An AI agent-based learning plan generation method and device

Abstract

The application discloses an AI intelligent agent-based learning plan generation method and device, relates to the technical field of education, and comprises the following steps: acquiring a knowledge point set corresponding to course material information of a student in a target time period from a companion learning system, determining basic learning information of knowledge points in the knowledge point set and prerequisite knowledge information and correlation information between the knowledge points; generating attribute information of the knowledge points based on the basic learning information, the prerequisite knowledge information and the correlation information, and constructing a knowledge graph corresponding to the knowledge point set based on the attribute information; mapping the knowledge graph in a knowledge point learning grid corresponding to the course material information to obtain a target knowledge point learning grid; and generating a learning plan of the student in the target time period based on the target knowledge point learning grid, and displaying the learning plan to the student in the target knowledge point learning grid in the form of a learning path. The application can guide the student to understand knowledge logic and improve learning efficiency and knowledge transfer ability.

Description

Technical Field

[0001] This application relates to the field of educational technology, and in particular to a method and apparatus for generating learning plans based on AI intelligent agents. Background Technology

[0002] With the rapid development of online education, personalized learning systems have become an important tool for improving learning efficiency. However, most existing systems can only generate learning plans based on students' basic information and preset goals. While they clarify "what to learn," they cannot explain "why to learn these things" or "how the content is related." Students often only passively execute planned tasks, struggling to understand the inherent logic between knowledge points and the design intent of the learning path. This results in a lack of systematic understanding of the learning process, hindering efficiency improvement.

[0003] Knowledge graph technology, as a mainstream solution for structuring knowledge, has been introduced into adaptive learning systems, aiming to present the relationships between knowledge points through semantic networks. However, its effectiveness still faces significant challenges in practical applications.

[0004] Existing knowledge graphs largely rely on manual construction by experts or automated data mining. The former is highly subjective, lacks scalability, and struggles to cover all implicit connections; the latter, while capable of discovering some data patterns, lacks dynamic optimization of connection strength and continuous verification of teaching effectiveness. Furthermore, learning plans generated from such graphs often simply transform static knowledge topologies into learning sequences, failing to clearly demonstrate the logical connections between knowledge points and the rationale behind the path design. Students face a "black box" learning plan, unaware of the connection between current learning content and previously learned knowledge, or the logical link between subsequent content and the current task. This results in the learning process remaining at the level of passively completing tasks, unable to achieve deep learning and effective transfer through understanding the knowledge structure. Summary of the Invention

[0005] In view of this, this application provides a learning plan generation method and apparatus based on AI intelligent agents. The main purpose is to improve the technical problem that students face a "black box" learning plan in the current technology. They are not clear about the connection between the current learning content and the knowledge they have already learned, nor do they understand the logical connection between the subsequent content and the current task. As a result, the learning process is still at the level of passively completing tasks and cannot achieve deep learning and effective transfer by understanding the knowledge structure.

[0006] Firstly, this application provides a learning plan generation method based on an AI agent, including:

[0007] The system retrieves the set of knowledge points corresponding to the course materials of students within the target time period from the learning companion system, and determines the basic learning information of the knowledge points in the set of knowledge points, as well as the prerequisite knowledge information and correlation information between the knowledge points.

[0008] Based on the basic learning information, the prerequisite knowledge information, and the relevance information, attribute information of knowledge points is generated, and a knowledge graph corresponding to the set of knowledge points is constructed based on the attribute information.

[0009] The knowledge graph is mapped into the knowledge point learning grid corresponding to the course material information to obtain the target knowledge point learning grid, which is a learning grid that includes the attribute information of knowledge points and the relationship information between knowledge points.

[0010] Based on the learning grid of target knowledge points, a learning plan for the student within the target time period is generated, and the learning plan is displayed to the student in the form of a learning path within the learning grid of target knowledge points.

[0011] Optionally, constructing a knowledge graph corresponding to the set of knowledge points based on the attribute information includes:

[0012] Based on the basic learning information, a first association type between the knowledge points is determined; based on the prerequisite knowledge information, a second association type between the knowledge points is determined; and based on the relevance information, a third association type between the knowledge points is determined.

[0013] Based on the first association type between knowledge points corresponding to the basic learning information and the second association type between knowledge points corresponding to the prerequisite knowledge information, directed edges corresponding to the knowledge point set are generated, and based on the first association type and the third association type between knowledge points corresponding to the relevance information, undirected edges corresponding to the knowledge point set are generated.

[0014] Based on the attribute information, the target association strength corresponding to the knowledge points in the knowledge point set is determined, and an association matrix is ​​generated based on the knowledge point set, the directed edges, the undirected edges, and the association strength.

[0015] The graph edges are constructed based on the association matrix, and the graph nodes are constructed based on the knowledge point set and the attribute information. The graph nodes and graph edges are combined to generate the knowledge graph.

[0016] Optionally, determining the target association strength corresponding to the knowledge points in the knowledge point set based on the attribute information includes:

[0017] Based on the attribute information, determine the frequency data of knowledge points appearing in the same historical learning session, and generate the first association strength corresponding to the knowledge points in the knowledge point set based on the frequency data.

[0018] Based on the attribute information, the mastery level data of knowledge points is determined, and a second association strength corresponding to the knowledge points in the knowledge point set is generated based on the mastery level data;

[0019] Based on the attribute information, the proximity of the learning order of knowledge points is determined, and a third association strength corresponding to the knowledge points in the knowledge point set is generated according to the proximity of the learning order.

[0020] Based on the attribute information, determine the probability data of incorrect answers for knowledge points, and generate the fourth association strength corresponding to the knowledge points in the knowledge point set based on the probability data of incorrect answers.

[0021] Generate the target association strength corresponding to the knowledge points in the knowledge point set based on the first association strength, the second association strength, the third association strength, and the fourth association strength.

[0022] Optionally, after constructing graph edges based on the association matrix, constructing graph nodes based on the knowledge point set and the attribute information, and combining the graph nodes with the graph edges to generate the knowledge graph, the method further includes:

[0023] The knowledge graph is subjected to quality assessment. If there are isolated nodes in the knowledge graph, the knowledge point set is refined. The knowledge graph is then adjusted based on the refined knowledge point set.

[0024] If there are nodes in the knowledge graph whose density is greater than a density threshold, the knowledge point set is merged, and the knowledge graph is adjusted based on the merged knowledge point set.

[0025] Optionally, after generating the student's learning plan for the target time period based on the target knowledge point learning grid, and displaying the learning plan to the student in the form of a learning path within the target knowledge point learning grid, the method further includes:

[0026] In response to obtaining the student's learning behavior data based on the learning plan, the student's question information and note information during the learning process are determined based on the learning behavior data.

[0027] Identify the implicit knowledge points in the question information and the note information, and determine the implicit attribute information of the implicit knowledge points. The implicit attribute information includes implicit prerequisite knowledge information and implicit relevance information.

[0028] When the implicit attribute information meets the knowledge graph update conditions, the knowledge graph is updated based on the implicit knowledge points and the implicit attribute information to obtain a target knowledge graph. The target knowledge graph is used to generate a learning grid that includes the implicit knowledge points and their attribute information.

[0029] Optionally, updating the knowledge graph based on the implicit knowledge points and the implicit attribute information when the implicit attribute information meets the knowledge graph update conditions includes:

[0030] Based on the implicit prerequisite knowledge information, the prerequisite knowledge points corresponding to the implicit knowledge points are determined;

[0031] Based on the question information, determine the student's mastery of the implicit knowledge points and the prerequisite knowledge points, as well as the answer results data of the questions containing the implicit knowledge points and the prerequisite knowledge points;

[0032] The confidence level of the implicit prior knowledge information is determined based on the level of mastery and the answer results data;

[0033] In response to the implicit prior knowledge information having a confidence level greater than a confidence level threshold and the number of times the implicit knowledge point is identified being greater than a number of times threshold, the graph nodes in the knowledge graph are updated based on the implicit knowledge point, and the directed edges in the knowledge graph are updated based on the implicit prior knowledge information.

[0034] Optionally, if the implicit attribute information meets the knowledge graph update conditions, the knowledge graph is updated based on the implicit knowledge points and the implicit attribute information, including:

[0035] Based on the implicit correlation information, the relevant knowledge points corresponding to the implicit knowledge points are determined;

[0036] Based on the notes information, determine the student's mastery of the implicit knowledge points and related knowledge points, as well as the information contribution data of the notes information to the implicit related information;

[0037] The confidence level of the implicit correlation information is determined based on the degree of mastery and the information contribution data.

[0038] In response to the implicit correlation information having a confidence level greater than a confidence level threshold and the number of times the implicit knowledge point is identified being greater than a number of times threshold, the graph nodes in the knowledge graph are updated based on the implicit knowledge points, and the undirected edges in the knowledge graph are updated based on the implicit prior knowledge information.

[0039] Secondly, this application provides a learning plan generation device based on an AI agent, comprising:

[0040] The acquisition module is configured to acquire the set of knowledge points corresponding to the course materials of students within the target time period from the learning companion system, and determine the basic learning information of the knowledge points in the set of knowledge points, as well as the prerequisite knowledge information and correlation information between the knowledge points.

[0041] The generation module is configured to generate attribute information of knowledge points based on the basic learning information, the prerequisite knowledge information and the relevance information, and to construct a knowledge graph corresponding to the set of knowledge points based on the attribute information.

[0042] The mapping module is configured to map the knowledge graph into the knowledge point learning grid corresponding to the course material information to obtain the target knowledge point learning grid, wherein the target knowledge point learning grid is a learning grid that includes the attribute information of knowledge points and the association information between knowledge points.

[0043] The display module is configured to generate a learning plan for the student within the target time period based on the target knowledge point learning grid, and to display the learning plan to the student in the form of a learning path within the target knowledge point learning grid.

[0044] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the AI ​​agent-based learning plan generation method described in the first aspect.

[0045] Fourthly, this application provides an electronic device, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the computer program to implement the AI-based learning plan generation method described in the first aspect.

[0046] By utilizing the above technical solutions, this application provides a learning plan generation method and apparatus based on AI intelligent agents. Compared with existing technologies, this application obtains course knowledge points and related information within the student's target time period from the learning companion system, and constructs a knowledge graph by combining basic learning, prerequisites, and related information. This transforms scattered knowledge points into a structured knowledge network and clearly presents the inherent connections between knowledge points. By mapping the knowledge graph into an intuitive learning grid of target knowledge points, and generating a learning plan based on the grid and displaying it as a learning path, this approach guides students to understand the logic of knowledge, improving learning efficiency and knowledge transfer ability. Attached Figure Description

[0047] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0048] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 A flowchart illustrating a learning plan generation method based on an AI agent provided in an embodiment of this application is shown.

[0050] Figure 2 A flowchart illustrating a learning plan generation method based on an AI agent provided in an embodiment of this application is shown.

[0051] Figure 3 A schematic diagram of an association matrix provided in an embodiment of this application is shown;

[0052] Figure 4 A schematic diagram of a directed edge knowledge graph provided in an embodiment of this application is shown;

[0053] Figure 5 A schematic diagram of an undirected edge knowledge graph provided in an embodiment of this application is shown;

[0054] Figure 6 A schematic diagram of a network centrality analysis provided in an embodiment of this application is shown;

[0055] Figure 7 A schematic diagram of a bidirectional feedback mechanism provided in an embodiment of this application is shown;

[0056] Figure 8 This illustration shows a schematic diagram of a learning plan generation device based on an AI agent, according to an embodiment of this application.

[0057] Figure 9 A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown. Detailed Implementation

[0058] The embodiments of this application will now be described in more detail with reference to the accompanying drawings. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0059] To address the technical problem of students facing a "black box" learning plan in current technologies, where they are unclear about the connection between current learning content and previously learned knowledge, as well as the logical link between subsequent content and the current task, resulting in a learning process that remains at the level of passively completing tasks and failing to achieve deep learning and effective transfer through understanding the knowledge structure, this embodiment provides a learning plan generation method based on an AI agent, such as... Figure 1 As shown, the method includes:

[0060] Step 101: Obtain the set of knowledge points corresponding to the student's course materials within the target time period from the learning companion system, determine the basic learning information of the knowledge points in the knowledge point set, as well as the prerequisite knowledge information and correlation information between the knowledge points.

