Teaching knowledge graph node mapping method and system based on learning state reasoning

Abstract

The invention belongs to the technical field of artificial intelligence, and relates to a teaching knowledge graph node mapping method and system based on learning state reasoning, and the method comprises the steps: S1, obtaining learning feedback information; s2, performing CoT multi-step reasoning analysis to obtain a learning state reasoning result; s3, judging the learning state to determine whether knowledge point nodes of the teaching knowledge graph need to be updated or not, generating a candidate knowledge point set in combination with the structure of the teaching knowledge graph when the knowledge point nodes need to be updated, matching a learning state reasoning result with the candidate knowledge point set according to a mapping rule, determining a target knowledge point, and generating a mapping relation; and S4, on the basis of the mapping relationship, updating the state of the corresponding knowledge point node in the teaching knowledge graph. A learning state reasoning mechanism based on CoT is introduced and a reasoning result is mapped to a teaching knowledge graph node, so that refined diagnosis of a learning state is realized, and teaching pertinence, interpretability and stability of an intelligent teaching system are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence technology and relates to a method and system for mapping nodes in a teaching knowledge graph, particularly a method and system for mapping nodes in a teaching knowledge graph based on learning state reasoning. Background Technology

[0002] With the continuous development of artificial intelligence technology in the field of educational informatization, Large Language Models (LLM), intelligent teaching systems, and digital human teaching platforms are gradually being applied to scenarios such as online teaching, intelligent tutoring, adaptive learning, and personalized teaching. In these applications, the teaching system not only needs to present teaching content to students, but also needs to analyze students' comprehension, error performance, and cognitive state during the learning process in order to achieve intelligent support for the teaching process.

[0003] In an ideal intelligent teaching scenario, the teaching system should be able to accurately diagnose students' knowledge mastery based on their learning feedback, and provide a basis for subsequent teaching decisions. However, in existing digital teaching systems, the modeling and expression of students' learning status are still relatively coarse-grained. Students' learning problems are often difficult to effectively link with specific teaching knowledge structures, which restricts the teaching system's ability to perform refined diagnosis and adaptive teaching.

[0004] Especially when students lack understanding of certain knowledge points, have a weak grasp of concepts, or experience cognitive gaps during the learning process, existing systems can usually only provide an overall evaluation or simple prompts, making it difficult to pinpoint the specific knowledge points corresponding to the student's problem, thus affecting the pertinence and effectiveness of teaching interventions.

[0005] In recent years, research and system implementations related to student learning status analysis and the organization of teaching knowledge can be mainly summarized into the following technical approaches: 1. A scheme for representing student learning status based on grades or outcome evaluation.

[0006] This type of approach primarily uses students' accuracy in answering questions, exam scores, or periodic assessment results to characterize their learning status, and uses this as a basis for teaching feedback or adjustments. Students' learning performance is usually summarized into overall status labels such as "mastered" or "not mastered."

[0007] Although such solutions are simple to implement and easy to deploy, they are essentially learning status assessments at the outcome level. They cannot reflect the specific problems and causes of errors students encounter in the process of understanding knowledge, and are therefore unable to support the accurate identification of weak links in students' knowledge structure.

[0008] 2. Student model construction scheme based on statistical features or behavioral data.

[0009] Some intelligent teaching systems are attempting to incorporate statistical features such as learning duration, click behavior, and number of practice sessions to build student learning models, which are used to analyze students' learning engagement or learning habits.

[0010] However, this type of method focuses on data analysis at the behavioral level, making it difficult to depict students' cognitive paths and reasoning states during the knowledge comprehension process, and unable to accurately distinguish different types of learning errors. Therefore, it has obvious limitations at the knowledge diagnosis level.

[0011] 3. A scheme for separating student models from teaching knowledge structures in modeling.

[0012] In some systems, student learning state models and instructional knowledge graphs are constructed separately: on the one hand, student models are used to record learning performance and state changes; on the other hand, instructional knowledge graphs are used to describe the logical relationships and prerequisite dependencies between knowledge points.

[0013] Then, due to the lack of a unified mapping mechanism between the two, students' learning status is difficult to be explicitly expressed in the teaching knowledge structure, and the teaching system cannot directly reflect the location of students' weak knowledge points at the knowledge graph level, thus affecting the interpretability and controllability of subsequent teaching decisions.

[0014] 4. Error localization schemes based on rules or human experience.

[0015] Some teaching systems use manually set rules or based on teaching experience to simply categorize student errors and associate them with a specific knowledge module or chapter.

[0016] This type of approach alleviates the problem of students misidentifying information to some extent, but it relies on human experience and predefined rules, lacks a unified reasoning mechanism, is difficult to adapt to the diversity and complexity of students' learning paths, and has limited scalability and generalization ability.

[0017] In summary, while existing student learning status analysis and knowledge association schemes have improved the intelligence level of teaching systems to some extent, they still generally suffer from the following key technical deficiencies: (1) The expression of student learning status is abstract and difficult to locate specific knowledge points; (2) There is a lack of deep reasoning ability regarding the causes of student errors and cognitive processes; (3) There is a lack of effective mapping mechanism between student learning status and teaching knowledge graph; (4) Learning diagnosis results are difficult to express explicitly at the knowledge structure level; and (5) It is difficult to provide a structured basis for subsequent adaptive teaching and digital human teaching strategies.

[0018] Based on the above analysis, there is an urgent need for a technical solution that can perform in-depth reasoning analysis on students' learning status and accurately map the reasoning results to nodes in the teaching knowledge graph. This would enable students' learning problems to be explicitly expressed at the knowledge structure level, thus providing a reliable structured foundation for intelligent teaching systems to achieve refined learning diagnosis, adaptive teaching decisions, and digital human teaching. Summary of the Invention

[0019] To overcome the shortcomings of existing technologies, this invention proposes a teaching knowledge graph node mapping method and system based on learning state reasoning. In the process of analyzing students' learning states, it introduces a cognitive state reasoning mechanism based on CoT multi-step reasoning, uses the reasoned causes of students' errors and learning states as the mapping basis, and accurately locates students' learning problems to the corresponding knowledge point nodes within the structural framework of the teaching knowledge graph, thereby achieving refined learning diagnosis oriented towards knowledge structure.

