the application are only used to describe the purpose of specific embodiments and are not intended to limit the various embodiments of the application. As used herein, the singular form is intended to also include the plural form, unless the context clearly indicates otherwise. Unless otherwise limited, all terms used here (including technical terms and scientific terms) have the same meaning as the meanings commonly understood by ordinary technicians in the field of the various embodiments of the application. The terms (such as the terms defined in the dictionary generally used) will be interpreted as having the same meaning as the contextual meaning in the relevant technical field and will not be interpreted as having an idealized meaning or an overly formal meaning, unless clearly defined in the various embodiments of the application.

[0048] Through the analysis of relevant background technologies, the existing AI-assisted teaching systems have the following problems: (1) the knowledge representation is not comprehensive and multi-dimensional enough; (2) there is a lack of personalized learning path planning; (3) there is a lack of dynamic generation of teaching content based on the learning path; (4) it is difficult to achieve adaptive course assessment. In view of this, the embodiments of the present invention provide a multi-scale course knowledge graph construction method, as well as an intelligent course teaching method and teaching system using a multi-scale course knowledge graph, which are suitable for auxiliary teaching and learning in any learning field, and provide teachers and learners with personalized learning content formulation, personalized knowledge recommendation, and personalized assessment based on learning progress, which is conducive to improving learning efficiency and learning effect.

[0049] An embodiment of the present invention provides a multi-scale course knowledge graph with rich knowledge dimensions, which is applied to an intelligent teaching system, supports personalized learning path generation based on the multi-scale course knowledge graph, provides personalized learning plan customization for teachers, and provides personalized learning experience for learners.

[0050] See also Figure 1 As shown, Figure 1 This is a flowchart of a method for constructing a multi-scale course knowledge graph according to an embodiment of the present invention. The multi-scale course knowledge graph includes a four-level knowledge structure to provide a comprehensive knowledge reasoning foundation and knowledge dimension. The construction method thereof is shown in the following steps S1-S4.

[0051] S1, establish the first layer of knowledge graph: extract knowledge point entities and knowledge point entity relationships in the course textbook text, and establish the first layer of knowledge graph based on the knowledge point entities and knowledge point entity relationships.

[0052] First, we collect various types of course materials related to the training objectives, such as course materials for various stages of compulsory education, quality education materials, adult vocational education materials, etc. We collect the types of course materials related to the training objectives for analysis and establish the first-level knowledge graph, that is, the first-level knowledge graph based on the textbooks. The establishment method is similar to the general knowledge graph. First, we process the course materials into text data, remove information such as pictures, and then extract entities and entity relationships in the course materials text. Here, we extract relevant entities of declarative knowledge points such as relevant entity objects, abstract concepts, and known laws of the course, and form knowledge point entity relationships based on the connections between the knowledge points.

[0053] The entity extraction process can be implemented through an existing entity extraction model or a large language model or agent that can perform entity extraction. For example, a large language model is used to build an agent suitable for identifying and creating the first-level knowledge graph. The agent analyzes the course materials section by section, identifies the knowledge point entities and the relationships between them, and forms the first-level knowledge graph. By building a knowledge graph to generate an agent, new course material texts can be input at any time to expand and update the knowledge graph.

[0054] The specific algorithm for constructing the first layer of knowledge graph is provided below.

[0055] S1-1, pre-process the course materials into text.

[0056] According to the layout of the textbook, the text is divided into blocks, including titles, chapters, and paragraphs. For content involving formulas and tables, the markdown syntax is used for text encoding. For image content, text encoding is used. The specific text is: the title text of the image, the paragraph in the textbook that quotes this image, and additional text descriptions added manually. Finally, the course textbook is converted into a plain text markdown document. In this process, the use of terminology also needs to be unified.

[0057] S1-2, use a large language model to perform entity recognition on the preprocessed course textbook text to extract entity objects, abstract concepts, and known rules.

[0058] S1-3, use a large language model to extract entity relationships from the course textbook text.

[0059] S1-4. Generate the first-layer knowledge graph containing entities and entity relationships.