[0061] In this embodiment of the application, the learning companion system can be the backend system corresponding to the AI ​​learning assistant installed on the learning tablet. The learning companion system can store data such as students' basic information, learning records, and course materials.

[0062] In this embodiment of the application, the target time period can be any learning cycle set by the student, such as a week, a month, or a semester. The specific target time period can be selected by the student according to their learning needs or set by default by the system in conjunction with the teaching progress.

[0063] For the embodiments of this application, course information may include relevant learning materials such as textbook content, teaching syllabus, online course videos, and exercise sets that students need to learn within the corresponding time period.

[0064] In the embodiments of this application, the knowledge point set can be a set of learning units extracted from course material information. Each unit can correspond to a specific knowledge content, and multiple knowledge points together constitute the knowledge point set of the corresponding course material information.

[0065] In the embodiments of this application, basic learning information may be basic attribute information related to the learning of knowledge points. Specifically, basic learning information may include the difficulty, importance, learning time, and chapter to which the knowledge point belongs.

[0066] In this embodiment of the application, the prerequisite knowledge information may be related information about other knowledge points that need to be mastered in advance before learning the knowledge point.

[0067] In the embodiments of this application, the relevance information can be the degree of association between different knowledge points. The relevance information can be used to identify whether there is a connection or complementarity between different knowledge points in terms of content.

[0068] In this embodiment, students need to fill in basic information such as class, grade, name, age, and semester when logging into the learning companion system via a learning tablet. The learning companion system can generate a user ID based on this information. The user ID can be used to associate all the student's learning data. When it is necessary to generate a learning plan for a target time period, the learning companion system can first retrieve the course materials for the corresponding semester from the resource library of the learning companion system based on the student's ID and the selected target time period.

[0069] In the embodiments of this application, the learning companion system can use natural language processing (NLP) technology to perform structured analysis on the retrieved course materials information. Specifically, it can analyze the content of textbook catalogs, teaching outlines, curriculum standards, etc., identify the core concepts, theorems, skill points, etc., and then establish a hierarchical knowledge structure, that is, the breakdown from course to chapter to knowledge point, and finally obtain the set of knowledge points corresponding to the course materials information within the target time period.

[0070] In this embodiment of the application, after determining the set of knowledge points, the system will further extract the basic learning information of each knowledge point. The difficulty of each knowledge point can be quantitatively evaluated by data such as the complexity of the content involved in the knowledge point and the average mastery time of students. The importance of each knowledge point can be determined by combining factors such as the requirements of the teaching syllabus and the proportion of the exam score. The learning time of each knowledge point can be estimated based on the amount of content and difficulty of the knowledge point. For example, each knowledge point requires about 45 minutes of learning time.

[0071] For example, the learning companion system can determine the prerequisite knowledge information between knowledge points by analyzing the content order in the teaching syllabus, the arrangement logic of textbook chapters, and the logical relationship between knowledge points; the learning companion system can determine the relevance information between knowledge points by analyzing the content similarity of knowledge points and their co-occurrence in practice questions.

[0072] Step 102: Generate attribute information for knowledge points based on basic learning information, prerequisite knowledge information, and relevance information, and construct a knowledge graph corresponding to the set of knowledge points based on the attribute information.

[0073] In the embodiments of this application, attribute information can be a comprehensive description of various features of knowledge points. In addition to including basic learning information, prerequisite knowledge information and relevance information, it can also include related information such as association type and association strength derived from these information, which can comprehensively reflect the features of knowledge points and the association between knowledge points.

[0074] In the embodiments of this application, a knowledge graph can be a graph that displays knowledge points and the relationships between knowledge points in a graphical structure. A knowledge graph can include nodes representing knowledge points and edges representing the relationships between knowledge points. A knowledge graph can present the structure and logical relationships of a knowledge system through a visual structure.

[0075] In this embodiment, attribute information of knowledge points is generated based on the acquired basic learning information, prerequisite knowledge information, and relevance information. The difficulty, importance, and learning time in the basic learning information, the specific prerequisite knowledge points corresponding to the prerequisite knowledge information, and the related knowledge points and degree of association corresponding to the relevance information are integrated to form unique attribute information for each knowledge point. A knowledge graph corresponding to the set of knowledge points is constructed based on the generated attribute information.

[0076] Step 103: Map the knowledge graph to the knowledge point learning grid corresponding to the course material information to obtain the target knowledge point learning grid. The target knowledge point learning grid is a learning grid that includes the attribute information of knowledge points and the relationship information between knowledge points.

[0077] In this embodiment of the application, the knowledge point learning grid can be a two-dimensional grid that presents knowledge points. Each cell in the knowledge point learning grid can correspond to a knowledge point, and the knowledge point learning grid can be used to intuitively display the distribution of knowledge points.

[0078] In this embodiment of the application, the target knowledge point learning grid can be a grid formed by mapping the knowledge point nodes and the relationships between nodes in the knowledge graph to the knowledge point learning grid. The target knowledge point learning grid can include the cell corresponding to each knowledge point, the attribute information of the knowledge point, and the relationship information between the knowledge points.

[0079] In the embodiments of this application, the system maps the knowledge graph to the knowledge point learning grid corresponding to the course material information. The basic structure of the knowledge point learning grid can be determined first, and an appropriate number of cells can be divided according to the content volume and number of knowledge points of the course material information, with each cell corresponding to one knowledge point.

[0080] For example, in the process of mapping the knowledge graph into the knowledge point learning grid corresponding to the course material information, in order to maintain the topological relationship between knowledge points in the knowledge graph so that the connected nodes in the graph are as adjacent as possible in the grid, a force-directed algorithm can be used. The force-directed algorithm simulates physical forces (such as repulsive and attractive forces) to determine the position of the nodes. Of course, other similar graph layout algorithms can also be used, including but not limited to.

[0081] In some examples, since the layout generated by the force-directed algorithm is a continuous coordinate system, the continuous coordinates can be discretized onto a grid. Specifically, the two-dimensional space can be divided into a predefined grid, and then each knowledge point node can be assigned to the nearest grid cell.

[0082] For example, in the process of assigning each knowledge point node to the nearest grid cell, a greedy algorithm can also be used to sort the nodes in the force-directed layout according to their importance (such as degree, i.e., the number of associated knowledge points corresponding to the node), and place the nodes in the grid in turn to the nearest free cell to their consecutive positions, until all knowledge point nodes are mapped to the corresponding cells in the grid.

[0083] In some examples, each cell in the target knowledge point learning grid can be labeled with the corresponding knowledge point's attribute information, such as difficulty, importance, and learning time. At the same time, the relationship information between knowledge points can be displayed in the form of lines, such as directed lines representing prerequisite relationships and undirected lines representing correlation relationships. The correlation strength can also be marked on the lines, thus forming a complete learning grid that contains knowledge point attribute information and relationship information.

[0084] Step 104: Generate a learning plan for students within the target time period based on the target knowledge point learning grid, and present the learning plan to students in the form of a learning path in the target knowledge point learning grid.

[0085] In the embodiments of this application, a learning plan can be a detailed learning arrangement formulated for students within a target time period. For example, the learning plan in the embodiments of this application may specifically include the knowledge points to be learned, the learning order, the learning duration, the learning content (such as online courses, practice questions, etc.), and the learning objectives.

[0086] In the embodiments of this application, the learning path can be a visual representation of the learning order of knowledge points in the learning plan. The learning path can be used to display the learning order and path from the starting knowledge point to the target knowledge point in the target knowledge point learning grid through specific markers (such as lines of different colors, arrows, etc.).

[0087] For example, the system can display the learning plan as a learning path in the target knowledge point learning grid. Specifically, the display can use arrows of specific colors to connect the cells corresponding to the knowledge points to be learned, and the direction of the arrows indicates the learning order. At the same time, the system can mark the learning time period and learning content prompts for each knowledge point in the grid, so that students can intuitively see their learning path and specific arrangements within the target time period.

[0088] Compared with existing technologies, this embodiment obtains course knowledge points and related information within the student's target time period from the learning companion system, and constructs a knowledge graph by combining basic learning, prerequisites, and related information. This can transform scattered knowledge points into a structured knowledge network and clearly present the inherent connections between knowledge. By mapping the knowledge graph into an intuitive learning grid of target knowledge points, and generating a learning plan based on the grid and displaying it as a learning path, it can guide students to understand the logic of knowledge and improve learning efficiency and knowledge transfer ability.

[0089] As an optional approach, when performing the "construction of a knowledge graph corresponding to a set of knowledge points based on attribute information", the following methods can be used, but are not limited to: Figure 2 As shown, the method includes:

[0090] Step 201: Determine the first type of association between knowledge points based on basic learning information, determine the second type of association between knowledge points based on prerequisite knowledge information, and determine the third type of association between knowledge points based on relevance information.

[0091] In the embodiments of this application, the first association type can be the association type between knowledge points determined based on the basic learning information of the knowledge points. The first association type can be a relatively basic association relationship that does not involve a strict sequential learning order or strong content association.

[0092] In the embodiments of this application, the second association type can be an association type determined based on prior knowledge information. Specifically, the second association type can be a prior association relationship between knowledge points, and the second association type can reflect the sequential requirements for learning knowledge points.

[0093] In this embodiment of the application, the third association type can be an association type determined based on relevance information, that is, there is a content-related or complementary relationship between knowledge points, but there is no need to strictly follow the order of learning.

[0094] For example, the first association type based on basic learning information can be determined by analyzing the chapter, learning duration range, difficulty level, etc. in the basic learning information of knowledge points. For example, knowledge points belonging to the same chapter can be classified into the same first association type, and knowledge points with the same difficulty level can also be classified into a first association type. The first association type can be a relatively broad association classification, and the first association type can be used to help distinguish the basic association between knowledge points.

[0095] For example, the second type of association based on prior knowledge information can be a prior association relationship, that is, one knowledge point is the prerequisite learning content of another knowledge point, which can reflect the sequential requirement of learning the former first and then the latter.

[0096] For example, the third type of association based on relevance information can be content association, which can indicate that knowledge points are related in terms of derivation ideas and calculation logic but do not need to be learned in sequence.

[0097] Step 202: Generate directed edges corresponding to the knowledge point set based on the first association type between knowledge points corresponding to basic learning information and the second association type between knowledge points corresponding to prerequisite knowledge information, and generate undirected edges corresponding to the knowledge point set based on the first association type and the third association type between knowledge points corresponding to relevance information.

[0098] In this embodiment of the application, a directed edge can be a connecting line in a knowledge graph used to represent the sequential relationship between knowledge points. A directed edge can have a clear direction and can indicate which knowledge point is a prerequisite knowledge point or which is a knowledge point to be learned later.

[0099] In the embodiments of this application, undirected edges can be connecting lines in a knowledge graph used to represent content associations between knowledge points without any order requirement, while directed edges may not have a clear direction and only indicate that there is an association between two knowledge points.

[0100] For example, when generating directed edges, a first association type and a second association type can be combined. The first association type can help filter out the range of knowledge points with potential sequential relationships, such as knowledge points belonging to the same chapter and with increasing difficulty. Then, the prior relationship determined by the second association type is combined to establish directed edges between knowledge points. The starting point of the directed edge can be the node corresponding to the prior knowledge point, and the ending point of the directed edge can be the node corresponding to the subsequent knowledge point.

[0101] For example, when generating undirected edges, a first association type and a third association type can be combined. The first association type can filter out groups of knowledge points with similar basic features, and then the content relevance determined by the third association type can be combined to establish undirected edges between knowledge points. By distinguishing between directed and undirected edges, different types of associations between knowledge points can be clearly presented. Among them, directed edges can have a higher priority than undirected edges. In the application of knowledge graphs, the preconditions represented by directed edges can be given priority.

[0102] Step 203: Determine the target association strength corresponding to the knowledge points in the knowledge point set based on the attribute information, and generate an association matrix based on the knowledge point set, directed edges, undirected edges and association strength.