[0020] To achieve the above objectives, the present invention provides the following technical solution: A method for mapping nodes in a teaching knowledge graph based on learning state reasoning, characterized by the following steps: S1: Obtain students' learning feedback information during the teaching process and perform structured preprocessing; S2: Based on the structured preprocessed learning feedback information, the CoT multi-step reasoning analysis is performed on the learning state of students in the learning process using a large language model, so as to map the learning feedback information into interpretable learning state reasoning results. S3: Based on the learning state reasoning result, determine the student's learning state during the learning process to determine whether the knowledge point nodes of the teaching knowledge graph need to be updated. When it is determined that the knowledge point nodes of the teaching knowledge graph need to be updated, generate a set of candidate knowledge points related to the knowledge points that the student is currently learning by combining the structure of the teaching knowledge graph. According to the preset or learned mapping rules, match the learning state reasoning result with the set of candidate knowledge points to determine the target knowledge point corresponding to the learning state reasoning result and generate the mapping relationship between the learning state reasoning result and the knowledge point nodes of the teaching knowledge graph. S4: Based on the mapping relationship between the learning state reasoning results and the knowledge point nodes of the teaching knowledge graph, update the state of the corresponding knowledge point nodes in the teaching knowledge graph.

[0021] Preferably, in step S1, the learning feedback information obtained includes one or more of the following: the student's answers to the teaching content, exercises, or test questions; the student's interactive behavior data during the teaching process; the types of errors, pauses, repeated answers, or repeated requests for explanations that the student makes during the learning process; the student's historical learning records, periodic learning results, or mastery of knowledge points.

[0022] Preferably, step S2 specifically includes: S21: Convert the structured preprocessed learning feedback information into an inference input package; S22: Based on the inference input package, a fixed multi-step inference template is used to guide the large language model to perform CoT multi-step inference analysis, so as to gradually output intermediate judgment results and finally output structured learning state inference results. The structured learning state inference results include state labels, error cause types, error cause evidence, and confidence level.

[0023] Preferably, the confidence level is determined based on a confidence score, and the confidence score is: , In the formula, the evidence consistency score The sufficiency of evidence score is used to measure whether the status label is consistent with objective evidence. The reasoning stability score is used to measure whether the evidence in a reasoning argument is sufficient and stable. Used to measure whether the state label is stable under the same input. , and It is a preset coefficient and satisfies ; Furthermore, determining the confidence level based on the confidence score specifically involves: when The confidence level is high confidence when The confidence level is medium confidence. The confidence level is low confidence, among which, For high confidence threshold, This is the low confidence threshold.

[0024] Preferably, in step S3, when the learning state reasoning result indicates that the student has at least one of the following situations in the current or subsequent knowledge points, it is determined that the knowledge point nodes of the teaching knowledge graph need to be updated: insufficient understanding of a specific knowledge point, confusion or misunderstanding between related concepts, or obstruction of understanding of the current knowledge point due to the lack of prior knowledge points.

[0025] Preferably, in step S3, generating a set of candidate knowledge points related to the knowledge points the student is currently learning, based on the teaching knowledge graph structure, and matching the learning state reasoning result with the set of candidate knowledge points according to preset or learned mapping rules, determining the target knowledge point corresponding to the learning state reasoning result, and generating the mapping relationship from the learning state reasoning result to the knowledge point node of the teaching knowledge graph specifically includes: S31: Based on the knowledge points that students are currently learning Centered on this, a set of candidate knowledge points is formed by retrieving them from the teaching knowledge graph. The candidate knowledge point set It should include at least: the knowledge points currently being learned. Knowledge points currently being learned Prerequisite knowledge points and / or dependent knowledge points, and knowledge points currently being learned Related knowledge points with connections, and knowledge points currently being learned. Knowledge points at the same level; S32: For the set of candidate knowledge points Each candidate knowledge point The learning state reasoning result is matched with the learning state reasoning result using rule matching and semantic similarity to determine the target knowledge point corresponding to the learning state reasoning result; S33: Based on the target knowledge points, establish a mapping relationship between the learning state reasoning results and the knowledge point nodes of the teaching knowledge graph.

[0026] Preferably, in step S4, when updating the state of the corresponding knowledge point node in the teaching knowledge graph, different update strategies are adopted according to the confidence level: when the confidence level is high, a strong update is performed on the corresponding knowledge point node; when the confidence level is medium, a weak update or smooth update is performed on the corresponding knowledge point node; when the confidence level is low, no update is performed on the corresponding knowledge point node.

[0027] Furthermore, this invention also provides a teaching knowledge graph node mapping system based on learning state reasoning, characterized in that it includes: The module for acquiring and preprocessing student learning feedback information is used to acquire students' learning feedback information during the teaching process and perform structured preprocessing. The CoT-based multi-step reasoning analysis module for student learning status is used to perform CoT multi-step reasoning analysis on the learning status of students in the learning process based on structured preprocessed learning feedback information and using a large language model, so as to map the learning feedback information into interpretable learning status reasoning results. The module for updating the requirement determination and knowledge point mapping relationship generation is used to determine the student's learning state during the learning process based on the learning state reasoning result to determine whether the knowledge point nodes of the teaching knowledge graph need to be updated. When it is determined that the knowledge point nodes of the teaching knowledge graph need to be updated, a set of candidate knowledge points related to the knowledge points that the student is currently learning is generated in combination with the structure of the teaching knowledge graph. According to the preset or learned mapping rules, the learning state reasoning result is matched with the set of candidate knowledge points to determine the target knowledge point corresponding to the learning state reasoning result and generate the mapping relationship from the learning state reasoning result to the knowledge point nodes of the teaching knowledge graph. The knowledge point node status update module of the teaching knowledge graph is used to update the status of the corresponding knowledge point node in the teaching knowledge graph based on the mapping relationship between the learning state reasoning result and the knowledge point node in the teaching knowledge graph.

[0028] Furthermore, the present invention also provides a teaching knowledge graph node mapping device based on learning state reasoning, characterized in that it includes: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the teaching knowledge graph node mapping method based on learning state reasoning as described above. Finally, the present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the program is executed by a processor, it implements the steps of the teaching knowledge graph node mapping method based on learning state reasoning as described above.