[0060] S2, establish the second-layer knowledge graph: extract questions and corresponding answers from the exercise materials related to the course materials, identify the knowledge nodes related to the extracted questions in the first-layer knowledge graph, form the related knowledge nodes into a knowledge graph microstructure, and establish the second-layer knowledge graph based on the knowledge graph microstructure.

[0061] Using a method similar to step S1 for extracting entities and entity relationships, the exercise entities and answer entities are extracted from the exercise materials related to the course materials. A large language model is used to build an intelligent agent suitable for identifying and creating the second-layer knowledge graph. The intelligent agent analyzes the course examples, exercises and other materials one by one to obtain questions and their corresponding answers. A question can correspond to at least one answer, and the answer includes the solution process for the question. And according to the knowledge points involved in the answer content, the nodes where the knowledge points involved in the answer are located are identified in the first-layer knowledge graph, and at least one related knowledge node is formed into a knowledge graph microstructure.

[0062] In some embodiments, at least one relevant knowledge node is formed into a knowledge graph microstructure corresponding to the extracted question and the corresponding answer, specifically including: according to the process and comments of solving the problem, these relevant knowledge nodes are organized in the form of program operation steps to form a knowledge graph microstructure involved in solving the specific problem.

[0063] The process of solving a problem can be abstracted as a directed graph. The starting point of the graph is the known conditions and known rules of the problem. The process is reasoning and calculation based on the rules. The final end point is the conclusion. See the example below.

[0064] Question: A free-falling object starts falling from rest. Find its speed after 2 seconds.

[0065] Answer: Since it is a free fall motion, the acceleration is the constant acceleration due to gravity;

[0066] According to the velocity formula under constant acceleration v=v0+a*t, since the object is stationary, v0=0, a=g, the replacement variables are velocity v=g*t;

[0067] Since the acceleration due to gravity is g = 9.8 m / s 2 , t = 2 seconds, substituting into the formula, we can calculate the speed after 2 seconds v = 9.8*2 = 19.6 m / s 2 .

[0068] Based on the above example problems and solution processes, the steps to construct the knowledge graph microstructure are as follows.

[0069] (1) Knowledge point extraction.

[0070] Based on the questions and answers, we can extract the following: free fall (quantitative meaning: a=g), stillness (quantitative meaning: speed is 0), acceleration due to gravity (quantitative meaning: g=9.8m / s 2 ), speed, falling time, the relationship between constant acceleration and speed (v=v0+a*t), variable substitution, numerical expression calculation and other knowledge points. The reasoning rules between the steps are also knowledge points, such as causal logic such as substituting variables, calculation rules, etc.

[0071] (2) Search the constructed knowledge graph for knowledge nodes that match the knowledge points in (1).

[0072] (3) Construction of process relationships between knowledge points, including microstructure naming, microstructure knowledge point extraction, and knowledge point relationship reasoning.

[0073] A. Naming of microstructures, for example, using "Solving the velocity of a free fall object starting from rest" as the name.

[0074] B. Extraction of knowledge points involved in microstructure: See the knowledge point extraction steps in (1).

[0075] C. Construct the microstructure in sequence according to the knowledge reasoning process: [free fall]->[relationship between constant acceleration and velocity](v=v0+a*t)->[variable substitution]->+[acceleration due to gravity]+[stationary]->[v=g*t]->[variable substitution]->[numerical expression calculation]->[velocity v=19.6m / s].

[0076] It should be noted that a problem may have multiple solutions. For different solutions, multiple microstructures can be generated according to the above method.

[0077] A knowledge graph microstructure is taken as a whole to form a node of the second-layer knowledge graph. When searching in the second-layer knowledge graph, the knowledge graph microstructure node corresponding to the question can be queried according to the question, and then the knowledge points involved in the answer process in the first-layer knowledge graph can be obtained according to the knowledge graph microstructure, and presented to learners in the order in which the knowledge points are organized.

[0078] The second-layer knowledge graph construction algorithm includes the following steps.