[0103] In this embodiment of the application, the target association strength can be a numerical value used to quantify the degree of association between knowledge points. The value range of the target association strength can be between [0,1]. The larger the value of the target association strength, the stronger the degree of association between the two knowledge points.

[0104] In this embodiment of the application, the association matrix can be a matrix with knowledge points as rows and columns, and matrix elements as the association strength between corresponding two knowledge points. The association matrix can be used to display the association strength between knowledge points, and the association matrix can also contain association type information corresponding to directed edges and undirected edges.

[0105] In this embodiment, after obtaining the target association strength between all knowledge points, an association matrix can be generated based on the knowledge point set, directed edges, undirected edges, and association strength. The specific generation steps may include: using each knowledge point in the knowledge point set as a row and column of the matrix, with the matrix elements representing the target association strength between the corresponding row and column knowledge points. Simultaneously, the association relationships between directed and undirected edges are represented by specific markings or distinctions; for example, for knowledge points connected by directed edges, the matrix elements can be labeled with direction information, while for knowledge points connected by undirected edges, no direction labeling is required, thus forming a complete association matrix. For example, if the knowledge point set includes k1, k2, and k3, the generated association matrix can be as follows: Figure 3 As shown, 0.8 can be represented as directed edge k1 to k2, and 0.9 can be represented as undirected edge k2 to k3.

[0106] Step 204: Construct graph edges based on the association matrix, and construct graph nodes based on the knowledge point set and attribute information. Combine the graph nodes and graph edges to generate a knowledge graph.

[0107] In this embodiment, the association matrix can be standardized first to ensure that all association strength values ​​are within the range of [0,1]; then, an association strength threshold can be set. For example, you can set =0.3, which allows for the inclusion of associations with a strength greater than a threshold in the association matrix. Graph edges are constructed between the knowledge points corresponding to the elements, and the correlation strength is less than or equal to a threshold. If the graph edges are not constructed, then meaningful relationships are selected.

[0108] For example, the directed edges in the embodiments of this application can maintain direction information, and the directed edge knowledge graph is as follows: Figure 4 As shown, Figure 4 The numbers (0, 1, 2, etc.) in the graph can represent knowledge point identifiers, and each identifier can correspond to one knowledge point. Arrows can indicate the direction of edges; undirected edges do not need to be assigned a direction. An undirected edge knowledge graph is shown below. Figure 5As shown, the corresponding target association strength can be used as the weight of the graph edge and labeled on the graph edge; when constructing graph nodes, each knowledge point corresponds to a graph node, and the attribute information of the knowledge point (including basic learning information, prerequisite knowledge information, relevance information, etc.) can be associated with the corresponding graph node.

[0109] In this embodiment, graph nodes can be connected by graph edges according to the relationships between knowledge points to form a complete knowledge graph. During the construction process, graph algorithms can be used to calculate the network centrality indices of nodes, including degree centrality, betweenness centrality, and proximity centrality. Network centrality analysis can be performed as follows: Figure 6 As shown, degree centrality represents the number of other nodes directly connected to a node, exhibiting local influence; betweenness centrality represents the frequency with which a node is a necessary node on the shortest path between other nodes in the network, exhibiting bridging control; and proximity centrality represents the reciprocal of the average distance from a node to all other nodes in the network, exhibiting global accessibility. For example, knowledge points with high degree centrality can be basic knowledge points, those with high betweenness centrality can be key transit knowledge points, and those with high proximity centrality can be core knowledge points.

[0110] For example, assuming that after standardization and threshold filtering, the association strength between knowledge points k1, k2, and k3 is 0.8 (greater than the threshold of 0.3), the association strength between k2 and k3 is 0.9 (greater than the threshold of 0.3), and the association strength between k1 and k3 is 0.2 (less than the threshold of 0.3), then a directed edge (weight 0.8) is constructed between k1 and k2, and an undirected edge (weight 0.9) is constructed between k2 and k3. The three corresponding knowledge points are used as graph nodes, and each node is associated with its attribute information, ultimately forming a knowledge graph containing these three nodes and two graph edges.

[0111] Optionally, when performing the "determining the target association strength corresponding to the knowledge points in the knowledge point set based on attribute information", the following methods may be used, but are not limited to: determining the frequency data of the knowledge points appearing in the same historical learning sessions based on attribute information, and generating a first association strength corresponding to the knowledge points in the knowledge point set based on the frequency data; determining the mastery level data of the knowledge points based on attribute information, and generating a second association strength corresponding to the knowledge points in the knowledge point set based on the mastery level data; determining the proximity of the learning order of the knowledge points based on attribute information, and generating a third association strength corresponding to the knowledge points in the knowledge point set based on the proximity of the learning order; determining the probability data of the knowledge points being answered incorrectly based on attribute information, and generating a fourth association strength corresponding to the knowledge points in the knowledge point set based on the probability data of being answered incorrectly; and generating a target association strength corresponding to the knowledge points in the knowledge point set based on the first association strength, second association strength, third association strength, and fourth association strength.

[0112] In the embodiments of this application, the target association strength can be determined by combining the attribute information of knowledge points from multiple dimensions and then fusing them together. The target association strength may include a first association strength, a second association strength, a third association strength, and a fourth association strength.

[0113] In the embodiments of this application, the frequency data of knowledge points appearing in the same historical learning session can be determined based on attribute information. The first association strength can be generated based on the frequency data. The higher the frequency, the greater the first association strength. For example, if two knowledge points are frequently learned in the same learning session, their first association strength will be higher.

[0114] In the embodiments of this application, the mastery level data of knowledge points determined based on attribute information can be used to calculate the correlation between the mastery levels of different knowledge points, and a second correlation strength can be generated based on the correlation data; if the mastery levels of two knowledge points are highly correlated, that is, when one knowledge point is mastered well, the other knowledge point is usually also mastered well, then the second correlation strength will be greater.

[0115] In this embodiment of the application, the proximity of the learning order of knowledge points is determined based on attribute information. For example, the shorter the interval between the learning time of two knowledge points, the closer the learning order, and the stronger the corresponding third association.

[0116] In this application embodiment, the probability data of knowledge points being answered incorrectly is determined based on attribute information. For example, if two knowledge points are frequently answered incorrectly by students at the same time, it can be proven that the two knowledge points are related in terms of learning difficulty or content association. Then, a fourth association strength can be generated based on the co-occurrence of error probabilities.

[0117] In this embodiment of the application, after obtaining the first association strength, second association strength, third association strength, and fourth association strength, the target association strength between knowledge points can be obtained by weighting and fusing the first association strength, second association strength, third association strength, and fourth association strength by setting weight coefficients. The weight coefficients can be adjusted according to the actual learning scenario and teaching needs. The calculation formula for the target association strength is as shown in Formula 1, where, This can be expressed as the weighting coefficient of the first association strength. This can be expressed as the weighting coefficient of the second association strength. This can be expressed as the weighting coefficient of the third association strength. It can be expressed as the weighting coefficient of the fourth association strength, and the weighting coefficient must satisfy... .

[0118] Target correlation strength =

[0119] ×First association strength+ ×Second association strength+ ×Third association strength+ ×Fourth association strength

[0120] (Formula 1)

[0121] Optionally, the target association strength can be further optimized by combining the teaching expert rating and the association strength based on text content similarity. For example, the target association strength = 0.4 × expert rating + 0.3 × student data association strength (i.e., the fusion result of the above four association strengths) + 0.3 × content similarity association strength. The content similarity can be obtained by calculating the cosine similarity after vectorizing the text description based on knowledge points using TF-IDF.

[0122] As an optional approach, after performing the steps of "constructing graph edges based on the association matrix, constructing graph nodes based on the knowledge point set and attribute information, and combining graph nodes with graph edges to generate a knowledge graph," the following methods can also be used, but are not limited to: performing a quality assessment of the knowledge graph; refining the knowledge point set when isolated nodes exist in the knowledge graph, and adjusting the knowledge graph based on the refined knowledge point set; and merging the knowledge point set when the node density in the knowledge graph exceeds a density threshold, and adjusting the knowledge graph based on the merged knowledge point set.

[0123] In the quality assessment process of this application embodiment, the existence of isolated nodes, i.e., nodes with a degree centrality of 0, can be checked first. These nodes have no connection with other nodes, which may be due to unreasonable knowledge point splitting or omission of relationship identification. When isolated nodes are found, the knowledge point set can be refined, and the content of the knowledge points corresponding to the isolated nodes can be re-analyzed to check whether there are problems with overly coarse splitting or unclear content boundaries. If the content of the knowledge points corresponding to the isolated nodes has problems with overly coarse splitting or unclear content boundaries, the content of the knowledge points corresponding to the isolated nodes can be refined into multiple sub-knowledge points or the content boundaries of the knowledge points corresponding to the isolated nodes can be adjusted so that the content of the knowledge points corresponding to the isolated nodes can establish reasonable relationships with other knowledge points. Then, the relationships are reconstructed based on the refined knowledge point set, and the knowledge graph is adjusted.

[0124] In this embodiment, the node density in the knowledge graph can be specifically measured by the average degree, i.e., the average number of connections of all nodes. The specific steps for checking the node density in the knowledge graph may include: first, setting a density threshold. When the node density is greater than the density threshold, it indicates that the associations between knowledge points are too dense, which may lead to a chaotic knowledge graph structure and affect the usability. At this time, it is necessary to merge the knowledge point set, merging some knowledge points that are too closely related and have extremely high content similarity, reducing the number of nodes and simplifying the association relationships; then, based on the merged knowledge point set, the association relationship matrix and knowledge graph are reconstructed to adjust the knowledge graph.

[0125] In this embodiment, a two-way feedback mechanism can also be established to form a closed-loop process of knowledge point extraction, relationship modeling, network analysis, and quality assessment. The two-way feedback mechanism can be as follows: Figure 7 As shown; if the quality assessment fails, the process returns to the corresponding step to adjust the granularity of the knowledge points until the knowledge graph meets the requirements. It should be noted that the quality assessment and adjustment mechanism continuously optimizes the structure of the knowledge graph, ensuring that it accurately reflects the relationships between knowledge points and provides reliable support for the generation of subsequent learning plans. For example, if the knowledge graph construction for course A reveals that the nodes corresponding to knowledge point A have high betweenness centrality and some relationships are disordered, knowledge point A can be refined into two knowledge points and the relationships can be reconstructed to make the knowledge graph structure more reasonable.

[0126] As an optional approach, after performing the step of "generating a student's learning plan within a target time period based on the learning grid of target knowledge points, and displaying the learning plan to the student in the form of a learning path within the learning grid of target knowledge points," the following method can also be used, but is not limited to: In response to obtaining learning behavior data of students learning based on the learning plan, determining the student's question information and note information during the learning process based on the learning behavior data; identifying implicit knowledge points in the question information and note information, determining the implicit attribute information of the implicit knowledge points, including implicit prerequisite knowledge information and implicit relevance information; and updating the knowledge graph based on the implicit knowledge points and implicit attribute information when the implicit attribute information meets the knowledge graph update conditions, obtaining a target knowledge graph, which is used to generate a learning grid including implicit knowledge points and their attribute information.

[0127] In this embodiment of the application, learning behavior data may include students' online course learning records, practice questions records, notes, and question records. The system can extract students' question information and note information from this learning behavior data. The question information may include the question ID, answer, accuracy rate, answer time, and a list of explicit knowledge points marked by the system. The note information may include text content recorded by the student during the learning process, the learning location, and the corresponding contextual knowledge points. For example, by setting the data unit as a single learning session (such as completing a question or reading a chapter), a triple D=(Student, Action, Context) can be recorded to capture key data from the student's learning process.

[0128] In this embodiment of the application, implicit knowledge points can be knowledge points that are not directly marked by the system in the questions or notes but actually exist and have an important impact on the learning of the current knowledge point; the system can continuously monitor the student's learning behavior, and when the student generates a question or note event, it can jump to the implicit knowledge point identification step.

[0129] In this embodiment of the application, implicit attribute information may include implicit prerequisite knowledge information and implicit relevance information. Implicit prerequisite knowledge information may be knowledge points that need to be mastered in advance to learn the implicit knowledge point, and implicit relevance information may be knowledge points that are related to the implicit knowledge point in terms of content.