[0029] Compared with existing teaching systems that rely solely on evaluation of answer results, statistical feature analysis, or manual rules to determine students' learning status, the teaching knowledge graph node mapping method and system proposed in this invention have the following significant advantages in terms of learning status diagnosis accuracy, knowledge structure expression ability, and interpretability of teaching decisions: (1) Deep diagnosis of learning state based on CoT reasoning, fundamentally improving the accuracy of locating weak knowledge points.

[0030] Existing teaching systems typically rely on accuracy rates, scores, or simple labels to analyze student learning status, lacking in-depth analysis of students' cognitive processes and the causes of errors. This makes it difficult to accurately identify students' true weaknesses in knowledge comprehension. This invention introduces a chain-of-thought-based multi-step reasoning mechanism to deeply analyze student learning feedback, characterizing the formation path of errors and their corresponding cognitive states. Based on the reasoning results, it determines students' understanding of specific knowledge points. Through this mechanism, the teaching system can diagnose student learning problems from the perspective of "why mistakes were made," significantly improving the accuracy and reliability of locating students' weak knowledge points at the mechanism level.

[0031] (2) The learning status is accurately mapped to the knowledge point nodes of the teaching knowledge graph to realize the explicit expression of the knowledge structure level.

[0032] Unlike existing methods that confine student learning status to abstract evaluations or representations within student models, this invention establishes a mapping between student error types and knowledge point nodes in a teaching knowledge graph. This allows students' learning status to be explicitly expressed within the teaching knowledge structure as changes in node status. This approach transforms student learning problems from isolated results into direct connections to specific knowledge point nodes, clearly revealing students' weaknesses at the knowledge structure level and providing the teaching system with analyzable and traceable structured learning diagnostic results.

[0033] (3) An efficient diagnostic mechanism that does not require additional teaching content or extra teaching burden, thereby improving system operating efficiency.

[0034] Unlike traditional methods that improve diagnostic accuracy by increasing assessment frequency, repeating explanations, or introducing additional teaching content, this invention does not change the teaching content itself. Instead, it analyzes and reasons from existing student learning feedback and maps learning states into a teaching knowledge graph. This approach achieves refined diagnosis and expression of student learning states without increasing teaching time or introducing additional teaching resources. It effectively avoids redundant teaching processes and increased system complexity, enabling the teaching system to obtain higher-quality learning state modeling capabilities at a lower cost.

[0035] (4) Expressing learning status under the constraints of the teaching knowledge graph structure significantly enhances the logical consistency and interpretability of diagnostic results.

[0036] In some existing solutions, the analysis results of student learning status lack clear knowledge structure support, making it difficult to explain the relationship between student errors and the knowledge system. This invention maps student learning status within the structural framework of a teaching knowledge graph, naturally subjecting diagnostic results to the constraints of prior knowledge relationships and teaching logic. Through this mechanism, the expression of student learning status always conforms to the internal structure of the teaching knowledge system, avoiding inconsistencies between diagnostic results and cognitive patterns. This significantly enhances the stability, interpretability, and credibility of the teaching system in complex teaching scenarios.

[0037] (5) Provide a unified and scalable structured foundation for adaptive teaching and digital human teaching.

[0038] By mapping student learning states to knowledge point nodes in a teaching knowledge graph, this invention provides a unified, structured foundation for subsequent teaching strategy generation, adaptive teaching path planning, and step-by-step explanations by digital humans. The teaching system can directly generate targeted teaching decisions and behaviors based on changes in the state of knowledge point nodes in the knowledge graph. This mechanism realizes the transformation of student learning states from "implicit internal representation" to "explicit knowledge structure representation," providing stable and scalable technical support for upgrading intelligent teaching systems from simple feedback modes to adaptive teaching modes with cognitive understanding capabilities.

[0039] In summary, this invention introduces a learning state reasoning mechanism based on chain-of-thought multi-step reasoning and maps the reasoning results to knowledge point nodes in the teaching knowledge graph, thereby achieving refined diagnosis and structured expression of students' learning states and significantly improving the teaching relevance, interpretability, and system stability of the intelligent teaching system. Attached Figure Description

[0040] Figure 1 This is a flowchart of the teaching knowledge graph node mapping method based on learning state reasoning of the present invention.

[0041] Figure 2 This is a schematic diagram of the teaching knowledge graph node mapping system based on learning state reasoning of the present invention.

[0042] Figure 3 This is a structural block diagram of the teaching knowledge graph node mapping device based on learning state reasoning of the present invention. Detailed Implementation

[0043] Before detailing any embodiment of the invention, it should be understood that the invention, in its application, is not limited to the details of the construction and arrangement of the components set forth in the following description or illustrated in the following figures. The invention can have other embodiments and can be practiced or carried out in various ways. Furthermore, it should be understood that the wording and terminology used herein are for descriptive purposes and should not be considered limiting. The use of “comprising” or “having” and variations thereof in this invention is intended to cover the items set forth below and their equivalents, as well as any additional items. Unless otherwise specified or limited, the terms “installation,” “connection,” “support,” and “linkage,” and variations thereof are used broadly and cover both direct and indirect installation, connection, support, and linking. Moreover, “connection” and “linkage” are not limited to physical or mechanical connections or links. Furthermore, firstly, in the disclosure of this invention, the terms "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the above terms should not be construed as limiting this invention. Secondly, the term "a" should be understood as "at least one" or "one or more," that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple. The term "a" should not be construed as a limitation on the quantity.

[0044] With the widespread application of large language models in intelligent teaching systems, digital human teaching, and personalized learning scenarios, teaching systems are gradually gaining the ability to automatically analyze and provide feedback on students' learning processes. However, limited by existing student state modeling methods and teaching knowledge structure representation mechanisms, current digital teaching systems still face the following prominent technical challenges in the refined diagnosis of students' learning states and the correlation of knowledge structures: (1) Students’ learning status is expressed in an abstract way, making it difficult to pinpoint specific weak knowledge points.