[0079] S2-1, Knowledge Node Identification: Analyze the text of examples and exercises, identify related entities and concepts, and extract the knowledge points involved in the problem and answer process. The extraction of entity objects and concepts can be found in the construction process of the first layer of knowledge graph.

[0080] S2-2, knowledge graph microstructure generation: match the knowledge points involved in the problem and the answer process with the knowledge nodes of the first-layer knowledge graph, find the relevant knowledge nodes in the first-layer knowledge graph, and organize the relevant knowledge nodes in sequence according to the steps in the answer process to form a knowledge graph microstructure;

[0081] S2-3, construction of the second-layer knowledge graph: the knowledge graph microstructure is taken as a whole to form the nodes of the second-layer knowledge graph, and the second-layer knowledge graph is formed by summarizing the relationship between the knowledge graph microstructure.

[0082] S3, construction of the third-layer knowledge graph: cluster the knowledge graph microstructure, generate summary nodes based on the similar knowledge graph microstructures obtained by clustering, and use the summary nodes as knowledge graph nodes to establish the third-layer knowledge graph.

[0083] The clustering method can adopt K-means clustering to cluster similar nodes, that is, the knowledge graph microstructure in the second-layer knowledge graph, together, conduct comparative analysis on similar nodes, generate the relationship between nodes, and further summarize and conclude to form a summary node with solutions to the course problems involved in this type of similar nodes.

[0084] It is also possible to build relationships for the nodes of the second-level knowledge graph through a large language model, and to summarize and generalize to generate an intelligent agent for the third-level knowledge graph. Through the intelligent agent, node clustering, relationship analysis and generalization are performed to generate summary nodes, and the summary nodes are used as nodes of the third-level knowledge graph to build the third-level knowledge graph.

[0085] The following algorithm for constructing the third-layer knowledge graph is provided.

[0086] S3-1, cluster the knowledge graph microstructure through K-means clustering, compare and analyze the clustering results, and generate the relationship between nodes.

[0087] S3-2, use the hierarchical clustering method to further summarize and generalize the node relationships to form summary nodes with similar knowledge graph microstructures, thereby forming the third-level knowledge graph based on the summary nodes and node relationships.

[0088] S4, establish the fourth-layer knowledge graph: obtain course objectives, establish the relationship between course objectives and the first-layer knowledge graph, the second-layer knowledge graph, and the third-layer knowledge graph to form a multi-scale course knowledge graph.

[0089] You can directly set the course objectives, or have a conversation with the course builder through the active dialogue system to extract information related to the course objectives. Identify the knowledge points and relationships related to the course objectives in the conversation, and connect them with the nodes of each layer of the knowledge graph across layers, thus forming a multi-layer knowledge graph from top to bottom with the course objectives as the highest layer and the course textbook knowledge points as the bottom layer.

[0090] By setting course objectives or inputting course objectives through dialogue, a personalized knowledge graph corresponding to different course objectives can be established, thereby better providing personalized knowledge for different teachers or learners.

[0091] In some embodiments, a knowledge graph operation agent is constructed on the basis of a multi-layer knowledge graph. The agent can generate program statements for querying or modifying a multi-scale course knowledge graph based on a natural language description, and call a database tool to perform knowledge graph query or modification operations, such as Neo4j. The knowledge graph operation agent receives the natural language description, parses and generates a query statement or a modification statement, and submits it to the database tool. The database tool performs knowledge query or knowledge modification operations on the knowledge graph based on the query statement or the modification statement. The query statement or the modification statement adopts a database statement generation method, such as cypher, SQL statement generation, etc.

[0092] Teachers and learners can make requests for knowledge query or knowledge modification by communicating with the knowledge graph operation agent. The knowledge graph operation agent first determines whether it is a knowledge query type or a knowledge modification type based on the received natural language description, and then generates the corresponding query statement or modification statement. This allows both knowledge query to be completed during the teaching process and modification to be made when errors are found in the knowledge during the learning process.