[0130] In this embodiment of the application, when the implicit attribute information meets the knowledge graph update conditions, the knowledge graph can be updated based on the implicit knowledge points and implicit attribute information to obtain the target knowledge graph. The target knowledge graph can be used to generate a learning grid containing implicit knowledge points and attribute information of implicit knowledge points.

[0131] As an optional approach, when performing the task of "updating the knowledge graph based on implicit knowledge points and implicit attribute information when the implicit attribute information meets the knowledge graph update conditions," the following methods can also be used, but are not limited to: determining the prerequisite knowledge points corresponding to the implicit knowledge points based on the implicit prerequisite knowledge information; determining the student's mastery of the implicit knowledge points and prerequisite knowledge points, as well as the answer result data of the questions containing the implicit knowledge points and prerequisite knowledge points, based on the question information; determining the confidence level of the implicit prerequisite knowledge information based on the mastery level and the answer result data; and updating the graph nodes in the knowledge graph based on the implicit knowledge points and updating the directed edges in the knowledge graph based on the implicit prerequisite knowledge information when the confidence level of the implicit prerequisite knowledge information is greater than the confidence level threshold and the number of times the implicit knowledge points are identified is greater than the number of times the implicit knowledge points are identified.

[0132] In the embodiments of this application, the level of mastery can be calculated using data such as the student's correct answer rate and review frequency.

[0133] For the embodiments of this application, the answer result data may include whether the answer is correct or not, the answer time, etc.

[0134] In this embodiment of the application, the confidence level of implicit prerequisite knowledge information can be determined based on the degree of mastery and the answer result data. For example, when a student answers a question containing implicit knowledge points and corresponding prerequisite knowledge points incorrectly and has not mastered the prerequisite knowledge points, the confidence level will increase significantly.

[0135] For example, to determine the confidence level of implicit prerequisite knowledge information based on the student's mastery and answer results data, one can first perform NLP analysis (such as entity linking) on ​​the question answer analysis to extract all knowledge point sets and calculate the candidate set of implicit knowledge points; then, perform mastery verification on each knowledge point and define a confidence threshold (e.g., 0.7); if the student answers incorrectly and does not master the implicit prerequisite knowledge points, a strong evidence strength can be set; if the student answers correctly but does not master the implicit prerequisite knowledge points, a weak evidence strength can be set; if the student answers correctly and masters the implicit prerequisite knowledge points, a moderate evidence strength can be set; otherwise, there is no evidence.

[0136] In this embodiment of the application, the accumulation of confidence using the confidence update rule may specifically include: when the confidence of implicit prior knowledge information is greater than the confidence threshold and the number of times implicit knowledge points are identified is greater than the number of times threshold, the graph nodes in the knowledge graph can be updated based on the implicit knowledge points, that is, the implicit knowledge points are added as new graph nodes. Furthermore, the directed edges in the knowledge graph can be updated based on the implicit prior knowledge information, that is, the directed edges pointing from the prior knowledge points to the implicit knowledge points are added.

[0137] As an optional approach, when performing the task of "updating the knowledge graph based on implicit knowledge points and implicit attribute information when the implicit attribute information meets the knowledge graph update conditions," the following methods can also be used, but are not limited to: determining the related knowledge points corresponding to the implicit knowledge points based on implicit relevance information; determining the student's mastery of the implicit knowledge points and related knowledge points, as well as the information contribution data of the note information to the implicit relevance information, based on the note information; determining the confidence level of the implicit relevance information based on the mastery level and information contribution data; updating the graph nodes in the knowledge graph based on the implicit knowledge points in response to the confidence level of the implicit relevance information being greater than the confidence level threshold and the number of times the implicit knowledge points are identified being greater than the number of times the threshold is reached; and updating the undirected edges in the knowledge graph based on the implicit prerequisite knowledge information.

[0138] In this embodiment, the information contribution data can be determined based on the frequency of mentioning two knowledge points simultaneously in the notes, the level of detail in the description, etc. For example, if the context of the notes contains one of the knowledge points, the basic contribution can be set to 1; otherwise, it can be set to 0.5. The information contribution is then calculated in combination with the student's level of mastery.

[0139] In the embodiments of this application, during the process of updating the graph nodes in the knowledge graph based on implicit knowledge points and updating the undirected edges in the knowledge graph based on implicit prerequisite knowledge information, a candidate relation library can also be established to store the implicit attribute information of all candidates. The candidate relation library can be verified by periodically scanning the candidate relation library. For candidate relations whose confidence and occurrence frequency reach the threshold, they can be added to the knowledge graph. For candidate relations that do not reach the threshold or have not been triggered for a long time (such as more than 3 months), the implicit attribute information of the candidates can be removed from the candidate relation library.

[0140] For example, if the implicit attribute information k1→k2 in the candidate relation database reaches a confidence level of 0.75 and an occurrence count of 6 after multiple evidence accumulations, and both the confidence level and the occurrence count exceed the corresponding thresholds, then the implicit attribute information of the word can be updated in the knowledge graph.

[0141] As an optional approach, when generating a learning plan based on the target knowledge point learning grid, the system can first locate the grid coordinates of the knowledge point that the student is currently learning. The grid position can be represented by Pvi=(xi;yi) to indicate the coordinates of knowledge point vi in ​​the grid.

[0142] Optionally, learning status data may include online course learning progress (such as video completion rate, number of repeated views, etc.), practice questions (such as accuracy rate, number of questions answered, time spent answering questions, etc.), and interaction data (such as frequency of asking questions, number of notes, etc.). The system can use this data to calculate the mastery level of each knowledge point. Specifically, the calculation can be performed by constructing a feature vector, incorporating online course learning data (video completion rate, etc.) into the data. Repeat viewing count Duration of stay ), practice data (accuracy rate) Answering speed Correct answer rate for redoing incorrect questions Interaction data (frequency of questions) Number of notes Features such as ) are included, and the feature vector is shown in Formula 2, where, This can be expressed as video completion rate, This can be expressed as the number of times it was viewed repeatedly. It can be expressed as the length of stay, It can represent the accuracy rate of solving problems, It can indicate the speed of answering questions, It can represent the accuracy rate of redoing incorrect questions. It can indicate the frequency of questions asked. It can represent the number of notes; and perform normalization processing, the expression for which is shown in Formula 3, normalizing each feature component to the range of [0,1].

[0143] (Formula 2)

[0144] (Formula 3)

[0145] In this embodiment of the application, the effective value of the learning state data can be calculated according to time decay, as shown in Formula 4, where It can represent whether the i-th answer is correct or incorrect (1 or 0). It can represent the time taken to answer the question for the i-th time. It can indicate the current time for answering the question. It can represent the attenuation coefficient.

[0146] (Formula 4)

[0147] Optionally, embodiments of this application can use logistic regression or softmax regression models to output the mastery probability, and then set a threshold for three-color classification. First, the mastery score can be calculated, as shown in Formula 5, where, It can represent a weight vector, which can represent the importance of each feature to the mastery level (obtained through training), and b is the bias term.

[0148] (Formula 5)

[0149] For example, the probability of a student mastering a knowledge point can be calculated using the sigmoid function. Mapping to probability , The expression is shown in Formula 6; a threshold can be set. and (For example, setting a threshold) =0.4 and =0.7), if the probability is known Less than 0.4, marked in red (not yet understood); if Values ​​greater than or equal to 0.4 and less than 0.7 are marked in yellow (indicating mastery but not proficiency); if... A score greater than or equal to 0.7 is marked in green (Mastery). The system can prioritize knowledge points marked in red and yellow as key learning content based on the level of mastery. Combining the prerequisite and relevance relationships between knowledge points, the attribute information of knowledge points (such as difficulty, learning time, and importance), and the duration of the target time period, the system can reasonably arrange the learning order and learning time, and generate a learning plan for the target time period.

[0150] (Formula 6)

[0151] Optionally, this application also provides an example, such as... Figure 8 As shown, in the process of knowledge graph construction, since the extraction of knowledge points is processed using natural language processing technology, knowledge points with explicit features can be extracted. However, in the actual application scenario of students doing a test question, if the question tests knowledge point A (Ab), then although the question tests knowledge point A, there is actually a hidden knowledge point C. Knowledge point C must be mastered before knowledge point A can be mastered. Some answer explanations will mention the implicit knowledge point C. Then, based on the student's notes on the question, it can be found that in addition to recording knowledge point A, the student also recorded notes on knowledge points C and D. Therefore, the relationship between implicit knowledge points is also particularly important. Determining the relationship between implicit knowledge points can specifically include: obtaining the student's test-taking records, notes, and answer explanations on the student's tablet (which can be grouped by class, grade, or academic year). This can be used to treat the initial knowledge graph as a prior model, using the student's test-taking and note-taking behavioral data as observational evidence. Through a series of calculations and analyses, the relationships in the graph can be updated, especially to discover and verify those implicit, undefined dependencies.

[0152] Optionally, data units can be set to record triples D=(Student, Action, Context) based on a single learning session (such as completing a question or reading a chapter); question-solving data... The expression can be shown in Formula 7, where q represents the question ID, ans represents the student's answer, correct belongs to {0,1} to indicate whether it is correct or not, and t represents the time. It can represent a list of explicit knowledge points annotated by the system.

[0153] (Formula 7)

[0154] Optional, note data The expression can be shown in Formula 8, where Text can represent the note text and Loc can represent the learning location. It can represent contextual knowledge points; the system can continuously monitor students' learning behavior, and when it generates... or In the event of an incident, you can skip to the hidden knowledge point identification step.

[0155] (Formula 8)

[0156] Optionally, the hidden knowledge point identification steps may specifically include: extracting hidden prerequisite relationships from the question-answering data (regardless of whether the answer is right or wrong, as long as the question involves hidden knowledge points and the student's mastery is related to the answer result, the discovery of the association and the update of the confidence level can be triggered).

[0157] Step 1: Triggering conditions can include any record of answering questions. This includes the question q, the student's answer ans, whether it is correct or not, and the explicit knowledge points marked by the system. ;

[0158] Step 2, hidden knowledge point discovery, may include: performing NLP analysis (such as entity linking) on ​​the answer parsing of question q, and extracting the set of all knowledge points. ; Calculate the candidate set of hidden knowledge points, using the formula shown in Formula 9, where, It can represent hidden knowledge points. It can represent a set of knowledge points (i.e., knowledge points mentioned in the answer analysis but not directly marked in the question).

[0159] (Formula Nine)

[0160] Step 3: Student Mastery Verification: If the probability of a student mastering each knowledge point is P(k), a mastery threshold can be defined. When P(k) is less than In such cases, it can be assumed that the student has not mastered k.

[0161] Step 4: Establish candidate relationships and calculate the strength of evidence. and In the case where Pce=0 in the initial graph (i.e., there is no prior relation), a candidate prior relation kc→ke can be proposed (if it does not already exist); the strength of the evidence can be calculated based on the students' answers and level of understanding. .

[0162] For example, if a student answers incorrectly (correct=0), and the system detects that the student has not mastered kc (i.e., P(kc) is less than 0), This is strong evidence that kc might be a priori of ke, and it can be set. .

[0163] For example, if a student answers correctly (correct=1), but the system detects that the student has not mastered kc (i.e., P(kc) is less than 1), the system will not detect the student's mastery of kc. If so, this is weak evidence because students might answer correctly through guessing or other means. It is possible to set... .

[0164] For example, if the student answers correctly and has mastered kc (i.e., P(kc) is greater than or equal to...), If this supports a connection, but the strength of the evidence is moderate, then it can be set to... In other cases, Snew=0 (i.e., no evidence).

[0165] Compared with existing technologies, this embodiment improves the accuracy and interpretability of the knowledge network by clarifying the association type, distinguishing between directed and undirected edges, and generating an association matrix. By calculating the target association strength from multiple dimensions such as co-occurrence frequency and mastery, the quantification of the degree of association of knowledge points is more in line with teaching practice and students' learning patterns. Through quality assessment, isolated nodes are refined and high-density nodes are merged to continuously optimize the knowledge graph structure, ensuring the long-term effectiveness of the graph. By mining implicit knowledge points in learning behavior data and verifying and updating the knowledge graph with confidence and recognition count, the accuracy of the learning plan is improved, further helping students understand the deep logic between knowledge points and improving the systematicness and efficiency of learning.