[0045] In existing teaching systems, students' learning status is usually represented by overall evaluation results such as "correct / incorrect" or "mastered / not mastered," or simply by statistical indicators such as scores and accuracy rates. This type of status expression lacks fine-grained analysis of the causes of students' errors and deviations in knowledge understanding, and cannot clearly point out the specific knowledge points corresponding to students' problems, making it difficult for the teaching system to accurately diagnose students' true knowledge mastery.

[0046] (2) The existing learning status analysis mechanism lacks the ability to reason about the causes of errors.

[0047] Even though some teaching systems have introduced intelligent analysis or recommendation algorithms, their judgments on students' learning status are mostly based on surface behavioral characteristics or historical statistical information, lacking in-depth analysis of students' cognitive processes and error formation paths. These methods cannot effectively distinguish between different types of error causes, such as conceptual confusion, missing prerequisite knowledge, or incorrect reasoning paths, thus limiting the depth of the teaching system's understanding and explanatory ability regarding students' learning status.

[0048] (3) There is no effective mapping mechanism between students’ learning status and teaching knowledge structure.

[0049] In existing technologies, student learning status and the structure of teaching knowledge points are usually modeled independently: on the one hand, the student model is used to record learning performance; on the other hand, the teaching knowledge graph is used to describe the logical relationships between knowledge points. Due to the lack of a unified mechanism to map student learning status to nodes in the teaching knowledge graph, the teaching system struggles to explicitly express students' weaknesses at the knowledge structure level, thus affecting the accuracy and controllability of adaptive teaching decisions.

[0050] To address the aforementioned problems, this invention proposes a method and system for mapping student learning states to knowledge point nodes in a teaching knowledge graph based on chain-of-thought reasoning. This invention introduces a cognitive state reasoning mechanism based on multi-step reasoning into the student learning state analysis process. The reasonsing-derived causes of student errors and learning states are used as the mapping basis. Within the structural framework of the teaching knowledge graph, student learning problems are precisely located to corresponding knowledge point nodes, thereby achieving refined learning diagnosis oriented towards knowledge structure.

[0051] Specifically, this invention effectively enhances students' learning state modeling and expression abilities through the following methods: when reasoning analysis shows that students have a sufficient understanding of the current knowledge point and a complete reasoning path, the normal state of the corresponding knowledge point node is maintained in the teaching knowledge graph to reflect their stable mastery; when reasoning analysis finds that students have weak understanding, conceptual confusion, or reasoning breaks in certain knowledge points, the learning state is mapped to the corresponding knowledge point node in the teaching knowledge graph, and the state of the knowledge point node is updated to explicitly mark the weak positions of students in the knowledge structure.

[0052] This invention achieves technological breakthroughs primarily through the following three aspects: (1) Introduce a chain-of-thought-based reasoning mechanism for students’ learning status: In the teaching process, perform multi-step reasoning analysis on students’ answer results, interactive behavior and other learning feedback information to characterize the reasons for students’ errors and their corresponding cognitive states, so as to achieve a deeper understanding of students’ learning status, rather than just making judgments based on superficial results.

[0053] (2) Establish a mapping mechanism from student error type (status label) to teaching knowledge graph node: Based on the reasoning results, construct the mapping relationship between student error type (status label) and knowledge point node in the teaching knowledge graph, so that students' learning problems can be accurately located to specific knowledge points, and realize effective mapping from student state space to knowledge structure space.

[0054] (3) Realize the explicit expression of students’ learning status at the level of teaching knowledge structure: By dynamically updating the status of knowledge point nodes in the teaching knowledge graph, students’ learning status is explicitly represented in the teaching knowledge graph in a structured form, providing a reliable structured basis for subsequent adaptive teaching strategy generation, step-by-step explanation of digital human and adjustment of teaching path.

[0055] Figure 1 A flowchart of the teaching knowledge graph node mapping method based on learning state reasoning of the present invention is shown. Figure 1 As shown, the teaching knowledge graph node mapping method based on learning state reasoning of the present invention includes the following steps: S1: Acquisition and structured preprocessing of student learning feedback information.

[0056] During the teaching process, student learning feedback information is acquired and used as the foundational input for subsequent student learning status reasoning analysis and instructional knowledge graph mapping. This feedback information is used to characterize students' current learning performance and behavioral traits.

[0057] Specifically, the learning feedback information includes at least one or more of the following: (1) Students' answers to teaching content, exercises, or test questions; (2) Student interactive behavior data during the teaching process, including behaviors such as clicking, pausing, repeating playback, and requesting prompts; (3) The types of errors students make during the learning process, pauses, and behavioral characteristics such as repeatedly answering questions or repeatedly requesting explanations; (4) Students’ historical learning records, periodic learning results or mastery of knowledge points.

[0058] After obtaining the learning feedback information, data cleaning and consistency processing can be performed on the learning feedback information, including deleting abnormal or missing feedback information, aligning feedback information from different sources in time, and associating learning behaviors with corresponding teaching units or knowledge points.

[0059] Subsequently, the processed learning feedback information is converted into a structured representation to form student learning feedback information for subsequent reasoning and analysis. This structured preprocessed learning feedback information includes at least: current teaching knowledge point identifiers, student response information, a summary of learning behavior characteristics, and optional historical learning context information.

[0060] S2: CoT-based multi-step reasoning analysis of student learning status.

[0061] Based on the structured preprocessed learning feedback information obtained in step S1, a Chain-of-Thought (CoT) multi-step reasoning analysis is performed on the student's current learning state using a large language model to achieve a deep understanding of the student's cognitive state, thereby mapping the learning feedback information into interpretable learning state reasoning results.

[0062] Specifically, the CoT multi-step inference analysis process is used to analyze at least one or more of the following aspects: (1) Students' understanding and mastery of the current knowledge points; (2) Students may have weak knowledge or confusion about concepts during the learning process; (3) The underlying causes of student errors and their corresponding cognitive paths; (4) Whether there is a break in reasoning or a lack of prior knowledge points in the process of students’ understanding of knowledge.

[0063] Through CoT multi-step reasoning analysis, students' learning feedback information is transformed into interpretable learning state reasoning results, which reflect students' current cognitive state and knowledge mastery level.