[0093] The embodiment of the present invention also provides a multi-scale course knowledge graph construction device, such as Figure 2 The figure shows a structural block diagram of a multi-scale course knowledge graph construction device, including:

[0094] The first agent is used to build the first layer of knowledge graph;

[0095] The second agent is used to build the second layer of knowledge graph;

[0096] The third agent is used to build the third-layer knowledge graph;

[0097] The fourth agent is used to construct a multi-scale course knowledge graph.

[0098] The specific method for constructing each layer of the knowledge graph is the same as the multi-scale course knowledge graph construction method of the aforementioned embodiment. The first agent, the second agent, the third agent and the fourth agent are constructed based on a large language model, and have the entity extraction, entity relationship summarization and knowledge graph construction functions required for establishing each layer of the knowledge graph, so that each layer of the knowledge graph corresponds to an agent, which is responsible for the construction of the knowledge graph at the corresponding level.

[0099] Specifically, the first agent analyzes the course materials section by section, identifies the knowledge point entities and the relationships between them, and forms the first layer of knowledge graph. The second agent analyzes the examples, exercises and other materials of the course one by one, obtains the questions and their corresponding answers, and identifies the knowledge nodes involved in the questions and their answers in the first layer of knowledge graph, forms the relevant knowledge nodes into a knowledge graph microstructure, and summarizes the relationship between the knowledge graph microstructure to form a second layer of knowledge graph with the knowledge graph microstructure as the node. The third agent clusters the knowledge graph microstructure, forms summary nodes with similar knowledge graph microstructures, and summarizes the relationship between the summary nodes to form a third layer of knowledge graph with the summary nodes as the nodes. The course objectives are obtained through the fourth layer of knowledge graph, and cross-layer connections between the course objectives and the nodes of each layer of knowledge graph are established to form a multi-scale course knowledge graph.

[0100] When the first agent receives the course materials, it automatically updates the first layer of the knowledge graph, and the second, third, and fourth agents update the knowledge graphs they manage accordingly, and finally completes the update of the entire multi-scale course knowledge graph. When the second agent receives the exercise materials, it automatically updates the second layer of the knowledge graph, and the third and fourth agents update the knowledge graphs they manage accordingly. By establishing agents corresponding to each layer of the knowledge graph, the automatic construction and automatic update of the multi-scale course knowledge graph is realized.

[0101] In one embodiment, the multi-scale course knowledge graph construction device also includes a knowledge graph operation agent, which is used to generate query statements or modification statements based on natural language, so as to perform knowledge query or knowledge modification operations in the multi-scale course knowledge graph through query statements or modification statements.

[0102] An embodiment of the present invention also provides an intelligent teaching method, which includes the following steps.

[0103] S4-1, before learners start learning the course objectives, test learners’ mastery of knowledge points through dialogue guidance.

[0104] S4-2, search for the corresponding knowledge graph subgraph in the multi-scale course knowledge graph based on what the learner has mastered.

[0105] S4-3, linearly expand the network structure of the knowledge graph subgraph to form multiple progressive linear substructures, and generate a knowledge graph learning path based on the progressive relationship of the linear substructures for learners to learn.

[0106] Among them, the multi-scale course knowledge graph is constructed by the multi-scale course knowledge graph construction method described in the above embodiment.

[0107] Specifically, the test of learners' mastery of knowledge points through dialogue can be done by guiding learners to give subjective dialogue-style descriptions of the courses they have learned, telling their initial understanding of the courses, and guiding them to express as much as possible through analogies and other methods. Then, based on the students' subjective statements, the agent matches the multi-scale knowledge graph through graph operation agents to find the knowledge graph subgraph that the learner is most familiar with. The knowledge graph subgraph refers to searching for other knowledge points that are directly related to the knowledge point in the multi-scale knowledge graph based on the students' learning situation mastered through dialogue, starting from the knowledge points that have been mastered, and the indirect knowledge points that are related to the direct knowledge points. These directly and indirectly related knowledge points and their connection relationships are extracted separately to form a knowledge graph subgraph corresponding to the learner's knowledge mastery. The knowledge graph subgraph includes both the content that the learner is familiar with and the content that is relatively unfamiliar or not mastered, and these knowledge contents are interconnected, so as to carry out learning in a hierarchical manner and improve learning effects and learning efficiency.