[0166] As an alternative approach, this application also provides the following embodiments, but is not limited thereto, including:

[0167] Example 1: In response to obtaining first learning behavior data of a student learning questions based on the learning plan, the system determines first question information and first note information of the student's learning target questions based on the first learning behavior data, and sends the first question information and first note information to the knowledge point learning grid module; the knowledge point learning grid module is used to determine the target implicit knowledge point set corresponding to the learning plan from the implicit knowledge point set based on the implicit knowledge point set corresponding to the first question information and the first note information, according to the prerequisite relationship and correlation information between the implicit knowledge point set and the first knowledge point set; the target implicit knowledge point set is updated in the target knowledge point learning grid; the learning path generation module is used to update the learning plan to include the target implicit knowledge point set.

[0168] Example 11: Extract knowledge points from the first question information and the first notes information respectively to obtain the second set of knowledge points corresponding to the first question information and the third set of knowledge points corresponding to the first notes information; determine the knowledge points in the second set that are not included in the first set of knowledge points as the first subset of implicit knowledge points corresponding to the target question; determine the knowledge points in the third set that are not included in the first set of knowledge points as the second subset of implicit knowledge points corresponding to the target question; determine the implicit knowledge point set based on the first subset of implicit knowledge points and the second subset of implicit knowledge points.

[0169] Example 12: Establish multiple prerequisite relationships for knowledge points based on a first subset of implicit knowledge points and a first set of knowledge points, so as to determine the knowledge points in the first subset of implicit knowledge points as prerequisite knowledge points for the knowledge points in the first set of knowledge points; establish multiple correlation relationships for knowledge points based on a second subset of implicit knowledge points and a first set of knowledge points, so as to associate the knowledge points in the second subset of implicit knowledge points with the knowledge points in the first set of knowledge points; determine the first confidence level of using the multiple prerequisite relationships for knowledge points as prerequisite relationships for target knowledge points corresponding to the learning plan and the second confidence level of using the multiple correlation relationships for knowledge points as correlation relationships for target knowledge points corresponding to the learning plan; determine the target implicit knowledge point set corresponding to the learning plan from the implicit knowledge point set based on the first confidence level and the second confidence level.

[0170] Example 13: Determine the student's first mastery data on the first subset of implicit knowledge points, and determine the student's answer to the question based on the student's answer data in the first question information; determine the first support data for each knowledge point prerequisite relationship in multiple knowledge point prerequisite relationships based on the first mastery data and the answer results; update the first confidence of each knowledge point prerequisite relationship based on the historical confidence of the first support data to obtain the first confidence of using multiple knowledge point prerequisite relationships as the target knowledge point prerequisite relationship.

[0171] Example 14: Determine the second mastery data of students on the second subset of implicit knowledge points, and the contribution data of the first notes information to the correlation relationship of multiple knowledge points; determine the second support data of each knowledge point correlation relationship in the multiple knowledge point correlation relationships based on the mastery data and the contribution data; update the second support data according to the historical confidence of each knowledge point correlation relationship, and obtain the current second confidence of taking the correlation relationship of multiple knowledge points as the correlation relationship of the target knowledge point.

[0172] Example 15: Multiple prerequisite relationships and multiple correlation relationships of knowledge points are stored in a candidate relation library corresponding to the learning plan, so as to monitor the first confidence level and the second confidence level through the candidate relation library; in response to the first confidence level of the prerequisite relationship of the first knowledge point being greater than the prerequisite confidence threshold and the number of times the first implicit knowledge point corresponding to the prerequisite relationship of the first knowledge point being identified being greater than the number of times threshold, the first implicit knowledge point is determined as the first target implicit knowledge point corresponding to the learning plan; in response to the first confidence level of the correlation relationship of the first knowledge point being greater than the correlation confidence threshold and the number of times the first implicit knowledge point corresponding to the correlation relationship of the first knowledge point being identified being greater than the number of times threshold, the first implicit knowledge point is determined as the first target implicit knowledge point corresponding to the learning plan; the target implicit knowledge point set corresponding to the learning plan is determined based on the first target implicit knowledge point and the second target implicit knowledge point.

[0173] Example 16: Determine the monitoring period for the first confidence level and the second confidence level of the candidate relation library; within the monitoring period, in response to the first confidence level of the second knowledge point prerequisite relation among multiple knowledge point prerequisite relations and / or the second confidence level of the second knowledge point prerequisite relation among multiple knowledge point correlation relations being lower than the monitoring threshold, remove the second knowledge point prerequisite relation and / or the second knowledge point prerequisite relation from the candidate relation library; in response to the end of the monitoring period, remove multiple knowledge point prerequisite relations and multiple knowledge point correlation relations from the candidate relation library.

[0174] As an alternative approach, this application also provides the following embodiments, but is not limited thereto, including:

[0175] Example 2: Based on the student's learning location information in the target knowledge point learning grid, determine the target learning course corresponding to the learning location information; divide the student's second learning behavior data in the target learning course into at least one learning behavior data set, and generate at least one behavioral feature set corresponding to the at least one learning behavior data set; analyze the student's mastery of knowledge points in the target learning course based on the at least one behavioral feature set, and determine the student's unmastered knowledge point set in the target learning course; match the unmastered knowledge point set with the target knowledge graph corresponding to the target knowledge point learning grid to obtain the target knowledge point network corresponding to the unmastered knowledge point set in the target knowledge graph; generate the student's target learning path in the target knowledge point learning grid based on the target knowledge point network, and the target learning path is used to assist the student in mastering the unmastered knowledge point set.

[0176] Example 21: Determine the time information corresponding to each behavioral feature in at least one set of behavioral features, and evaluate the effectiveness of each behavioral feature in at least one set of behavioral features based on the time information to obtain effective value data for each behavioral feature in at least one set of behavioral features; filter the behavioral features in at least one set of behavioral features based on the effective value data to obtain target behavioral features that meet the effective conditions in at least one set of behavioral features, and form at least one set of target behavioral features; analyze the students' mastery of knowledge points in the target learning course based on at least one set of target behavioral features to determine the set of knowledge points that students have not mastered in the target learning course.

[0177] Example 22: Match at least one set of target behavioral features with each knowledge point in the target learning course to determine at least one target behavioral feature corresponding to each knowledge point in the target learning course; identify the importance data of at least one target behavioral feature corresponding to each knowledge point in the target learning course to each knowledge point in the target learning course through a target model, which is trained based on students' historical behavioral features and historical knowledge point mastery levels; determine students' mastery data of each knowledge point in the target learning course based on at least one target behavioral feature and importance data corresponding to each knowledge point; determine the set of unmastered knowledge points in the target learning course based on the mastery data.

[0178] Example 23: Based on importance data, determine the influence coefficient of at least one behavioral feature corresponding to each knowledge point on the degree of mastery of each knowledge point; analyze at least one behavioral feature based on the influence coefficient to generate data on the student's mastery of each knowledge point in the target learning course.

[0179] Example 24: Map the mastery data of each knowledge point to the objective function to obtain the mastery probability data of each knowledge point; select knowledge points that meet the unmastered conditions from the knowledge corresponding to the target learning course based on the mastery probability data, and form a set of unmastered knowledge points from the knowledge points selected from the knowledge corresponding to the target learning course.

[0180] Example 25: Generate a sequence of unmastered knowledge points based on the set of unmastered knowledge points; mark the sequence of unmastered knowledge points with unmastered identifiers in the target learning grid, and extract the paths of unmastered knowledge points corresponding to the sequence of unmastered knowledge points from the target learning grid; match the paths of unmastered knowledge points with the target knowledge graph according to the unmastered identifiers to obtain the target knowledge point network corresponding to the set of unmastered knowledge points in the target knowledge graph.

[0181] Example 26: This method is used to match the paths of unmastered knowledge points with the target knowledge graph based on the unmastered identifier, thereby obtaining the network of knowledge points to be screened corresponding to the set of unmastered knowledge points in the target knowledge graph; and to determine the target knowledge point network from the network of knowledge points to be screened based on the prerequisite relationships and relevance information of the paths of unmastered knowledge points in the network of knowledge points to be screened.

[0182] As an alternative approach, this application also provides the following embodiments, but is not limited thereto, including:

[0183] Example 3: Obtain third learning behavior data of the student learning knowledge points in the set of unmastered knowledge points within the knowledge point learning grid; determine first note information and first note information based on the third learning behavior data; evaluate the student's learning needs for knowledge points in the set of unmastered knowledge points based on the first note information and first note information to obtain learning intention intensity data corresponding to knowledge points in the knowledge point learning grid; identify the student's target learning intention for knowledge points in the set of unmastered knowledge points based on the learning intention intensity data, the first note information and first note information; update the set of unmastered knowledge points to obtain a target set of unmastered knowledge points based on the target learning intention and the knowledge point location information of the set of unmastered knowledge points in the knowledge point learning grid, and recommend the target set of unmastered knowledge points to the student.

[0184] Example 31: Based on note information, determine the update frequency of students' notes within the target time range, and evaluate students' learning activity data within the target time range based on the update frequency; based on note information and question information, determine the question frequency of students' doubts about knowledge points in the set of unmastered knowledge points, and evaluate students' difficulty in understanding knowledge points in the set of unmastered knowledge points based on the question frequency; evaluate students' learning needs for knowledge points in the set of unmastered knowledge points based on learning activity data and difficulty in understanding data, and obtain the learning intention intensity data corresponding to knowledge points in the knowledge point learning grid.

[0185] Example 32: Based on the third learning behavior data, determine the learning end time for students to learn the knowledge points in the set of unmastered knowledge points; determine the decay factor of learning intention intensity over time based on the learning end time and time decay coefficient; evaluate the urgency data of students learning the knowledge points in the set of unmastered knowledge points based on the decay factor; perform weighted analysis on learning activity data, comprehension difficulty data and urgency data to evaluate students' learning needs for the knowledge points in the set of unmastered knowledge points, and obtain the learning intention intensity data corresponding to the knowledge points in the knowledge point learning grid.

[0186] Example 33: Based on the first note information, generate student learning content data and corresponding behavioral sequence data for the third learning behavior data; evaluate the student's learning intention for the knowledge points in the set of unmastered knowledge points as a first probability data of deep understanding intention, a second probability data of basic consolidation intention, and a third probability data of conceptual inquiry intention based on the learning intention intensity data, learning content data, and behavioral sequence data; determine the target learning intention from deep understanding intention, basic consolidation intention, and conceptual inquiry intention based on the first probability data, second probability data, and third probability data.

[0187] Example 34: Evaluate the student's required level of understanding of the knowledge points in the set of unmastered knowledge points based on the first probability data, the second probability data, and the third probability data; if the required level of understanding data is greater than the first threshold for required understanding, determine the student's intention to deepen understanding as the target learning intention for the knowledge points in the set of unmastered knowledge points; if the required level of understanding data is less than or equal to the first threshold for required understanding and greater than the second threshold for required understanding, determine the student's intention to consolidate basic knowledge as the target learning intention for the knowledge points in the set of unmastered knowledge points, wherein the second threshold for required understanding is less than the first threshold for required understanding; if the required level of understanding data is less than the second threshold for required understanding, determine the student's intention to raise conceptual questions as the target learning intention for the knowledge points in the set of unmastered knowledge points.

[0188] Example 35: Generate student learning content data based on the first note information; map the third learning behavior data into the knowledge graph corresponding to the knowledge point learning grid to obtain the student's learning knowledge point data in the knowledge graph; perform multi-dimensional information matching between the learning content data and the learning knowledge point data to generate knowledge point location information for the set of unmastered knowledge points; update the set of unmastered knowledge points according to the target learning intention and the knowledge point location information to obtain the target set of unmastered knowledge points, generate the target set of unmastered knowledge points and the target learning path corresponding to the target set of unmastered knowledge points; generate recommendation information for the target set of unmastered knowledge points and the target learning path so that students can master the knowledge points in the target set of unmastered knowledge points based on the target set of unmastered knowledge points and the target learning path.