[0064] In this invention, the learning state reasoning result can be further structured into state labels, error cause descriptions (including error cause types and error cause evidence), and corresponding confidence levels.

[0065] Preferably, a consistency check can also be performed on the reasoning results of the learning state. When there is a significant conflict between the reasoning conclusion of the learning state reasoning results and the student's answer, a second reasoning can be triggered or a conservative judgment strategy can be adopted to improve the stability and reliability of the analysis results.

[0066] In this invention, the Chain-of-Thought (CoT) multi-step reasoning analysis specifically includes the following steps: 1. Construction of the Evidence Pack for Inference.

[0067] The structured learning feedback information obtained in step S1 is organized into a reasoning input package. The reasoning input package includes at least: the current knowledge point ID / name, the question and standard answer (or scoring points), the student's answer, key process information (such as steps / intermediate steps), a summary of learning behavior (pause duration, repeated viewing, number of prompts, number of retakes, etc.), and necessary historical context (previous knowledge point mastery records, etc.).

[0068] 2. Multi-step reasoning template constraint.

[0069] To ensure stable and controllable reasoning, a fixed multi-step reasoning template is used to guide the large language model to perform CoT multi-step reasoning analysis, so as to gradually output intermediate judgment results (the intermediate judgment structure may not be exposed to the outside world, but it must be traceable) and finally output the structured learning state reasoning results.

[0070] In this invention, the step of using a fixed multi-step inference template to guide a large language model for CoT multi-step inference analysis may include: Step A: Identify the target ability and scoring points for the student's current knowledge point (e.g., what is being tested in this question / what are the key conditions, etc.). Step B: Compare the student's answers to extract "correct evidence / incorrect evidence" (e.g., which steps were correct, where they were wrong, etc.); Step C: Map the error to candidate cause types (e.g., conceptual confusion, missing prerequisites, calculation errors, misinterpretation of the question, break in reasoning, etc.). Step D: Verify the reasons by combining behavioral characteristics and historical records (e.g., repeatedly watching a video → getting stuck on a concept; frequently requesting hints → weak prior knowledge, etc.). Step E: Learn the state inference results (state label + error cause description + confidence level).

[0071] 3. Structured output of learning status.

[0072] The learning state inference results are organized into structured fields, including: Status labels: such as mastered / partially mastered / not mastered / conceptual confusion / missing prior knowledge / broken reasoning, etc.; Error causes include: misreading the question, misuse of concepts, omission of conditions, missing steps, calculation errors, and incomplete conclusions. Evidence of the causes of errors: Key terms extracted from students' answers and reasoning explanations, such as "missed solution", "unclear definition", "unable to complete the square", "misremembered formula", etc. The confidence level (low / medium / high, or 0~1 score) is used to select subsequent knowledge point node update strategies.

[0073] In this invention, the confidence level can be obtained by quantitatively assessing the consistency and sufficiency of "reasoning conclusion - response evidence - behavioral evidence - historical evidence". The confidence score can be expressed as a numerical value. The confidence level is determined based on the confidence score, i.e., the confidence level is classified as high / medium / low. The confidence score calculation includes at least the following factors: (1) Evidence consistency score : Used to measure whether the reasoning conclusion (status label) is consistent with objective evidence.

[0074] For example, if the reasoning conclusion is "mastered", but students are missing points / key steps and their scores are significantly low, then consistency decreases; if the reasoning conclusion is "prerequisites are missing", and students consistently make mistakes on the question types corresponding to the prerequisite knowledge points, then consistency increases.

[0075] In practice, each rule can be assigned a value based on a preset consistency rule and then aggregated to obtain the result. .

[0076] (2) Sufficiency of evidence score : Used to measure whether the evidence used for reasoning is sufficient and stable.

[0077] For example: near The higher the error repetition rate of similar questions, the more times they are redone / reminded, and the more concentrated the pauses, the stronger the evidence; the stronger the evidence is when there is only one answer or when there is a lack of process information.

[0078] In practice, various observables can be normalized and weighted to obtain... .

[0079] (3) Rating of reasoning stability Used to measure whether the reasoning conclusion is stable under the same input.

[0080] In practice, the following approach can be used: repeatedly infer the same input using different prompt templates or different random sampling parameters. Alternatively, cross-validation with two inferencers / models can be used, with the proportion of consensus as the criterion. .

[0081] Finally, the confidence score is obtained by merging the above scores. ,For example: .

[0082] in, , and It is a preset coefficient and satisfies .

[0083] And Mapping to hierarchical results: when At that time, high confidence was maintained. At that time, Zhongzhixin, when The timeframe is considered low confidence. Low confidence triggers a conservative update or verification mechanism to suppress graph state fluctuations and erroneous update contamination. Specifically, For high confidence threshold, It is a low confidence threshold and can be set as needed.

[0084] 4. Consistency verification and re-inference triggering.

[0085] When there is a significant conflict between the reasoning conclusion and the objective answer result or scoring point (e.g., the reasoning judges "mastered", but key steps are completely missing; or the judgment is "careless calculation", but the same type of error is repeated multiple times), a consistency check is triggered. This can involve re-reasoning, introducing a more rigorous template for review, or adopting a conservative judgment (lowering the confidence level / outputting "partially mastered") to improve stability and reliability. Taking the knowledge point "quadratic formula for solving a quadratic equation in one variable" as an example, suppose we collect data on students' answers to the question x... 2 The solution 5x+6=0 gives the answer x=2 (missing the solution x=3), and exhibits the behavior of "pausing the video twice, requesting a hint once, and redoing twice".

[0086] First, evidence extraction: The student was able to complete part of the completing the square / factoring, but did not perform a "completeness check of the result", resulting in missing solutions; Secondly, the reason is more consistent with "insufficient awareness of the integrity of the solution set / missing verification steps" than simply a calculation error; Next, output: Status label = "Partially mastered (risk of incomplete solution set)"; Error cause description = "The completeness check of the root was not performed, resulting in missing solution"; Confidence level = Medium (because the student repeatedly redid the problem and still missed the solution, and did not complete it after being prompted).

[0087] If similar problems continue to be left unsolved, the weight of "conceptual weakness" will be increased in the consistency check, and the confidence level will be raised to high.