[0108] The agent traverses the knowledge nodes in the knowledge graph subgraph one by one, queries the related questions and answers, arranges and organizes them in the form of a chat dialogue, and starts a dialogue with each question until the student cannot answer, and then terminates the dialogue. Based on the student's answer, the agent rates and scores the student's mastery of each subgraph structure.

[0109] The substructure of the knowledge graph is a network, and the network structure needs to be gradually expanded and traversed linearly to form multiple progressive linear substructures that are gradually exposed to students. In each expanded partial linear substructure, most of the nodes and relationships are already familiar to students, and only a small number are unfamiliar.

[0110] In some implementations, after the learner has learned a linear substructure, the learner's learning effect is assessed through questions and answers corresponding to the knowledge graph subgraph, and the learner's knowledge point mastery is updated.

[0111] Repeat the above steps of sub-graph learning and knowledge point mastery evaluation and update until the entire knowledge graph is covered by linear expansion. Since there are many different ways to linearly expand the network structure, each student's initial graph mastery and graph mastery progress during the learning process are different, so each student's learning path is completely personalized.

[0112] The method for generating test questions is as follows.

[0113] (1) Traverse the knowledge graph substructure and use the information in the knowledge graph (such as the question bank) to generate corresponding test questions and answers.

[0114] (2) For science and engineering questions, generate a Python program to verify the answer.

[0115] In some implementations, the process also includes explaining the learned knowledge points during the learning process.

[0116] When students learn about linear substructures, the knowledge graph explanation agent generates a story-like description based on the linear subgraph structure and explains it to students in an easy-to-understand chat format. During the explanation process, it accepts students' feedback at any time and gives targeted answers to focus the topic on the learning of linear subgraphs.

[0117] The embodiment of the present invention also provides an intelligent course teaching system, such as Figure 3 Shown is the structural block diagram of the intelligent course teaching system, which includes a multi-scale course knowledge graph, a knowledge graph operation agent, a course pre-test agent, and a learning path planning agent.

[0118] The multi-scale course knowledge graph is constructed by the multi-scale course knowledge graph construction method described above;

[0119] The knowledge graph operation agent is used to generate query statements or modification statements according to natural language, so as to perform knowledge query or knowledge modification in the multi-scale course knowledge graph through the query statements or modification statements;

[0120] The pre-course test agent is used to test learners’ mastery of knowledge points through dialogue before they start learning the course objectives.

[0121] The learning path planning agent is used to generate a knowledge graph learning path based on the learner's mastery.

[0122] The pre-course test agent tests the learner's mastery of knowledge points through dialogue. Specifically, the pre-course test agent guides students to give a subjective dialogue description of the course they have learned, talk about their initial understanding of the course, and guide them to express as much as possible through analogies and other methods. Then, based on the students' subjective statements, the agent matches the multi-scale knowledge graph through the graph operation agent to find the knowledge graph subgraph that the students are most familiar with. The agent traverses the knowledge nodes one by one in the knowledge graph subgraph, queries the related questions and answers, arranges and organizes them in the form of a chat dialogue, and conducts a dialogue for each question until the student cannot answer, and then terminates the dialogue. Based on the student's answer, the agent rates and scores the student's mastery of each subgraph structure.

[0123] The substructure of the knowledge graph is a mesh. The learning path planning agent needs to gradually expand and traverse the mesh structure linearly, thereby forming multiple progressive linear substructures that are gradually exposed to students. In each expanded partial linear substructure, most of the nodes and relationships are already familiar to students, and only a small number are unfamiliar. Since there are many ways to expand the mesh structure linearly, each student's initial mastery of the graph and the progress of mastering the graph during the learning process are different, so each student's learning path is completely personalized. Thus, a learning path suitable for the learner is formed according to the linear expansion of the knowledge graph.

[0124] In some embodiments, a course assessment agent is also included, which is used to assess the learner's learning effect through questions and answers corresponding to the knowledge graph subgraph after the learner has learned a linear substructure, and update the learner's knowledge point mastery.