[0189] Example 36: Obtain third learning behavior data from the learning support system, showing students learning knowledge points in the set of unmastered knowledge points within the knowledge point learning grid; divide the third learning behavior data of students in the target knowledge point learning grid into at least one learning behavior data set, and generate a behavioral feature set corresponding to at least one learning behavior data set; determine the time information corresponding to each behavioral feature in the behavioral feature set, and evaluate the effectiveness of each behavioral feature in the behavioral feature set based on the time information, obtaining the effective value data of each behavioral feature in the behavioral feature set; filter the behavioral features in the behavioral feature set based on the effective value data, obtain the target behavioral features that meet the effective conditions in the behavioral feature set, and form a target behavioral feature set; analyze the students' mastery of knowledge points in the target knowledge point learning grid based on the target behavioral feature set, and determine the set of unmastered knowledge points of students in the target knowledge point learning grid; identify the box-selection behavior from the target behavioral feature set, and classify the learning content selected by the box-selection behavior to obtain the first note information and first note information of students learning knowledge points in the set of unmastered knowledge points.

[0190] As an alternative approach, this application also provides the following embodiments, but is not limited thereto, including:

[0191] Example 4: The third learning behavior data is mapped into the knowledge graph corresponding to the knowledge point learning grid to obtain the student's learning knowledge point data in the knowledge graph; the student's learning content data is generated based on the question information and the note information, and the learning content data is matched with the learning knowledge point data in multiple dimensions to generate the knowledge point location information of the set of unmastered knowledge points.

[0192] Example 41: Generate student learning content data based on question information and note information; perform similarity matching based on the first semantic embedding vector corresponding to the learning content data and the second semantic embedding vector corresponding to the learning knowledge point data to match the content similarity between the learning content data and the learning knowledge point data, and obtain a first similarity matching result; perform similarity matching based on the context data corresponding to the learning content data and the adjacent knowledge point data corresponding to the learning knowledge point data to match the overall similarity between the learning content data and the learning knowledge point data in the knowledge graph, and obtain a second similarity matching result; perform similarity matching based on the first position data of the learning content data in the textbook and the second position data of the learning knowledge point data in the textbook to match the position similarity between the learning content data and the learning knowledge point data, and obtain a third similarity matching result; based on the first similarity matching result, the second similarity matching result, and the third similarity matching result, generate knowledge point location information for the set of unmastered knowledge points.

[0193] Example 42: A fusion analysis is performed based on the first similarity matching result, the second similarity matching result, and the third similarity matching result to evaluate the accuracy of the location of knowledge points in the set of unmastered knowledge points, and to obtain the location reliability data of the learning content data and the learning knowledge point data; based on the location reliability data, unmastered knowledge points that meet the confidence conditions are selected from the set of unmastered knowledge points to form a location knowledge point set; based on the prerequisite relationships and relevance relationships of the knowledge points in the location knowledge point set in the knowledge graph, the knowledge point location information of the set of unmastered knowledge points is generated.

[0194] Example 43: Based on the prerequisite relationships of knowledge points in the knowledge point set located in the knowledge graph, determine the first knowledge point set consisting of the prerequisite knowledge points corresponding to the knowledge point set located in the knowledge graph, and the second knowledge point set with the knowledge points in the knowledge point set as prerequisite knowledge points; extract the knowledge point dependency relationship chain corresponding to the unmastered knowledge point set from the knowledge graph based on the knowledge point set located, the first knowledge point set, and the second knowledge point set; determine the third knowledge point set related to the knowledge point set located in the knowledge graph based on the correlation relationships of knowledge points in the knowledge point set located in the knowledge graph, and extract the knowledge point related relationship chain corresponding to the unmastered knowledge point set based on the knowledge point dependency relationship chain and the knowledge point related relationship chain; generate the knowledge point location information of the unmastered knowledge point set based on the knowledge point dependency relationship chain and the knowledge point related relationship chain.

[0195] As an alternative approach, this application also provides the following embodiments, but is not limited thereto, including:

[0196] Example 5: Based on the student's review path for already learned knowledge points within the target knowledge point learning grid, determine the knowledge points to be reviewed indicated by the review path; extract the network of knowledge points to be reviewed corresponding to the knowledge points to be reviewed from the knowledge graph corresponding to the target knowledge point learning grid, wherein the network of knowledge points to be reviewed includes at least one prerequisite knowledge point and at least one related knowledge point corresponding to the knowledge point to be reviewed; update the review path based on the student's mastery of the knowledge point to be reviewed, the at least one prerequisite knowledge point, and the at least one related knowledge point to obtain the target knowledge point review path corresponding to the knowledge point to be reviewed; determine the student's target knowledge points to be reviewed based on the target knowledge point review path, generate recommendation information for the target knowledge points to be reviewed, and recommend the review of the target knowledge points to the student.

[0197] Example 51: Determine the knowledge point mastery level indicator of the knowledge point to be reviewed from the knowledge point network; when the knowledge point mastery level indicator of the knowledge point to be reviewed is not mastered, determine the first knowledge point mastery level indicator of at least one prerequisite knowledge point and the second knowledge point mastery level indicator of at least one related knowledge point in the knowledge point network to be reviewed; based on the first knowledge point mastery level indicator and the second knowledge point mastery level indicator, update the knowledge point review path to obtain the target knowledge point review path corresponding to the knowledge point to be reviewed.

[0198] Example 52: Based on the mastery level identifier of the first knowledge point and the mastery level identifier of the second knowledge point, determine the set of first knowledge points that the student has mastered from at least one prerequisite knowledge point and at least one related knowledge point; remove the set of first knowledge points from the network of knowledge points to be reviewed, and use the knowledge points to be reviewed as the starting point of the review path to update the network of knowledge points to be reviewed, so as to obtain the target knowledge point review path corresponding to the knowledge points to be reviewed.

[0199] Example 53: Based on the mastery level indicators of the first and second knowledge points, determine the set of second knowledge points that the student has not mastered from at least one prerequisite knowledge point and at least one related knowledge point; determine the learning priority information of the knowledge points in the second knowledge point set; based on the learning priority information, select the first target knowledge point that meets the priority condition from the second knowledge point set; use the first target knowledge point as the starting point of the review path, update the network of knowledge points to be reviewed, and obtain the review path of the target knowledge point.

[0200] Example 54: Determine the mastery probability data and knowledge point difficulty data for each knowledge point in the second knowledge point set; evaluate the learning priority of each knowledge point in the second knowledge point set based on the mastery probability data and knowledge point difficulty data to obtain the learning priority information of each knowledge point in the second knowledge point set; determine the knowledge point with the highest learning priority information in the second knowledge point set as the first target knowledge point.

[0201] Example 55: When the knowledge point mastery level indicator of the knowledge point to be reviewed is marked as "mastered", the knowledge point traversal range is determined based on the knowledge point to be reviewed; the knowledge points are traversed within the knowledge point traversal range to obtain the third set of knowledge points that the student has not mastered; the set of prerequisite knowledge points and the set of related knowledge points corresponding to the third set of knowledge points are determined in the knowledge graph; the second target knowledge point that meets the priority condition is determined from the third set of knowledge points, the set of prerequisite knowledge points, and the set of related knowledge points; the knowledge point to be reviewed is removed from the knowledge point network to be reviewed, and the second target knowledge point is used as the starting point of the review path to update the knowledge point network to obtain the target knowledge point review path.

[0202] As an alternative approach, this application also provides the following embodiments, but is not limited thereto, including:

[0203] Example 6: In response to obtaining the fourth learning behavior data of the student reviewing the target knowledge point to be reviewed, the first mastery level identifier of the target knowledge point to be reviewed is updated; the update influence range of the mastery level of the target knowledge point in the knowledge graph corresponding to the knowledge point learning grid is determined, and a set of knowledge points to be updated is formed based on the knowledge points covered by the update influence range; the mastery level data of the knowledge points in the set of knowledge points to be updated is updated, and the second mastery level identifier of the knowledge points in the set of knowledge points to be updated is updated according to the updated mastery level data; the knowledge point learning grid is updated to the target knowledge point learning grid based on the second mastery level identifier, and the target review content of the student is determined according to the target knowledge point learning grid.

[0204] Example 61: Based on the target prerequisite relationships and target relevance relationships of target knowledge points in the knowledge graph, determine the direct dependencies of target knowledge points, and determine the direct impact range based on the direct dependencies; based on the indirect connection relationships between target knowledge points and other knowledge points in the knowledge graph, determine the indirect dependencies of target knowledge points, and determine the indirect impact range based on the indirect dependencies; determine the set of directly impacted knowledge points covered by the direct impact range, and the set of indirect impacted knowledge points covered by the indirect impact range; generate a set of knowledge points to be updated based on the set of directly impacted knowledge points and the set of indirect impacted knowledge points.

[0205] Example 62: Determine the first indirect dependency relationship established by a target knowledge point through an indirect knowledge point in the knowledge graph, and the second indirect dependency relationship connected by multiple indirect knowledge points; based on the first influence decay degree corresponding to the first indirect dependency relationship, determine the first influence range of the target knowledge point based on the first indirect dependency relationship in the knowledge graph; based on the second influence decay degree corresponding to the second indirect dependency relationship, determine the second influence range of the target knowledge point based on the second indirect dependency relationship in the knowledge graph; determine the indirect influence range based on the first influence range and the second influence range.

[0206] Example 63: Based on direct and indirect dependencies, determine the dependency strength data between knowledge points in the set of knowledge points to be updated and the target knowledge points; identify the importance data of knowledge points in the set of knowledge points to be updated through the target model, which is trained based on students' historical behavioral characteristics and their mastery of historical knowledge points; determine the update priority information for updating knowledge points in the set of knowledge points to be updated based on dependency strength data, importance data, and mastery level data; update the mastery level data according to the update priority information, and update the second mastery level label of the knowledge points in the set of knowledge points to be updated based on the updated mastery level data.

[0207] Example 64: Determine the knowledge point update order list corresponding to the knowledge point set to be updated according to the update priority information, so as to determine the update order of knowledge points in the knowledge point set to be updated; for the first target knowledge point in the knowledge point set to be updated, determine the centrality adjustment coefficient corresponding to the first target knowledge point based on the connection strength data of the first target knowledge point in the knowledge graph, the distance data between the first target knowledge point and other knowledge points, and the degree centrality coefficient. The first target knowledge point is any knowledge point in the knowledge point set to be updated; update the mastery data of the first target knowledge point based on the centrality adjustment coefficient and the change in the mastery data of the target knowledge point; update the second mastery label of the knowledge points in the knowledge point set to be updated based on the updated mastery data of the knowledge points in the knowledge point set to be updated.

[0208] Example 65: When the connection strength data in the centrality adjustment coefficient is greater than the connection strength threshold and the change in the mastery level data is less than the first change threshold, the first target knowledge point is identified as a key knowledge point, and the second mastery level identifier of the first target knowledge point is marked as an unmastered identifier; when the degree centrality coefficient in the centrality adjustment coefficient is greater than the degree centrality threshold and the mastery level data is greater than the second mastery level threshold, the first target knowledge point is identified as a basic knowledge point, and the second mastery level identifier of the first target knowledge point is marked as a mastered identifier.

[0209] As an alternative approach, this application also provides the following embodiments, but is not limited thereto, including:

[0210] Example 7: Obtain fifth learning behavior data of students learning knowledge points in the knowledge point learning grid, and obtain third mastery level identifiers of knowledge points in the knowledge point learning grid; determine second mastery level data of students learning knowledge points in the knowledge point learning grid based on the fifth learning behavior data, and evaluate second knowledge point status data of knowledge points in the knowledge point learning grid based on the second mastery level data; determine abnormal knowledge points from the knowledge point learning grid based on the first knowledge point status data corresponding to the second mastery level data and the second knowledge point status data; update the knowledge graph corresponding to the knowledge point learning grid based on the abnormal knowledge points to obtain an updated target knowledge graph, which is used to generate the target knowledge point learning grid.