[0088] S3: Update the relationship between requirement determination and knowledge point mapping.

[0089] Based on the learning state reasoning results obtained in step S2, the student's learning state during the learning process is determined, and the knowledge comprehension problem corresponding to the learning state is further analyzed.

[0090] Specifically, based on the inference results of the learning state, it is identified whether the student falls into at least one of the following situations: (1) Insufficient understanding of a specific knowledge point; (2) There is confusion or misunderstanding between related concepts; (3) The lack of prior knowledge points hinders the understanding of current knowledge points.

[0091] If the above problems exist, it indicates that the student has made errors during the learning process, and the error type can be determined. If errors are identified, the knowledge point nodes in the teaching knowledge graph need to be updated. If no errors are identified, no updates to the knowledge point nodes in the teaching knowledge graph are needed, and the original state of the knowledge point nodes in the teaching knowledge graph can be maintained.

[0092] When it is necessary to update the knowledge point nodes of the teaching knowledge graph, a set of candidate knowledge points related to the reasoning results of the learning state is generated based on the structure of the teaching knowledge graph. The set of candidate knowledge points includes at least the knowledge point currently being learned and its predecessor or related knowledge points in the teaching knowledge graph.

[0093] Subsequently, based on the preset or learned mapping rules, the learning state reasoning results are matched with the candidate knowledge point set to determine the target knowledge point corresponding to the student's learning state, and a mapping relationship from the student's learning state to the teaching knowledge graph node is generated.

[0094] Specifically, after obtaining the learning state reasoning result output in step S2, the present invention further realizes the mapping of "learning state → knowledge point node" based on the teaching knowledge graph, which can be implemented in the following way: 1. Construct a candidate knowledge point set.

[0095] Based on the knowledge points currently being learned by students Centered on this, a set of candidate knowledge points is formed by retrieving them from the teaching knowledge graph. The set of candidate knowledge points It should include at least: the knowledge points currently being learned. The knowledge points currently being studied prerequisite knowledge points / dependent knowledge points (e.g., upstream knowledge points with relationships such as "prerequisites", "includes", "based on"); and knowledge points currently being studied. Related knowledge points that have connections such as "easily confused / similar / related examples". The knowledge points currently being studied Knowledge points at the same level (e.g., belonging to the same chapter or question type label).

[0096] By constructing a set of candidate knowledge points, it is ensured that subsequent matching is only within a "reasonable range," that is, only within the set of candidate knowledge points, and the learning state will not be mismapped to knowledge points of irrelevant chapters, thereby improving the efficiency of matching.

[0097] 2. Candidate knowledge point matching and scoring (Rule + Similarity).

[0098] For the set of candidate knowledge points Each candidate knowledge point The learning state reasoning result is matched with the learning state reasoning result using rule matching and semantic similarity to determine the target knowledge point corresponding to the learning state reasoning result.

[0099] Specifically, a combined strategy of rule matching as the primary method and similarity as a secondary method can be used to identify each candidate knowledge point. Matching with the reasoning results of the learning state.

[0100] (1) Rule matching (strong constraints).

[0101] If the status label is "preceding missing", prioritize matching the knowledge point currently being learned. The set of prerequisite knowledge points; If the status is "Conceptual Confusion", priority will be given to matching the knowledge point currently being learned. Knowledge points that are related by "easily confused / similar concepts"; If the status label is “Missing step / Broken reasoning”, priority will be given to matching process-related knowledge points (such as “verification steps”, “completeness check”, and “condition determination”) that are related to the question type steps.

[0102] (2) Semantic similarity (soft matching).

[0103] The text similarity between the error cause type and error cause evidence and the candidate knowledge points' "node name / definition / example description / tag" is calculated (using vector retrieval or keyword matching) and used to rank the candidate knowledge points in the same category.

[0104] Finally, the top-1 or top-N candidate knowledge points of the same type are selected as the target knowledge point set. The knowledge points in the text.

[0105] 3. Generate the mapping relationship between "state → node" (with state labels and weights).

[0106] Based on the target knowledge points, establish a mapping relationship between the learning state reasoning results and the knowledge point nodes of the teaching knowledge graph.

[0107] In other words, the mapping results are written into the teaching knowledge graph to form traceable relationships. For example: (Student, Mastery, (weight=0.82), (Student, Weakness) (weight=0.76), (Student, confusion, (e.g., weight=0.68)

[0108] Among them, the relationship type, such as "master" or "weak", is determined by the status label, and the weight can be obtained by fusing the confidence score and the semantic similarity score.

[0109] 4. Conflict checking and rollback (to prevent mismapping). If a conflict occurs (e.g., the reasoning judgment is "mastered" but the objective score is very low), the weight is reduced and the algorithm reverts to the previous level. Or output a set of multi-objective knowledge points. (For example, simultaneously label) ("Partially mastered" + "weak" in a certain prerequisite knowledge point) to avoid misjudgment on a single point.

[0110] For example, if a student repeatedly omits one root in a problem about finding the roots of a quadratic equation, and the error is attributed to "missing a completeness check of the result," then this error can be mapped to the knowledge point node "checking the roots / checking the completeness of the solution set" in the teaching knowledge graph. This generates a mapping relationship of (student, weak point, this knowledge point node) and uses it as the target knowledge point for subsequent personalized remedial learning.

[0111] S4: Update the status of knowledge point nodes in the teaching knowledge graph.

[0112] After determining the mapping relationship between the learning state reasoning result and the knowledge point node of the teaching knowledge graph in step S3, the state of the corresponding knowledge point node in the teaching knowledge graph is updated.

[0113] The instructional knowledge graph is used to describe the prerequisite dependencies, inclusion relationships, or hierarchical relationships between knowledge points. For the mapped target knowledge point, the node state corresponding to the target knowledge point in the instructional knowledge graph is adjusted based on the reasoning results of the student's learning state to reflect the student's actual mastery of that knowledge point.

[0114] In practical implementation, the node status of the knowledge point nodes in the teaching knowledge graph includes at least information such as the knowledge point mastery status, cognitive risk markers, and error cause records. Different node update strategies can be adopted based on the confidence level of the reasoning results to avoid excessive fluctuations in the knowledge graph state due to reasoning uncertainty.