[0125] Build a course assessment agent that traverses the knowledge graph. The agent traverses the substructure of the knowledge graph through the knowledge graph operation agent, and queries or designs corresponding examination questions and answers for each substructure. For science and engineering subjects involving calculations and formula derivation, the agent can generate Python programs to verify them.

[0126] In some embodiments, a knowledge graph explanation agent is also included, which can generate a story-like description based on the linear subgraph structure and explain the learned knowledge points during the learning process. When students learn the linear substructure, the knowledge graph explanation agent generates a story-like description based on the linear subgraph structure and explains it to the students in an easy-to-understand chat format. During the explanation process, the agent can also accept students' questions at any time and give targeted answers, with the topic focusing on the learning of linear subgraphs.

[0127] In some embodiments, a task scheduling agent is further included, and the task scheduling agent performs task allocation according to the task description completed by each agent. The task scheduling agent forms an agent cluster collaboration. Task allocation includes: (1) using the task scheduling agent to allocate different tasks to corresponding agents; (2) dynamically adjusting according to the task complexity and agent load.

[0128] The internal structure of each of the above-mentioned intelligent agents is composed of a perception module, a processing module, and a response module. The perception module is responsible for receiving and processing input data (such as text, dialogue, knowledge subgraph, etc.). The processing module: uses a large language model and algorithm to process data, generate a knowledge graph or respond to operations, and uses a shared graph database to store and manage knowledge graph data. The response module: outputs the generated knowledge graph, query results, or test questions.

[0129] In addition, message queues are used for communication between agents to ensure efficient data transmission and processing. A REST API interface is provided for each agent, and the agent accesses shared data through the API to achieve data sharing and collaboration, and facilitate calls from other agents or system components.

[0130] An embodiment of the present invention provides an electronic device, such as Figure 4 As shown, Figure 4 FIG. 4 is a schematic diagram of the structure of an electronic device of the present invention, wherein the electronic device comprises a processor 40, a memory 41, an input device 42, an output device 43 and a communication device 44; the number of processors 40 in the computer device can be one or more. Figure 4 The processor 40, the memory 41, the input device 42 and the output device 43 in the electronic device can be connected by a bus or other means. Figure 4 The example of connecting through bus is taken in the following.

[0131] The memory 41 is a computer-readable storage medium that can be used to store software programs, computer executable programs, and modules. The processor 40 executes various functional applications and data processing of the electronic device by running the software programs, instructions, and modules stored in the memory 41 to implement the multi-scale course knowledge graph construction method and / or intelligent course teaching method of any of the above embodiments of the present invention.

[0132] The memory 41 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system and at least one application required for a function; the data storage area may store data created according to the use of the terminal, etc. In addition, the memory 41 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 41 may further include a memory remotely arranged relative to the processor 40, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0133] The input device 42 may be used to receive a user ID and instructions, and the output device 43 may be used to output a network configuration page and the like.

[0134] In a fourth aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, a multi-scale course knowledge graph construction method and / or intelligent course teaching method according to any embodiment of the present invention is implemented. The storage medium may be a ROM / RAM, a magnetic disk, an optical disk, etc.

[0135] A fifth aspect of the present invention provides a computer program product, which, when executed on a computer, enables the computer to execute the multi-scale course knowledge graph construction method and / or intelligent course teaching method of any of the above-mentioned embodiments of the present invention.

[0136] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A multi-scale course knowledge graph construction method, characterized by: include: Extracting knowledge point entities and knowledge point entity relationships from the course textbook text, and establishing a first-level knowledge graph based on the knowledge point entities and the knowledge point entity relationships; Extracting questions and corresponding answers from the exercise materials, identifying knowledge nodes related to the questions in the first-layer knowledge graph, organizing the knowledge nodes into a knowledge graph microstructure according to the order of steps in the solution process, and establishing a second-layer knowledge graph based on the knowledge graph microstructure; Clustering the knowledge graph microstructures, generating summary nodes according to similar knowledge graph microstructures obtained through clustering, and establishing a third-layer knowledge graph based on the summary nodes; Obtain course objectives, establish connections between the course objectives and the first-layer knowledge graph, the second-layer knowledge graph, and the third-layer knowledge graph to form a multi-scale course knowledge graph.