[0211] Example 71: Based on the first knowledge point status data and the second knowledge point status data, perform a status data difference analysis to evaluate the degree of difference between the knowledge point mastery status displayed in the knowledge point learning grid and the student's mastery status of the knowledge points in the knowledge point learning grid; based on the fifth learning behavior data, determine the question information and note information of the student's learning of the knowledge points in the knowledge point learning grid; based on the question information, note information, and the second mastery level data, determine the student's learning status information of the knowledge points in the knowledge point learning grid; based on the learning status information and the degree of difference, identify the knowledge points with abnormal status from the knowledge point learning grid; update the knowledge graph corresponding to the knowledge point learning grid based on the knowledge points with abnormal status to obtain the updated target knowledge graph.

[0212] Example 72: Based on note information, determine the frequency of student note updates within a target time range, and evaluate the student's learning activity data for knowledge points in the knowledge point learning grid within the target time range based on the update frequency; based on note information and question information, determine the frequency of student questions about knowledge points in the knowledge point learning grid, and evaluate the student's difficulty in understanding knowledge points in the knowledge point learning grid based on the question frequency; if the learning activity data is greater than the activity threshold and / or the difficulty in understanding data is greater than the difficulty threshold and / or the second mastery level data is less than the mastery level threshold, determine the student's learning status information for knowledge points in the knowledge point learning grid as abnormal learning status information.

[0213] Example 73: When the student's learning status information for knowledge points in the knowledge point learning grid is abnormal, the stability data of knowledge points in the knowledge point learning grid in the knowledge graph is evaluated based on the betweenness centrality data and proximity centrality data of knowledge points in the knowledge point learning grid; the credibility data of abnormal learning status information is determined based on the stability data and the degree of difference; abnormal knowledge points are identified from the knowledge point learning grid based on the credibility data, learning status information, and degree of difference; the knowledge graph corresponding to the knowledge point learning grid is updated based on the abnormal knowledge points to obtain the updated target knowledge graph.

[0214] Example 74: Divide the fifth learning behavior data of students in the target knowledge point learning grid into at least one learning behavior data set, and generate a behavior feature set corresponding to at least one learning behavior data set; determine the time information corresponding to each behavior feature in the behavior feature set, and evaluate the effectiveness of each behavior feature in the behavior feature set based on the time information to obtain the effective value data of each behavior feature in the behavior feature set; filter the behavior features in the behavior feature set based on the effective value data to obtain the target behavior features in the behavior feature set that meet the effective conditions, and form a target behavior feature set; identify the box selection behavior from the target behavior feature set, and classify the learning content selected by the box selection behavior to obtain the question information and note information of students learning knowledge points in the abnormal knowledge point set.

[0215] As an alternative approach, this application also provides the following embodiments, but is not limited thereto, including:

[0216] Example 8: Determine the state correction content corresponding to the abnormal knowledge point; based on the correction content, perform state correction on the abnormal knowledge point in the knowledge point learning grid, and determine the mastery level impact data after the abnormal knowledge point has been state corrected; determine the influence range of the mastery level impact data in the knowledge graph corresponding to the knowledge point learning grid based on the mastery level impact data, and update the knowledge graph within the influence range.

[0217] Example 81: When the first mastery level identifier is a mastery identifier, determine the question information and note information of the student's learning of knowledge points in the knowledge point learning grid based on learning behavior data; evaluate the student's learning needs for knowledge points with abnormal status based on question information and note information, and obtain the learning intention intensity data corresponding to the knowledge points with abnormal status; determine the first mastery level data corresponding to the knowledge points with abnormal status based on the first mastery level identifier; when the learning intention intensity data is greater than the learning intention intensity threshold and the first mastery level data is less than the mastery level threshold, determine the first non-mastery identifier as the second mastery level identifier; determine the first state correction content corresponding to the knowledge points with abnormal status based on the mastery identifier and the first non-mastery identifier.

[0218] Example 82: When the first mastery level identifier is the first non-mastery identifier, the student's question information and note information for learning knowledge points in the knowledge point learning grid are determined based on learning behavior data; the frequency of questions the student has about knowledge points with abnormal status is determined based on the note information and question information, and the student's difficulty in understanding knowledge points with abnormal status is evaluated based on the question frequency; the student's score data for the questions corresponding to the knowledge points with abnormal status is determined; when the difficulty in understanding data is greater than the difficulty in understanding threshold and the score data is less than the score threshold, the second non-mastery identifier is determined as the second mastery level identifier; based on the first non-mastery identifier and the second non-mastery identifier, the second status correction content corresponding to the knowledge points with abnormal status is determined.

[0219] As an alternative approach, this application also provides the following embodiments, but is not limited thereto, including:

[0220] Example 9: Obtain the student's historical learning data; based on the student's historical learning data, determine the student's learning ability data and the student's learning status information for the knowledge points in the knowledge point learning grid; based on the learning ability data and the learning status information, determine the student's learning plan type, and determine the set of knowledge points to be learned in the knowledge point learning grid according to the learning plan type; generate the student's learning path in the knowledge point learning grid based on the set of knowledge points to be learned; determine the set of learning materials that match the learning path from the learning materials, so as to recommend that the student learn the set of knowledge points to be learned based on the set of learning materials.

[0221] Example 91: Extract students' learning characteristics from their historical learning data, and determine their historical learning status for the corresponding historical knowledge points based on these characteristics. The learning characteristics include at least one of learning speed, learning memory, and answering characteristics. Determine students' learning ability data based on their historical learning status. Determine the mastery probability data and centrality data of knowledge points in the knowledge point learning grid, and determine the intensity of students' learning intention for the knowledge points in the knowledge point learning grid based on their historical learning data. Construct a learning status matrix for students' knowledge points in the knowledge point learning grid based on the mastery probability data, centrality data, and intensity of learning intention data. This learning status matrix represents the students' learning status information for the knowledge points in the knowledge point learning grid.

[0222] Example 92: Based on learning ability data and learning status information, determine the time limit of the student's learning plan, and generate at least one learning task information corresponding to the student according to the learning plan time limit; match the at least one learning task information with the knowledge graph corresponding to the knowledge point learning grid to map the at least one learning task information to at least one candidate knowledge point set in the knowledge point learning grid; determine the priority data of the knowledge points in the at least one candidate knowledge point set based on mastery probability data, centrality data, and learning intention intensity data; and form a learning knowledge point set based on the priority data and the knowledge points in the at least one candidate knowledge point set that meet the priority conditions.

[0223] Example 93: Based on the dependencies of knowledge points in the knowledge graph corresponding to the knowledge point learning grid, generate a first knowledge point sequence corresponding to the knowledge points in the set of knowledge points to be learned. The dependencies include prerequisite relationships and relevance relationships. Determine the learning speed information corresponding to the first knowledge point sequence. If the learning speed is greater than the speed threshold, determine the parallel knowledge points corresponding to the first knowledge point sequence, and generate a learning path based on the parallel knowledge points and the first knowledge point sequence. If the learning speed is less than or equal to the speed threshold, generate a learning path based on the first knowledge point sequence.

[0224] Example 94: Based on the prerequisite relationships of knowledge points in the set of knowledge points to be learned in the knowledge graph, determine the set of prerequisite knowledge points corresponding to the set of knowledge points to be learned; generate an initial knowledge point sequence based on the set of knowledge points to be learned, and insert the knowledge points in the set of prerequisite knowledge points into the preceding positions of the knowledge points in the set of knowledge points to be learned in the initial knowledge point sequence to obtain the first knowledge point sequence.

[0225] Example 95: Based on at least one first material feature in the content dimension and at least one second material feature in the question dimension, a feature vector corresponding to the learning material is constructed; the feature vector and student ability data are subjected to correlation matching and similarity matching to obtain a comprehensive matching result of the feature vector and student ability data, and candidate learning materials suitable for the student are selected from the learning materials according to the comprehensive matching result; the student's learning duration is determined based on the student's ability data, and at least one target learning plan is generated for the student according to the learning duration and the learning path to be learned; a set of learning materials matching each of the at least one target learning plan is determined from the learning materials.

[0226] Example 96: Determine the students' already-learned plans in at least one target learning plan, and the set of questions that the students have already learned in the already-learned plans; based on the students' learning of the set of questions they have already learned, analyze the students' mastery data of the knowledge points corresponding to the set of questions they have already learned; update the students' unlearned plans and the set of learning materials corresponding to the unlearned plans in at least one target learning plan based on the mastery data.

[0227] Furthermore, as Figure 1 and Figure 2 The specific implementation of the method shown in this embodiment provides a learning plan generation device based on an AI agent, such as... Figure 8 As shown, the device includes: an acquisition module 31, a generation module 32, a mapping module 33, and a display module 34.

[0228] In some examples of this embodiment, the generation module 32 is specifically configured to: determine a first association type between knowledge points based on basic learning information, determine a second association type between knowledge points based on prerequisite knowledge information, and determine a third association type between knowledge points based on relevance information; generate directed edges corresponding to the knowledge point set based on the first association type between knowledge points corresponding to basic learning information and the second association type between knowledge points corresponding to prerequisite knowledge information, and generate undirected edges corresponding to the knowledge point set based on the first association type and the third association type between knowledge points corresponding to relevance information; determine the target association strength corresponding to the knowledge points in the knowledge point set based on attribute information, and generate an association matrix based on the knowledge point set, directed edges, undirected edges, and association strength; construct graph edges based on the association matrix, and construct graph nodes based on the knowledge point set and attribute information, and combine graph nodes and graph edges to generate a knowledge graph.

[0229] In some examples of this embodiment, the generation module 32 is further configured to: determine the frequency data of knowledge points appearing in the same historical learning session based on attribute information; generate a first association strength corresponding to the knowledge points in the knowledge point set based on the frequency data; determine the mastery level data of knowledge points based on attribute information; generate a second association strength corresponding to the knowledge points in the knowledge point set based on the mastery level data; determine the proximity of the learning order of knowledge points based on attribute information; generate a third association strength corresponding to the knowledge points in the knowledge point set based on the proximity of the learning order; determine the probability data of incorrect answers to knowledge points based on attribute information; generate a fourth association strength corresponding to the knowledge points in the knowledge point set based on the probability data of incorrect answers; and generate a target association strength corresponding to the knowledge points in the knowledge point set based on the first association strength, second association strength, third association strength, and fourth association strength.

[0230] In some examples of this embodiment, the generation module 32 is further configured to perform quality assessment on the knowledge graph; if there are isolated nodes in the knowledge graph, refine the knowledge point set and adjust the knowledge graph based on the refined knowledge point set; if there are nodes in the knowledge graph with a density greater than a density threshold, merge the knowledge point set and adjust the knowledge graph based on the merged knowledge point set.

[0231] In some examples of this embodiment, module 34 is specifically configured to, in response to acquiring learning behavior data of students learning based on a learning plan, determine the students' question information and note information during the learning process based on the learning behavior data; identify implicit knowledge points in the question information and note information, determine the implicit attribute information of the implicit knowledge points, the implicit attribute information including implicit prerequisite knowledge information and implicit relevance information; and, if the implicit attribute information meets the knowledge graph update conditions, update the knowledge graph based on the implicit knowledge points and implicit attribute information to obtain a target knowledge graph, the target knowledge graph being used to generate a learning grid including implicit knowledge points and implicit knowledge point attribute information.

[0232] In some examples of this embodiment, the demonstration module 34 is further configured to: determine the prerequisite knowledge points corresponding to the implicit knowledge points based on implicit prerequisite knowledge information; determine the student's mastery level of the implicit knowledge points and prerequisite knowledge points based on the question information, as well as the answer result data of the questions containing the implicit knowledge points and prerequisite knowledge points; determine the confidence level of the implicit prerequisite knowledge information based on the mastery level and answer result data; and update the graph nodes in the knowledge graph based on the implicit knowledge points and update the directed edges in the knowledge graph based on the implicit prerequisite knowledge information when the confidence level of the implicit prerequisite knowledge information is greater than the confidence level threshold and the number of times the implicit knowledge points are identified is greater than the number of times the implicit knowledge points are identified.