[0115] Specifically, when the confidence level is high, a strong update is performed on the corresponding knowledge point node, so that the node state of the teaching knowledge graph quickly converges to the new diagnostic conclusion.

[0116] When the confidence level is medium confidence, weak or smooth updates are applied to the corresponding knowledge point nodes to avoid drastic fluctuations in the node state of the teaching knowledge graph caused by a single inference.

[0117] When the confidence level is low, the corresponding knowledge point node is not updated to avoid accidental updates that could contaminate the teaching knowledge graph.

[0118] In this invention, the results of node state updates can also be propagated in association under the structural constraints of the teaching knowledge graph. For example, when a current knowledge point is determined to be unmastered, potential learning obstacles are marked for subsequent knowledge points.

[0119] Furthermore, the teaching knowledge graph node mapping method based on learning state reasoning of the present invention may further include: S5: Learning diagnostic results output and closed-loop update.

[0120] Based on the updated teaching knowledge graph in step S4, a learning diagnostic result for students at the knowledge structure level is generated, and the learning diagnostic result is output to support subsequent teaching decisions.

[0121] The learning diagnostic results include at least the identification of the student's weak knowledge points, the corresponding explanations of the causes of the errors, and suggested directions for teaching intervention. These results can be provided to the teaching system or digital human teaching module via a structured interface to generate targeted teaching strategies or instructional content.

[0122] As the teaching process continues, new learning feedback information from students can be continuously collected and used as input for the next round of student learning state reasoning analysis and node updates of the teaching knowledge graph, thus forming a closed-loop learning state modeling process of "learning feedback information acquisition - learning state reasoning analysis - knowledge point mapping - node state update - diagnostic output".

[0123] This invention can be widely applied to intelligent teaching scenarios that require high accuracy and personalization in learning diagnosis, including but not limited to online education platforms, digital human teaching systems, adaptive learning systems, and intelligent tutoring and Q&A systems.

[0124] In the aforementioned applications, this invention can perform in-depth reasoning analysis on students' learning status based on real-time learning feedback, and accurately map the diagnostic results to the teaching knowledge structure, effectively alleviating the problem of insufficient teaching relevance caused by the rough expression of learning status. Without increasing teaching content or burden, it achieves accurate identification and interpretable presentation of students' weak knowledge points, making it particularly suitable for teaching scenarios with significant student differences and complex knowledge structures. Furthermore, this invention can be integrated into existing intelligent teaching systems as a general learning status diagnosis and knowledge structure modeling module, possessing good engineering feasibility and broad industrial application prospects.

[0125] Figure 2 A schematic diagram of the structure of the teaching knowledge graph node mapping system based on learning state reasoning of the present invention is shown. Figure 2 As shown, the teaching knowledge graph node mapping system based on learning state reasoning of the present invention includes: 1. Module for acquiring and preprocessing student learning feedback information in a structured manner.

[0126] The student learning feedback information acquisition and structured preprocessing module is used to acquire students' learning feedback information during the teaching process and perform structured preprocessing on it.

[0127] 2. CoT-based multi-step reasoning analysis module for student learning status.

[0128] The CoT-based multi-step reasoning analysis module for student learning states is used to perform CoT multi-step reasoning analysis on the learning states during the learning process based on the structured preprocessed learning feedback information and a large language model, so as to map the learning feedback information into interpretable learning state reasoning results.

[0129] 3. Update the module for generating the relationship between requirements assessment and knowledge point mapping.

[0130] The update requirement determination and knowledge point mapping relationship generation module is used to determine the student's learning state during the learning process based on the learning state reasoning result to determine whether the knowledge point nodes of the teaching knowledge graph need to be updated. When it is determined that the knowledge point nodes of the teaching knowledge graph need to be updated, a set of candidate knowledge points related to the knowledge points that the student is currently learning is generated in combination with the teaching knowledge graph structure. According to the preset or learned mapping rules, the learning state reasoning result is matched with the set of candidate knowledge points to determine the target knowledge point corresponding to the learning state reasoning result and generate the mapping relationship from the learning state reasoning result to the knowledge point nodes of the teaching knowledge graph.

[0131] 4. Knowledge point node status update module for teaching knowledge graph.

[0132] The knowledge point node state update module of the teaching knowledge graph is used to update the state of the corresponding knowledge point node in the teaching knowledge graph based on the mapping relationship between the learning state reasoning result and the knowledge point node in the teaching knowledge graph.

[0133] Preferably, the teaching knowledge graph node mapping system based on learning state reasoning of the present invention may further include: 5. Learn the diagnostic results output and closed-loop update module.

[0134] The learning diagnostic results output and closed-loop update module is used to generate learning diagnostic results for students at the knowledge structure level based on the updated teaching knowledge graph, and output the learning diagnostic results to support subsequent teaching decisions.

[0135] Furthermore, this invention also provides a teaching knowledge graph node mapping device based on learning state reasoning. For example... Figure 3 As shown, the teaching knowledge graph node mapping device based on learning state reasoning of the present invention includes: a memory 11 for storing one or more programs; one or more processors 12; when the one or more programs are executed by the one or more processors 12, the one or more processors 12 implement the teaching knowledge graph node mapping method based on learning state reasoning of the present invention. Finally, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the teaching knowledge graph node mapping method based on learning state reasoning in the present invention.

[0136] The computer-readable storage medium includes both permanent and non-permanent, removable and non-removable media, and information storage can be achieved by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device. As defined in this invention, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0137] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0138] The steps of the methods or algorithms described in conjunction with the embodiments disclosed in this invention can be implemented in hardware, software modules executed by a processor, or a combination of both. The software modules can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art.

[0139] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the present invention. Those skilled in the art can modify or make equivalent substitutions to the technical solutions of the present invention based on the concept of the present invention, without departing from the essence and scope of the technical solutions of the present invention.