2. The multi-scale course knowledge graph construction method according to claim 1 is characterized in that: The knowledge point entities include entity objects, abstract concepts and known rules.

3. The multi-scale course knowledge graph construction method according to claim 1 is characterized in that: The step of clustering the knowledge graph microstructures, generating summary nodes according to similar knowledge graph microstructures obtained through clustering, and establishing a third-layer knowledge graph based on the summary nodes includes: Performing K-means clustering on the knowledge graph microstructure, analyzing the clustering results, and generating relationships between the knowledge graph microstructures; Hierarchical clustering method is used to further summarize and generalize the relationships between knowledge graph microstructures, and summary nodes of similar knowledge graph microstructures are generated based on the relationships between knowledge graph microstructures, and the third-level knowledge graph is generated based on the node relationships.

4. A multi-scale course knowledge graph construction device, characterized in that: include: The first agent is used to construct the first layer of knowledge graph: extract the knowledge point entities and knowledge point entity relationships in the course textbook text, and establish the first layer of knowledge graph according to the knowledge point entities and the knowledge point entity relationships; The second agent is used to construct a second-layer knowledge graph: extract the questions in the exercise materials and the answers corresponding to the questions, identify the knowledge nodes related to the questions in the first-layer knowledge graph, organize the knowledge nodes into a knowledge graph microstructure according to the step sequence in the answering process, and establish a second-layer knowledge graph based on the knowledge graph microstructure; A third intelligent agent is used to construct a third-layer knowledge graph: cluster the knowledge graph microstructure, generate summary nodes according to similar knowledge graph microstructures obtained by clustering, and establish a third-layer knowledge graph based on the summary nodes; The fourth intelligent agent is used to construct a multi-scale course knowledge graph: obtain the course objectives, establish a connection relationship between the course objectives and the first-layer knowledge graph, the second-layer knowledge graph and the third-layer knowledge graph, and form a multi-scale course knowledge graph.

5. The multi-scale course knowledge graph construction device according to claim 4 is characterized in that: Also includes: The knowledge graph operation agent is used to generate query statements or modification statements based on natural language, so as to perform knowledge query or knowledge modification in the multi-scale course knowledge graph through query statements or modification statements.

6. Intelligent course teaching method, characterized by: include: Before learners start to study the course objectives, test their mastery of knowledge points through dialogue guidance; According to the mastery status, searching for a corresponding knowledge graph subgraph in a multi-scale course knowledge graph, wherein the knowledge graph subgraph includes knowledge that the learner has mastered and related knowledge that the learner has not mastered, and the multi-scale course knowledge graph is constructed by the multi-scale course knowledge graph construction method according to any one of claims 1 to 3; The network structure of the knowledge graph subgraph is linearly expanded to form multiple progressive linear substructures, and a knowledge graph learning path is generated according to the progressive relationship of the linear substructures for learners to learn.

7. The intelligent course teaching method according to claim 6, characterized in that: The method also includes, after the learner has learned each linear substructure, assessing the learner's learning effect through questions and answers corresponding to the knowledge graph subgraph, and updating the learner's mastery of knowledge points.

8. Intelligent course teaching system, characterized by: include: A multi-scale course knowledge graph constructed by the multi-scale course knowledge graph construction method described in any one of claims 1 to 3; A knowledge graph operation agent is used to generate query statements or modification statements according to natural language, so as to perform knowledge query or knowledge modification in a multi-scale course knowledge graph through the query statements or modification statements; Pre-course test agent, used to test learners’ mastery of knowledge points through dialogue before they start learning the course objectives; The learning path planning agent is used to generate a knowledge graph learning path based on the learner's mastery.

9. The intelligent course teaching system according to claim 8, characterized in that: Also includes: The course assessment agent is used to assess the learner's learning effect through the questions and answers corresponding to the knowledge graph subgraph after the learner has learned a linear substructure, and update the learner's knowledge mastery.

Images