[0233] In some examples of this embodiment, the demonstration module 34 is further configured to: determine the related knowledge points corresponding to the implicit knowledge points based on implicit relevance information; determine the student's mastery of the implicit knowledge points and related knowledge points based on note information, as well as the information contribution data of the note information to the implicit relevance information; determine the confidence level of the implicit relevance information based on the mastery level and information contribution data; and update the graph nodes in the knowledge graph based on the implicit knowledge points and update the undirected edges in the knowledge graph based on the implicit prerequisite knowledge information when the confidence level of the implicit relevance information is greater than the confidence level threshold and the number of times the implicit knowledge points are identified is greater than the number of times the implicit knowledge points are identified.

[0234] Based on the above, Figure 1 and Figure 2 Accordingly, this embodiment also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. Figure 1 and Figure 2 The method shown.

[0235] Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as CD-ROM, USB flash drive, mobile hard drive, etc.) and includes several instructions to cause a computer device (such as personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of this application.

[0236] like Figure 9 The diagram shown is a hardware structure schematic of an electronic device according to the present invention, comprising:

[0237] At least one processor 801; and,

[0238] A memory 802 is communicatively connected to at least one processor 801; wherein,

[0239] The memory 802 stores instructions that can be executed by at least one processor to enable the at least one processor to perform the learning plan generation method based on the AI ​​agent as described above.

[0240] Figure 9 Take the 801 processor as an example.

[0241] The electronic device may also include an input device 803 and a display device 804.

[0242] The processor 801, memory 802, input device 803, and display device 804 can be connected via a bus or other means. Figure 9 Taking the example of a connection between China and Israel via a bus.

[0243] The memory 802, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the program instructions / modules corresponding to the AI ​​agent-based learning plan generation method in the embodiments of this application, for example, Figure 1 and Figure 2 The method flow is shown. The processor 801 executes various functional applications and data processing by running non-volatile software programs, instructions, and modules stored in the memory 802, thereby realizing the AI ​​agent-based learning plan generation method in the above embodiments.

[0244] The memory 802 may include a program storage area and a data storage area. The program storage area may store an operating system and applications required for at least one function; the data storage area may store data created according to the use of the AI ​​agent-based learning plan generation method. Furthermore, the memory 802 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 802 may optionally include memory remotely located relative to the processor 801, and these remote memories may be connected via a network to the apparatus executing the AI ​​agent-based learning plan generation method. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0245] The input device 803 can receive user clicks and generate signal inputs related to user settings and function control of the AI ​​agent-based learning plan generation method. The display device 804 may include a display screen or other display device.

[0246] One or more modules are stored in memory 802, and when run by one or more processors 801, the learning plan generation method based on AI intelligent agents in any of the above method embodiments is executed.

[0247] Optionally, the aforementioned physical devices may also include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, input units such as a keyboard, etc., and optional user interfaces may also include USB interfaces, card reader interfaces, etc. The network interface may optionally include standard wired interfaces, wireless interfaces (such as Wi-Fi interfaces), etc.

[0248] Those skilled in the art will understand that the physical device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or combine certain components, or have different component arrangements.

[0249] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the aforementioned physical device, supporting the operation of information processing programs and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software in the information processing physical device.

[0250] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platform, or it can be implemented by hardware. Compared with existing technologies, this embodiment, by acquiring course knowledge points and related information within a student's target time period from the learning companion system and constructing a knowledge graph by combining basic learning, prerequisites, and relevance information, can transform scattered knowledge points into a structured knowledge network and clearly present the inherent connections between knowledge. By mapping the knowledge graph into an intuitive learning grid of target knowledge points and generating learning plans based on the grid and displaying them as learning paths, it can guide students to understand the logic of knowledge, improving learning efficiency and knowledge transfer ability. By clarifying the association type, distinguishing between directed and undirected edges, and generating an association relationship matrix, it can improve the accuracy and interpretability of the knowledge network. By calculating the target association strength from multiple dimensions such as co-occurrence frequency and mastery, the quantification of the degree of association of knowledge points is more in line with teaching practice and students' learning patterns. By refining isolated nodes and merging high-density nodes through quality assessment, the knowledge graph structure is continuously optimized, ensuring the long-term effectiveness of the graph. By mining implicit knowledge points in learning behavior data and verifying and updating the knowledge graph by combining confidence and recognition counts, the accuracy of learning plans is improved, further helping students understand the deep logic between knowledge points and improving the systematicness and efficiency of learning.

[0251] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0252] The above are merely specific embodiments of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A learning plan generation method based on AI intelligent agents, characterized in that, include: The system retrieves the set of knowledge points corresponding to the course materials of students within the target time period from the learning companion system, and determines the basic learning information of the knowledge points in the set of knowledge points, as well as the prerequisite knowledge information and correlation information between the knowledge points. Based on the basic learning information, the prerequisite knowledge information, and the relevance information, attribute information of knowledge points is generated, and a knowledge graph corresponding to the set of knowledge points is constructed based on the attribute information. The knowledge graph is mapped into the knowledge point learning grid corresponding to the course material information to obtain the target knowledge point learning grid, which is a learning grid that includes the attribute information of knowledge points and the relationship information between knowledge points. Based on the target knowledge point learning grid, a learning plan for the student within the target time period is generated, and the learning plan is displayed to the student in the target knowledge point learning grid in the form of a learning path; In response to obtaining the student's learning behavior data based on the learning plan, the student's question information and note information during the learning process are determined based on the learning behavior data. Identify the implicit knowledge points in the question information and the note information, and determine the implicit attribute information of the implicit knowledge points. The implicit attribute information includes implicit prerequisite knowledge information and implicit relevance information. Based on the implicit prerequisite knowledge information, the prerequisite knowledge points corresponding to the implicit knowledge points are determined; Based on the question information, determine the student's mastery of the implicit knowledge points and the prerequisite knowledge points, as well as the answer results data of the questions containing the implicit knowledge points and the prerequisite knowledge points; The confidence level of the implicit prior knowledge information is determined based on the level of mastery and the answer results data; In response to the implicit prior knowledge information having a confidence level greater than a confidence level threshold and the number of times the implicit knowledge point is identified being greater than a number of times threshold, the graph nodes in the knowledge graph are updated based on the implicit knowledge point, and the directed edges in the knowledge graph are updated based on the implicit prior knowledge information. Based on the implicit correlation information, the relevant knowledge points corresponding to the implicit knowledge points are determined; Based on the notes information, determine the student's mastery of the implicit knowledge points and related knowledge points, as well as the information contribution data of the notes information to the implicit related information; The confidence level of the implicit correlation information is determined based on the degree of mastery and the information contribution data. In response to the fact that the confidence level of the implicit correlation information is greater than the confidence level threshold and the number of times the implicit knowledge point is identified is greater than the number of times the implicit knowledge point is identified, the graph nodes in the knowledge graph are updated based on the implicit knowledge point, and the undirected edges in the knowledge graph are updated based on the implicit correlation information; A target knowledge graph is obtained, which is used to generate a learning grid that includes the implicit knowledge points and their attribute information.

2. The method according to claim 1, characterized in that, The construction of the knowledge graph corresponding to the knowledge point set based on the attribute information includes: Based on the basic learning information, a first association type between the knowledge points is determined; based on the prerequisite knowledge information, a second association type between the knowledge points is determined; and based on the relevance information, a third association type between the knowledge points is determined. Based on the first association type between knowledge points corresponding to the basic learning information and the second association type between knowledge points corresponding to the prerequisite knowledge information, directed edges corresponding to the knowledge point set are generated, and based on the first association type and the third association type between knowledge points corresponding to the relevance information, undirected edges corresponding to the knowledge point set are generated. Based on the attribute information, the target association strength corresponding to the knowledge points in the knowledge point set is determined, and an association matrix is ​​generated based on the knowledge point set, the directed edges, the undirected edges, and the association strength. The graph edges are constructed based on the association matrix, and the graph nodes are constructed based on the knowledge point set and the attribute information. The graph nodes and graph edges are combined to generate the knowledge graph.

3. The method according to claim 2, characterized in that, Determining the target association strength corresponding to the knowledge points in the knowledge point set based on the attribute information includes: Based on the attribute information, determine the frequency data of the knowledge points appearing in the same historical learning session, and generate the first association strength corresponding to the knowledge points in the knowledge point set based on the frequency data. Based on the attribute information, the mastery level data of the knowledge points is determined, and the second association strength corresponding to the knowledge points in the knowledge point set is generated according to the mastery level data; Based on the attribute information, the proximity of the learning order of knowledge points is determined, and a third association strength corresponding to the knowledge points in the knowledge point set is generated according to the proximity of the learning order. Based on the attribute information, determine the probability data of incorrect answers for knowledge points, and generate the fourth association strength corresponding to the knowledge points in the knowledge point set based on the probability data of incorrect answers. Generate the target association strength corresponding to the knowledge points in the knowledge point set based on the first association strength, the second association strength, the third association strength, and the fourth association strength.

4. The method according to claim 2, characterized in that, After constructing graph edges based on the association matrix and graph nodes based on the knowledge point set and the attribute information, and combining the graph nodes with the graph edges to generate the knowledge graph, the method further includes: The knowledge graph is subjected to quality assessment. If there are isolated nodes in the knowledge graph, the knowledge point set is refined. The knowledge graph is then adjusted based on the refined knowledge point set. If there are nodes in the knowledge graph whose density is greater than a density threshold, the knowledge point set is merged, and the knowledge graph is adjusted based on the merged knowledge point set.

5. A learning plan generation device based on an AI intelligent agent, characterized in that, include: The acquisition module is configured to acquire the set of knowledge points corresponding to the course materials of students within the target time period from the learning companion system, and determine the basic learning information of the knowledge points in the set of knowledge points, as well as the prerequisite knowledge information and correlation information between the knowledge points. The generation module is configured to generate attribute information of knowledge points based on the basic learning information, the prerequisite knowledge information and the relevance information, and to construct a knowledge graph corresponding to the set of knowledge points based on the attribute information. The mapping module is configured to map the knowledge graph into the knowledge point learning grid corresponding to the course material information to obtain the target knowledge point learning grid, wherein the target knowledge point learning grid is a learning grid that includes the attribute information of knowledge points and the association information between knowledge points. The display module is configured to generate a learning plan for the student within the target time period based on the target knowledge point learning grid, and to display the learning plan to the student in the target knowledge point learning grid in the form of a learning path; In response to obtaining the student's learning behavior data based on the learning plan, the student's question information and note information during the learning process are determined based on the learning behavior data. Identify the implicit knowledge points in the question information and the note information, and determine the implicit attribute information of the implicit knowledge points. The implicit attribute information includes implicit prerequisite knowledge information and implicit relevance information. Based on the implicit prerequisite knowledge information, the prerequisite knowledge points corresponding to the implicit knowledge points are determined; Based on the question information, determine the student's mastery of the implicit knowledge points and the prerequisite knowledge points, as well as the answer results data of the questions containing the implicit knowledge points and the prerequisite knowledge points; The confidence level of the implicit prior knowledge information is determined based on the level of mastery and the answer results data; In response to the implicit prior knowledge information having a confidence level greater than a confidence level threshold and the number of times the implicit knowledge point is identified being greater than a number of times threshold, the graph nodes in the knowledge graph are updated based on the implicit knowledge point, and the directed edges in the knowledge graph are updated based on the implicit prior knowledge information. Based on the implicit correlation information, the relevant knowledge points corresponding to the implicit knowledge points are determined; Based on the notes information, determine the student's mastery of the implicit knowledge points and related knowledge points, as well as the information contribution data of the notes information to the implicit related information; The confidence level of the implicit correlation information is determined based on the degree of mastery and the information contribution data. In response to the fact that the confidence level of the implicit correlation information is greater than the confidence level threshold and the number of times the implicit knowledge point is identified is greater than the number of times the implicit knowledge point is identified, the graph nodes in the knowledge graph are updated based on the implicit knowledge point, and the undirected edges in the knowledge graph are updated based on the implicit correlation information; A target knowledge graph is obtained, which is used to generate a learning grid that includes the implicit knowledge points and their attribute information.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 4.

7. An electronic device comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 4.

Images