Claims

1. A method for mapping nodes in a teaching knowledge graph based on learning state reasoning, characterized in that, Includes the following steps: S1: Obtain students' learning feedback information during the teaching process and perform structured preprocessing; S2: Based on the structured preprocessed learning feedback information, the CoT multi-step reasoning analysis is performed on the learning state of students in the learning process using a large language model, so as to map the learning feedback information into interpretable learning state reasoning results. S3: Based on the learning state reasoning result, determine the student's learning state during the learning process to determine whether the knowledge point nodes of the teaching knowledge graph need to be updated. When it is determined that the knowledge point nodes of the teaching knowledge graph need to be updated, generate a set of candidate knowledge points related to the knowledge points that the student is currently learning by combining the structure of the teaching knowledge graph. According to the preset or learned mapping rules, match the learning state reasoning result with the set of candidate knowledge points to determine the target knowledge point corresponding to the learning state reasoning result and generate the mapping relationship between the learning state reasoning result and the knowledge point nodes of the teaching knowledge graph. S4: Based on the mapping relationship between the learning state reasoning results and the knowledge point nodes of the teaching knowledge graph, update the state of the corresponding knowledge point nodes in the teaching knowledge graph.

2. The teaching knowledge graph node mapping method based on learning state reasoning according to claim 1, characterized in that, In step S1, the learning feedback information obtained includes one or more of the following: students' answers to teaching content, exercises, or test questions; students' interactive behavior data during the teaching process; and the types of errors, pauses, repeated answers, or repeated requests for explanations that students make during the learning process. Students' historical learning records, periodic learning outcomes, or mastery of knowledge points.

3. The teaching knowledge graph node mapping method based on learning state reasoning according to claim 1, characterized in that, Step S2 specifically includes: S21: Convert the structured preprocessed learning feedback information into an inference input package; S22: Based on the inference input package, a fixed multi-step inference template is used to guide the large language model to perform CoT multi-step inference analysis, so as to gradually output intermediate judgment results and finally output structured learning state inference results. The structured learning state inference results include state labels, error cause types, error cause evidence, and confidence level.

4. The teaching knowledge graph node mapping method based on learning state reasoning according to claim 3, characterized in that, The confidence level is determined based on the confidence score, and the confidence score is: ， In the formula, the evidence consistency score The sufficiency of evidence score is used to measure whether the status label is consistent with objective evidence. The reasoning stability score is used to measure whether the evidence in a reasoning argument is sufficient and stable. Used to measure whether the state label is stable under the same input. , and It is a preset coefficient and satisfies ; Furthermore, determining the confidence level based on the confidence score specifically involves: when The confidence level is high confidence when The confidence level is medium confidence, when The confidence level is low confidence, among which, For high confidence threshold, This is the low confidence threshold.

5. The teaching knowledge graph node mapping method based on learning state reasoning according to claim 4, characterized in that, In step S3, when the learning state reasoning results indicate that the student has at least one of the following situations in the current or subsequent knowledge points, it is determined that the knowledge point nodes of the teaching knowledge graph need to be updated: insufficient understanding of a specific knowledge point, confusion or misunderstanding between related concepts, or obstruction of understanding of the current knowledge point due to the lack of prior knowledge points.

6. The teaching knowledge graph node mapping method based on learning state reasoning according to claim 5, characterized in that, In step S3, the process of generating a set of candidate knowledge points related to the knowledge points the student is currently learning, based on the teaching knowledge graph structure, and matching the learning state reasoning result with the set of candidate knowledge points according to preset or learned mapping rules, determining the target knowledge point corresponding to the learning state reasoning result, and generating the mapping relationship from the learning state reasoning result to the knowledge point node of the teaching knowledge graph specifically includes: S31: Based on the knowledge points that students are currently learning Centered on this, a set of candidate knowledge points is formed by retrieving them from the teaching knowledge graph. The candidate knowledge point set It should include at least: the knowledge points currently being learned. Knowledge points currently being studied Prerequisite knowledge points and / or dependent knowledge points, and knowledge points currently being learned Related knowledge points with connections, and knowledge points currently being learned. Knowledge points at the same level; S32: For the set of candidate knowledge points Each candidate knowledge point The learning state reasoning result is matched with the learning state reasoning result using rule matching and semantic similarity to determine the target knowledge point corresponding to the learning state reasoning result; S33: Based on the target knowledge points, establish a mapping relationship between the learning state reasoning results and the knowledge point nodes of the teaching knowledge graph.

7. The teaching knowledge graph node mapping method based on learning state reasoning according to claim 6, characterized in that, In step S4, when updating the state of the corresponding knowledge point node in the teaching knowledge graph, different update strategies are adopted according to the confidence level: when the confidence level is high, a strong update is performed on the corresponding knowledge point node; when the confidence level is medium, a weak update or smooth update is performed on the corresponding knowledge point node; when the confidence level is low, no update is performed on the corresponding knowledge point node.

8. A teaching knowledge graph node mapping system based on learning state reasoning, characterized in that, include: The module for acquiring and preprocessing student learning feedback information is used to acquire students' learning feedback information during the teaching process and perform structured preprocessing. The CoT-based multi-step reasoning analysis module for student learning status is used to perform CoT multi-step reasoning analysis on the student's learning status during the learning process based on the structured preprocessed learning feedback information and using a large language model, so as to map the learning feedback information into interpretable learning status reasoning results. The module for updating the requirement determination and knowledge point mapping relationship generation is used to determine the student's learning state during the learning process based on the learning state reasoning result to determine whether the knowledge point nodes of the teaching knowledge graph need to be updated. When it is determined that the knowledge point nodes of the teaching knowledge graph need to be updated, a set of candidate knowledge points related to the knowledge points that the student is currently learning is generated in combination with the structure of the teaching knowledge graph. According to the preset or learned mapping rules, the learning state reasoning result is matched with the set of candidate knowledge points to determine the target knowledge point corresponding to the learning state reasoning result and generate the mapping relationship from the learning state reasoning result to the knowledge point nodes of the teaching knowledge graph. The knowledge point node status update module of the teaching knowledge graph is used to update the status of the corresponding knowledge point node in the teaching knowledge graph based on the mapping relationship between the learning state reasoning result and the knowledge point node in the teaching knowledge graph.

9. A teaching knowledge graph node mapping device based on learning state reasoning, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the teaching knowledge graph node mapping method based on learning state reasoning as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the teaching knowledge graph node mapping method based on learning state reasoning as described in any one of claims 1-7.

Images